Validation of artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine in Gabon

Brice Meulah; Prosper Oyibo; Pytsje T Hoekstra; Paul Alvyn Nguema Moure; Moustapha Nzamba Maloum; Romeo Aime Laclong-Lontchi; Yabo Josiane Honkpehedji; Michel Bengtson; Cornelis Hokke; Paul L A M Corstjens; Temitope Agbana; Jan Carel Diehl; Ayola Akim Adegnika; Lisette van Lieshout

doi:10.1371/journal.pntd.0011967

. 2024 Feb 23;18(2):e0011967. doi: 10.1371/journal.pntd.0011967

Validation of artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine in Gabon

Brice Meulah ^1,^2,^*, Prosper Oyibo ³, Pytsje T Hoekstra ¹, Paul Alvyn Nguema Moure ^2,⁴, Moustapha Nzamba Maloum ², Romeo Aime Laclong-Lontchi ², Yabo Josiane Honkpehedji ^1,^2,⁵, Michel Bengtson ¹, Cornelis Hokke ¹, Paul L A M Corstjens ⁶, Temitope Agbana ³, Jan Carel Diehl ⁷, Ayola Akim Adegnika ^1,^2,^4,^5,⁸, Lisette van Lieshout ¹

Editor: Gabriel Rinaldi⁹

¹Leiden University Center for Infectious Diseases (LUCID), Leiden University Medical Center, Leiden, The Netherlands

²Centre de Recherches Médicales des Lambaréné, CERMEL, Lambaréné, Gabon

³Mechanical, Maritime and Material Engineering, Delft University of Technology, Delft, The Netherlands

⁴Ecole doctorale régionale d’Afrique centrale en infectiologie tropicale de Franceville, Gabon

⁵Fondation pour la Recherche Scientifique, Cotonou, Benin

⁶Department of Cell and Chemical Biology, Leiden University Medical Center, Leiden, The Netherlands

⁷Industrial Design Engineering, Delft University of Technology, Delft, The Netherlands

⁸Institut fur Tropenmedizin, Universitat Tubingen, Tubingen, Germany

⁹Aberystwyth University - Penglais Campus: Aberystwyth University, UNITED KINGDOM

The authors have declared that no competing interests exist.

^✉

* E-mail: b.meulah_tcheubousou@lumc.nl

Roles

Brice Meulah: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing

Prosper Oyibo: Data curation, Investigation, Software, Writing – review & editing

Pytsje T Hoekstra: Conceptualization, Data curation, Formal analysis, Methodology, Project administration, Supervision, Validation, Writing – review & editing

Paul Alvyn Nguema Moure: Investigation

Moustapha Nzamba Maloum: Investigation, Writing – review & editing

Romeo Aime Laclong-Lontchi: Investigation

Yabo Josiane Honkpehedji: Investigation, Writing – review & editing

Michel Bengtson: Conceptualization, Investigation, Methodology, Project administration, Supervision, Writing – review & editing

Cornelis Hokke: Conceptualization, Funding acquisition, Supervision, Writing – review & editing

Paul L A M Corstjens: Investigation, Writing – review & editing

Temitope Agbana: Funding acquisition, Supervision, Writing – review & editing

Jan Carel Diehl: Funding acquisition, Supervision, Writing – review & editing

Ayola Akim Adegnika: Conceptualization, Funding acquisition, Resources, Supervision, Writing – review & editing

Lisette van Lieshout: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Validation, Writing – review & editing

Gabriel Rinaldi: Editor

PMCID: PMC10917302 PMID: 38394298

Abstract

Introduction

Schistosomiasis is a significant public health concern, especially in Sub-Saharan Africa. Conventional microscopy is the standard diagnostic method in resource-limited settings, but with limitations, such as the need for expert microscopists. An automated digital microscope with artificial intelligence (Schistoscope), offers a potential solution. This field study aimed to validate the diagnostic performance of the Schistoscope for detecting and quantifying Schistosoma haematobium eggs in urine compared to conventional microscopy and to a composite reference standard (CRS) consisting of real-time PCR and the up-converting particle (UCP) lateral flow (LF) test for the detection of schistosome circulating anodic antigen (CAA).

Methods

Based on a non-inferiority concept, the Schistoscope was evaluated in two parts: study A, consisting of 339 freshly collected urine samples and study B, consisting of 798 fresh urine samples that were also banked as slides for analysis with the Schistoscope. In both studies, the Schistoscope, conventional microscopy, real-time PCR and UCP-LF CAA were performed and samples with all the diagnostic test results were included in the analysis. All diagnostic procedures were performed in a laboratory located in a rural area of Gabon, endemic for S. haematobium.

Results

In study A and B, the Schistoscope demonstrated a sensitivity of 83.1% and 96.3% compared to conventional microscopy, and 62.9% and 78.0% compared to the CRS. The sensitivity of conventional microscopy in study A and B compared to the CRS was 61.9% and 75.2%, respectively, comparable to the Schistoscope. The specificity of the Schistoscope in study A (78.8%) was significantly lower than that of conventional microscopy (96.4%) based on the CRS but comparable in study B (90.9% and 98.0%, respectively).

Conclusion

Overall, the performance of the Schistoscope was non-inferior to conventional microscopy with a comparable sensitivity, although the specificity varied. The Schistoscope shows promising diagnostic accuracy, particularly for samples with moderate to higher infection intensities as well as for banked sample slides, highlighting the potential for retrospective analysis in resource-limited settings.

Trial registration

NCT04505046 ClinicalTrials.gov.

Author summary

Assessment of schistosomiasis control programs is a crucial step to understanding the success rate of these control programs. The Schistoscope: an AI-powered automated digital microscope could overcome the limitations of conventional microscopy in endemic resource limited settings as well as in settings lacking microscopy experts. In this study, we carried out an extensive validation of the Schistoscope’s diagnostic performance for diagnosis of urogenital schistosomiasis compared to conventional microscopy as well as more accurate diagnostic tests such as real-time PCR and the up-converting particle (UCP) lateral flow (LF) test for the detection of circulating anodic antigen (CAA) on freshly collected urines. We also assessed the performance of the Schistoscope for the diagnosis of schistosomiasis on banked sample slides, using a simple and sustainable storage method, for approximately two years. Having a tool that can prospectively and retrospectively analyse samples in an easy and sustainable way could facilitate schistosomiasis control programs in settings with little or no access to microscopists. Overall, we found the Schistoscope to be as good as conventional microscopy for the diagnosis of schistosomiasis, and given its downstream advantages of digital health, it would serve as a valuable diagnostic/screening tool in resource limited endemic settings.

1. Introduction

Schistosomiasis is a tropical parasitic disease of significant public health concern, with an estimated 700 million individuals at risk of infection in areas known for transmission. Out of approximately 250 million people requiring preventive chemotherapy worldwide, Sub-Saharan Africa, including the centrally located country of Gabon, bears the highest proportion [1–3]. In order to control the disease morbidity and work towards its elimination as a public health problem, the World Health Organization (WHO) recommends annual preventive chemotherapy using a single dose of praziquantel for all individuals aged two years and above in communities where the prevalence of schistosomiasis is 10% or higher [4]. For communities with a prevalence below 10%, an optional test-and-treat strategy is recommended [4]. In both cases, reliable diagnostic tools are essential to support the monitoring and evaluation of these control strategies [5,6].

Conventional microscopy is the standard diagnostic procedure for schistosomiasis. However, the need for expert microscopists limits its application in resource-limited settings. Real-time polymerase chain reaction (PCR) for amplification and detection of schistosome-specific nucleic acid sequences, as well as a lateral flow test (LF) for the detection of schistosome-specific circulating anodic antigen (CAA), offer higher sensitivity and specificity than conventional microscopy [7,8]. Nevertheless, the requirement for specialized skills and advanced infrastructure currently limits their application in resource-limited settings.

Alternatively, automated digital microscopes have shown promising results in the diagnosis of schistosomiasis by detecting parasite eggs in stool or urine [9–12]. The application of artificial intelligence (AI) algorithms in the diagnosis and surveillance of infectious diseases has received significant attention [13–15]. Automated digital microscopes are designed to capture images of samples with simultaneous analysis by an AI algorithm trained to detect parasite components. Such innovative tools are relatively easy to use and can be customised for rural endemic settings. These tools also have propitious downstream applications including digital health [11,16–18]. In particular for the detection of S. haematobium eggs in urine, multiple studies have validated the diagnostic accuracy of AI-based digital microscopes, demonstrating sensitivities ranging from 32% to 91% compared to conventional microscopy, as summarised in a recent review [11].

The Schistoscope is an automated digital microscope with an integrated AI algorithm to detect S. haematobium eggs in urine samples. It was developed for use at point-of-need and is relatively easy to operate requiring minimal training [19,20]. The Schistoscope was first assessed in Nigeria for diagnosing urogenital schistosomiasis, revealing a high sensitivity but a rather low specificity compared to conventional microscopy [12]. Based on these results, the AI model was re-designed, retrained and embedded onboard the Schistoscope, and then validated using a set of field sample images, yielding better sensitivity and specificity [21]. A limitation of the previous studies has been the small size of validation sample dataset and the lack of an accurate reference standard. To perform more in-depth validation of the diagnostic accuracy of the Schistoscope in detecting S. haematobium eggs, urine samples were collected and analysed in a laboratory setting in Lambaréné, Gabon. The diagnostic performance of the Schistoscope was compared to conventional microscopy as well as to a composite reference standard (CRS), consisting of real-time PCR and UCP-LF CAA.

2. Methods

2.1. Ethics statement

Ethical approval for the study was obtained from the Comité d’Éthique Institutionnel (CEI) du Centre de Recherches Médicales de Lambaréné in Lambaréné, Gabon (reference no. CEI-CERMEL 005/2020). Prior to sample collection, written consent was obtained from adults and from parents or legal guardians of children and teenagers who wished to participate, indicated by their signatures. To ensure confidentiality and anonymity of the results, unique codes were assigned to the samples. Participants with detectable S. haematobium eggs/10 mL of urine based on microscopy were treated with praziquantel (40 mg/Kg of body weight) following local guidelines. The study was registered at ClinicalTrials.gov (NCT04505046).

2.2. Study design

The validation study was conducted in Lambaréné and surrounding areas, located in the Moyen-Ogooué province in Gabon, a region known to be endemic for S. haematobium with a prevalence of about 30% [22]. It was carried out in two parts: study A and study B (Fig 1). Study A was an independent cross-sectional study focusing on school-age children and adults from whom urine samples were collected and analysed by the Schistoscope, conventional microscopy, real-time PCR and UCP-LF CAA (see details below). Study B was partly embedded in several ongoing studies at Centre de Recherches Médicales de Lambaréné (CERMEL) in Gabon, where urine samples were collected from different populations (school-age children, adults and pregnant women) and analysed with a range of diagnostic methods including conventional microscopy, real-time PCR and UCP-LF CAA (see details below). Microscopy slides were subsequently biobanked at 4°C for retrospective analysis with the Schistoscope (~2 years later). All diagnostic procedures were conducted at CERMEL.

2.3. Sample size calculations

The Schistoscope was assumed to have a sensitivity and specificity non-inferior to conventional microscopy, which were realistically assumed to be 80% and 90%, respectively based on field expert estimates using real-time PCR and UCP-LF CAA. The sample size for both study A and B were determined based on a 30% prevalence of schistosomiasis in Lambaréné and its surrounding areas using a two-sample matched paired design, resulting in a required sample size of 350 urine samples [23]. A power of 80% and a 5% degree of error was considered for the calculations.

2.4. Sample collection and processing

Collection of urine samples in study A was carried out starting in 2023 while the urine sample biobanking (study B) was initiated in 2020. Study participants were provided with sterile containers labelled with unique identifiers and instructed to provide urine samples between 11 am and 2 pm. The samples were transported to CERMEL within 2 hours of collection for analysis. Microscopy slides were prepared by pressing 10 mL of homogenised urine through a 25mm membrane (pore size 30 μm; Whatmann International Ltd) with the use of a syringe and a filter holder and transferred onto a glass slide. For study A, the slides were examined on the same day using conventional microscopy and the Schistoscope. For study B, the slides were examined using conventional microscopy and stored at 4°C for about 2 years awaiting analysis with the Schistoscope. For both studies, 1 mL of urine from each sample was used for UCP-LF CAA analysis and 10 mL of homogenised urine was centrifuged and the resulting 1mL pellet was used for DNA extraction and amplification before biobanking the sample slides for retrospective analysis with the Schistoscope.

2.5. Diagnostic methods

2.5.1. The Schistoscope

Five Schistoscopes were used in this study (Fig 2A and S1 Video). Analysis was done following the standard operating procedure (S1 Manual) of the Schistoscope. Briefly, the slide was placed on the slide holder of the Schistoscope such that its microscope objective aligned with the filter membrane of the slide. The device’s autofocus algorithm positioned the microscope objective in the optimal focal plane. High resolution images of the sample were registered and analysed simultaneously by the integrated AI algorithm. The number of detected eggs (expressed in eggs/10~ml of urine) is displayed on a pop-up result window which also indicated the end of the sample analysis. Detected eggs are marked as shown in Fig 2B and 2C. The results from the Schistoscope were exported as an Excel-compatible file.

2.5.2.Conventional microscopy

Slides from both studies were analysed immediately after urine filtration under 10x objective of a Leica microscope (model: DM1000 LED, Microsystems CMS GmbH Ernt-Leitz-Str.17-37 Wetzlar, Germany). Each slide was examined by two independent microscopists and the mean egg count was calculated. In case of a >20% discrepancy in egg count, an additional reading by a third independent microscopist was required and the final egg count was determined by calculating the mean of the two closest egg counts obtained from the three readings. All egg counts were expressed as eggs/10 mL of urine. In addition, the storage conditions (4°C) and quality of the biobanked slides were monitored using conventional microscopy once every four months during the storage period. This was done by monitoring daily temperature of the fridge as well as by determining the egg counts of three known slides. Additionally, during the Schistoscope analysis the integrity of the biobank was quality controlled by re-examining a random selection of 10% of the slides by conventional microscopy and comparing the results to the outcomes before storage.

2.5.3. Nucleic acid extraction and real-time PCR

Genomic DNA extraction was carried out using the QIAamp Mini kit (cat: 51306; Qiagen) according to the manufacturer’s instructions. Briefly, 195μL of each centrifuged urine pellet was mixed with 5μL of internal control DNA commercially available as a DNA Extraction Control (DEC) 670 kit (Cat: BIO-35028; Bioline). The DEC 670 kit is supplied as a vial of internal control DNA sequence (with no known homology to sequences of any organism) and a vial of control mix containing primers and probes complementary to the internal control DNA sequence. The final mixture was then processed as previously described [8].

Real-time PCR was performed as previously described [8, 24] using a set of primers (Ssp48F and Ssp124R) and probe (Ssp78T) complementary to the 77-bp internal transcribed spacer-2 (ITS2) sequence, with minor modifications on the internal control (see above) as well as on the reaction mixture and conditions used (see below).

Amplification reactions were performed in a 15μL reaction mixture containing 1x No-ROX master mix (Cat: BIO-86005; Bioline), 4.5pmol of each Schistosoma-specific primer, 1.5pmol Schistosoma-specific probe, 0.4μL of control mix, 1.2μL of nuclease free water and 2.5μL DNA extract. The PCR runs consisted of an initial step of 5 min at 95°C followed by 40 successive cycles of 10 sec at 95°C and 60 sec at 60°C. The reaction was run on a Light cycler 480 II real-time PCR system (Roche Diagnostics). Schistosoma DNA detection was expressed in threshold (Ct) cycles. For every run, a non-template control and a positive control (S. haematobium DNA, Ct-value 23–25) was included. A test was considered positive when the threshold was attained within 40 PCR cycles (Ct-value ≤ 40). Each sample was run in duplicate and was considered positive when at least one of the duplicates was positive. Amplification of the internal control at the expected Ct-value showed success of nucleic acid extraction and no evidence of PCR inhibitors.

2.5.4. UCP-LF CAA

Urine CAA concentration was determined by the UCP-LF CAA assay using the UCAAhT417 format as previously described [7]. Briefly, 500μL of each urine sample was mixed with 100μL of 12% trichloroacetic acid, incubated and centrifuged. The clear supernatant obtained was concentrated to 20μL using an Amicon Ultra-4 concentration column (Millipore, Merck Chemicals B.V., Amsterdam, The Netherlands) and subsequently mixed with 50μL run buffer and added to 50 μL UCP solution. The resulting mixture was then used for the lateral flow assay. A set of CAA standards was used to validate the cut-off (2 pg/ml) as well as to reliably quantify the amount of CAA per sample up to 1000 pg/ml [7].

2.6. Statistical analyses

In study A, only samples with all four diagnostic test results available were included in the final analysis. For study B, samples with both the Schistoscope and conventional microscopy test results only were first analysed (B1). Additionally, a subset of samples (B2) which had outcomes of all four diagnostic tests was analysed separately (Fig 1). The percentage positive samples for a Schistosoma infection was determined for each diagnostic test. The sensitivity and specificity of the Schistoscope were assessed using conventional microscopy as the reference (study A, B1 and B2). Sensitivity and specificity of the Schistoscope and conventional microscopy were further evaluated using a combination of real-time PCR and/or UCP-LF CAA as a CRS (study A and B2). A sample was deemed positive by the CRS if it showed the presence of Schistosoma spp DNA and/or CAA. Conversely, a sample was considered negative if both diagnostic tests showed a negative outcome. A ≤10% difference in sensitivity and specificity between the Schistoscope and conventional microscopy based on the CRS was considered non-inferior. To determine the performance of the Schistoscope at different infection intensities, egg counts based on conventional microscopy were categorised into very low intensity infection (1–9 eggs/10 mL), low-intensity infection (10–49 eggs/10 mL) and high-intensity infection (≥50 eggs/10 mL) [25,26]. Cohen’s Kappa (k) statistics was computed to assess the qualitative agreement between the Schistoscope and conventional microscopy, and the CRS. Spearman’s correlation (r) was used to assess the strength of association between the Schistoscope and conventional microscopy, real-time PCR and UCP-LF CAA. Bland-Altman analysis was further used to assess the quantitative agreement between the Schistoscope and conventional microscopy. Wilcoxon sign rank test was used to compare the microscopy egg count of the randomly selected banked slides before and after storage. Statistical analysis was performed using IBM Statistical Package for Social Sciences version 25 (SPSS Inc., Chicago, United States of America) and GraphPad Prism version 9.0.1 for Windows (GraphPad Software, San Diego, California USA, www.graphpad.com).

3. Results

3.1. Study A: Diagnostic performance of the Schistoscope on freshly prepared samples

A total of 339 samples had outcomes available for all four diagnostic tests and were included in the analysis. Table 1 shows the proportion of positive results per diagnostic test. Real-time PCR found the highest proportion of positives (51.0%) followed by the UCP-LF CAA assay (46.6%). The proportion of positives detected by the Schistoscope (46.0%) was higher than that of conventional microscopy (38.3%). The median egg count of the Schistoscope (17 eggs/10ml) was lower than that of microscopy (31 eggs/10ml). The proportion of positives with egg count ≥50 eggs/10 mL by the Schistoscope and microscopy were comparable, 47 (30.1%) and 49 (37.7%), respectively (S1A Fig).

Table 1. Diagnostic outcomes of the Schistoscope in comparison to conventional microscopy, real-time PCR and UCP-LF CAA in study A and B.

	Study A (N = 339)				Study B1 (N = 798)		Study B2 (N = 349)
Diagnostic test	Schistoscope	Microscopy	Real-time PCR	UCP-LF CAA	Schistoscope	Microscopy	Schistoscope	Microscopy	Real-time PCR	UCP-LF CAA
Positive (%)	156 (46.0%)	130 (38.3%)	173 (51.0%)	158 (46.6%)	374 (46.9%)	307 (38.5%)	204 (58.5%)	190 (54.4%)	217 (62.2%)	225 (64.5%)
Range	1–1623 eggs/10mL	1–2516 eggs/10mL	20.2–37.0 Ct	2.6–1000.0 pg/mL	1–2879 eggs /10mL	1–9350 eggs/10 mL	1–1943 eggs/10mL	1–9350 eggs/10mL	19.1–38.7 Ct	2.1–1000.0 pg/mL
Median of the positives	17 eggs/10mL	31 eggs/10mL	29.0 Ct	65.0 pg/mL	17 eggs/10mL	105 eggs/10mL	32 eggs/10mL	209 eggs/10mL	26.6 Ct	189.6 pg/mL
Mean of the positives	78 eggs/10mL	119 eggs/10mL	29.7 Ct	134.0 pg/mL	136 eggs/10mL	464 eggs/10mL	160 eggs/10mL	565 eggs/10mL	27.8 Ct	310.5 pg/mL

Open in a new tab

Qualitatively, a moderate agreement between the Schistoscope and conventional microscopy was observed (K = 0.579, P<0.001). However, the agreement was only fair when compared to the CRS (K = 0.396, P<0.001) whereas a moderate agreement was observed between conventional microscopy and the CRS (K = 0.537, P<0.001) (Table 2). The sensitivity and specificity of the Schistoscope were 83.1% and 77.0%, respectively, when conventional microscopy was used as reference. In addition, when the Schistoscope and conventional microscopy were evaluated using the CRS, the sensitivity of the Schistoscope was 62.9% comparable to that of conventional microscopy, 61.9%. However, the specificity of the Schistoscope was significantly lower compared to the specificity of conventional microscopy. All samples with an egg count of ≥50 eggs/10mL defined by conventional microscopy were detected by the Schistoscope (S1A Fig). Of the microscopy positive samples with 1–9 eggs/10mL and 10–49 eggs/10mL, the Schistoscope detected 52.6% and 90.7% respectively. Conversely, the Schistoscope found 48 additional cases (of which 40 had <50 eggs/10mL) which were not detected by conventional microscopy. Of these additional cases, 35.4% and 27.1% were confirmed by real-time PCR and the UCP-LF CAA assay, respectively.

Table 2. Diagnostic performance and pairwise level of agreement by Cohen’s Kappa statistics between the Schistoscope and conventional microscopy and the composite reference for the detection of S. haematobium infection in study A and B.

Sample set	Diagnostic test	Reference test		Diagnostic test Sensitivity % (95% CI)	Diagnostic test Specificity % (95% CI)	Kappa	P value	Interpretation*
Study A (N = 339)		Microscopy
	Schistoscope	Positive	Negative	83.1 (75.5–89.1)	77.0 (70.7–82.5)	0.579	<0.001	Moderate
	Positive	108	48
	Negative	22	161
		Composite reference
	Schistoscope	Positive	Negative	62.9 (55.8–69.6)	78.8 (71.0–85.3)	0.396	<0.001	Fair
	Positive	127	29
	Negative	75	108
		Composite reference
	Microscopy	Positive	Negative	61.9 (54.8–68.6)	96.4 (91.7–98.8)	0.537	<0.001	Moderate
	Positive	125	5
	Negative	77	132
Study B1 (N = 798)		Microscopy
	Schistoscope	Positive	Negative	93.2 (89.7–95.7)	82.1 (78.4–85.4)	0,723	<0.001	Substantial
	Positive	286	88
	Negative	21	403
Study B2 (N = 349)		Microscopy
	Schistoscope	Positive	Negative	96.3 (92.6–98.5)	86.8 (80.5–91.6)	0.837	<0.001	Almost perfect
	Positive	183	21
	Negative	7	138
		Composite reference
	Schistoscope	Positive	Negative	78.0 (72.3–83.0)	90.9 (83.4–95.8)	0.604	<0.001	Moderate
	Positive	195	9
	Negative	55	90
		Composite reference
	Microscopy	Positive	Negative	75.2 (69.4–80.4)	98.0 (93.0–99.8)	0.619	<0.001	Substantial
	Positive	188	2
	Negative	62	97

Open in a new tab

A strong correlation was observed between egg counts estimated by the Schistoscope and conventional microscopy (r = 0.71, P<0.0001) (Fig 3A). A moderate correlation was observed between the Schistoscope egg counts and real-time PCR Ct-value (r = -0.58, P<0.0001), and CAA concentration (r = 0.58, P<0.0001) (Fig 3B and 3C, respectively). Bland-Altman analysis revealed that the Schistoscope tended to underscore egg counts compared to conventional microscopy, but approximately 95% of the difference in the egg count estimates between both methods fell within the limit of agreement (Fig 4A).

Fig 4 — Bland-Altman analysis demonstrating the quantitative agreement between the Schistoscope and conventional microscopy in study A (a) and B (b, c).

3.2. Study B: Diagnostic performance of the Schistoscope on banked samples

A total of 798 samples, for which both Schistoscope and conventional microscopy results were available, were included in the analysis (Study B1). Quality control of the biobank revealed no significant difference in microscopy egg count before and after storage which confirmed the integrity of the biobank. The percentage of positive cases detected by the Schistoscope (46.9%) was higher than by conventional microscopy (38.5%). The proportion of positives with an egg count of ≥50 eggs/10 mL was substantially lower by the Schistoscope (32.6%) than by conventional microscopy (59.3%) (S1B and S1C Fig).

A subset of 349 samples had test results available from all four diagnostic tests and were further analysed (Study B2). Based on real-time PCR and UCP-LF CAA a high percentage positive was observed, 62.2% and 64.5%, respectively. The percentage of positive cases detected by the Schistoscope and conventional microscopy were similar, 58.5% and 54.4% respectively, with a significantly different median egg count (Table 1). All samples with high infection intensity were detected by the Schistoscope (S1C Fig). In addition, the Schistoscope detected 76.5% and 93.0% of samples with microscopy egg count 1–9 eggs/10mL and 10–49 eggs/10mL respectively. On the contrary, the Schistoscope found 21 additional cases with low infection intensity not detected by conventional microscopy. Of the 21 cases, 57.6% and 38.1% were confirmed by real-time PCR and UCP-LF CAA assay respectively.

A substantial to almost perfect qualitative agreement between the Schistoscope and conventional microscopy was observed in study B1 and B2 respectively (Table 2). The agreement between the Schistoscope and the CRS was similar to the agreement between conventional microscopy and the CRS. The sensitivities and specificities of the Schistoscope in study B1 and B2 when conventional microscopy was used as a reference were comparable. Furthermore, a comparable sensitivity and specificity between the Schistoscope and conventional microscopy was observed when evaluating both methods against the CRS.

A very strong correlation between the egg counts of the Schistoscope and conventional microscopy was observed in study B1 (r = 0.87; P<0.0001, Fig 3D) and study B2 (r = 0.93, P<0.0001, Fig 3E). In study B2, a lower though significant correlation was observed between the Schistoscope egg counts and PCR Ct-values (r = -0.82, P<0.0001) and CAA concentration (r = 0.78, P<0.0001) (Fig 3F and 3G). Bland-Altman analysis further demonstrated a strong quantitative agreement between the Schistoscope and conventional microscopy in both study B1 and B2 (Fig 4B and 4C) with a trend in the Schistoscope underestimating egg count.

4. Discussion

For the first time, we demonstrate the sensitivity and specificity of the Schistoscope with an onboard integrated AI on fresh (study A) and stored (study B) sample sets in comparison to conventional microscopy as well as to a more sensitive CRS consisting of real-time PCR and UCP-LF CAA. Five Schistoscopes were successfully transported and implemented in the parasitology laboratory of CERMEL, a reference laboratory setting within a rural part of Gabon, which is a region endemic for S. haematobium. All other diagnostic tests were also performed at CERMEL. Overall, the performance of the Schistoscope was non-inferior to conventional microscopy with a comparable sensitivity and a slightly lower specificity. The Schistoscope is a promising tool for urogenital schistosomiasis screening in endemic settings and offers the advantage of data connectivity and the possibility of task shifting [27–29].

Qualitatively, a moderate to almost perfect agreement between the Schistoscope and conventional microscopy was found while a fair to moderate agreement was observed when compared to the CRS. This lower agreement can mainly be attributed to the fact that the two additional diagnostic tests included in the CRS (PCR and CAA) are more accurate, especially at low infection intensities, and these tend to be missed by the Schistoscope and/or conventional microscopy. The sensitivity of the Schistoscope was found to be non-inferior to conventional microscopy in both study A and B2. The specificity of the Schistoscope was however inferior to conventional microscopy in study A, but comparable in study B2. This is thought to be a consequence of the presence of relatively more artifacts in the freshly prepared slides (study A) compared to stored slides (study B), which the AI algorithm could not differentiate from eggs. Secondly, although samples in study A and B were obtained from the same geographical area in Gabon (Lambaréné and its surrounding villages), they were collected at different time points (~2 years apart) as well as from different populations, i.e. community-based in study A versus specific populations including pregnant women in study B. Differences in urine composition due to differential seasonal concomitant bacterial infections was assumed to explain increase in egg-like crystals formation in urine that interfered with AI detection. Manual re-analysis of the images of a selection of samples that were positive by the Schistoscope but negative by conventional microscopy, revealed that indeed crystals were present in these slides, which the AI incorrectly identified as eggs (S3 Fig).

Although the sensitivity of the Schistoscope in study A (83.1%) was comparable to previously reported results from a field setting in Nigeria (87.3%) [12], the observed specificity was significantly higher (77.0% compared to 48.9% in Nigeria) as well as the correlation between egg counts by the Schistoscope and conventional microscopy, indicating the successful re-designing and re-training of the AI algorithm [21]. The slightly lower correlation observed between the Schistoscope egg count and real-time PCR Ct-values or CAA concentration could be because of the differences in diagnostic target; eggs, egg-DNA and circulating antigen, respectively. The correlation between conventional microscopy and real-time PCR or UCP-LF CAA resulted in a similar observation (S2 Fig). The correlation observed between egg counts by conventional microscopy and Ct-values is comparable to previous findings [30]. Furthermore, although a better correlation would be expected between egg counts and Ct-values (egg-DNA), considering that they are both egg-based detection methods, it is important to note that, an egg does not have a fixed target DNA copy number. This variation is influenced by the egg developmental stage, which could account for the broad spectrum of Ct-values observed across varying infection intensities or egg count [31].

In study B overall, over 50% of the samples had only results for microscopy and the Schistoscope and rather than discarding this number of samples from our data set, it was analysed separately as B1. For both studies, all cases with high infection intensity based on conventional microscopy, known to correlate strongly with morbidity of the disease [32], were detected by the Schistoscope. Following Bland-Altman’s analysis, a constant but clinically fair bias (absolute error) between Schistoscope and conventional microscopy egg counts was observed in both the fresh and stored sample sets, suggesting that the Schistoscope is slightly underestimating egg counts at a constant rate. This could be explained by the fact that with increasing infection intensity, eggs tend to overlap which could not be accurately counted by the AI algorithm (Fig 2C), as also observed previously [12]. Furthermore, the AI algorithm was designed and optimised for specificity at the expense of sensitivity, i.e. it was programmed to refrain from detecting truncated eggs located on boundaries of images so as to reduce the chances of detecting artifacts as well as eggs with lower morphological attributes. Such errors can be corrected by further optimisation of the AI algorithm in order to quantify the number of eggs more accurately. Also, in both the fresh and stored sample set, the majority of missed cases by the Schistoscope had a very low intensity of infection based on conventional microscopy (≤ 5 eggs/10mL), highlighting another area of focus for the next generation of the Schistoscope.

Our results indicated a better sensitivity of the Schistoscope on banked sample slides compared to fresh samples. This could be due to the difference in the infection intensity observed in the two studies. The median egg count, based on conventional microscopy, was lower in the fresh samples compared to the banked samples, which implies that the Schistoscope performs better on samples with a higher infection intensity, as also previously reported [12, 21]. Nevertheless, our results demonstrate a good performance of the Schistoscope on banked sample slides, indicating the possibility for retrospective analysis of banked sample slides in settings lacking direct access to microscopists.

In study A, the sensitivity of conventional microscopy estimated based on the CRS (62%) was lower than the sensitivity (80%) assumed for power calculations, in contrast to study B where the sensitivity (75.2%) observed was comparable. Retrospectively, the sample size calculation was limited in that it did not take into account the proportion of high-intensity infections but only incorporated prevalence, which could have had a significant impact on the sensitivity of conventional microscopy as it is known that the sensitivity of microscopy is limited in case of low intensity infections [33]. Overall, the proportion of high-intensity infections in study A was significantly lower compared to study B, resulting in a lower sensitivity of conventional microscopy as observed in study A. Despite the difference between the assumed and obtained sensitivity of conventional microscopy, we still believe our study had sufficient power to accurately determine the performance of the Schistoscope. So far, the performance of the Schistoscope has been evaluated in two endemic settings in urine samples producing promising outcomes. There is need for more performance evaluation in diverse schistosomiasis endemic settings (in urine and stool) with different climatic conditions, such as the northern part of Sahel region. Also, a cost effective analysis should be performed to support the integration of such a tool in large scale control programmes.

Limitations of this study include the time it took to analyse a slide by the Schistoscope, which on average was ~25 mins. for samples with egg counts ≥200egg/10mL even more time was needed. Furthermore, in this study a filter membrane of diameter 25mm was used (following the standard protocol of CERMEL), which also increased the time of analysis by 3-fold compared to the use of a 13mm filter membrane [12]. If a smaller filter membrane is used and the Schistoscope is programmed to stop counting when reaching 50 eggs/10mL–as this is classified as a high infection intensity and in such cases a detailed egg count is often not required [34]–the total reading time could be reduced to less than 10 mins. A tool as such would complement the existing POC-CCA urine test, which has been recommended by the WHO for S. mansoni infections, in settings co-endemic for S. haematobium. Although the Schistoscope has been fully automated, the aesthetics are currently unsatisfactory [20]. Furthermore, there is need to make the Schistoscope field-friendly and compatible to very rural settings, including the addition of a power source, improving the user interface and making it more compact and portable.

To conclude, in this study a follow-up assessment of the Schistoscope was conducted in a rural laboratory setting in Gabon, further validating its potential as a digital diagnostic tool for the identification and quantification of S. haematobium eggs in freshly collected as well as banked urine sample slides. Although the specificity of the Schistoscope could still be improved, its overall performance was non-inferior to conventional microscopy hence, a promising tool for urogenital schistosomiasis screening in endemic settings.

Supporting information

S1 Checklist. STARD-2015-Checklist.

(DOCX)

pntd.0011967.s001.docx^{(34.2KB, docx)}

S1 Fig

Agreement between the Schistoscope and microscopy per category of infection intensity in study A and B.

(TIF)

pntd.0011967.s002.tif^{(1,022.9KB, tif)}

S2 Fig

Correlation between S. haematobium egg counts measured by the conventional microscopy and Ct-values determined by by real-time PCR (a, c) and urine CAA concentration.

(TIF)

pntd.0011967.s003.tif^{(1.7MB, tif)}

S3 Fig. Image showing crystal incorrectly detected as an egg by the Schistoscope.

(TIF)

pntd.0011967.s004.tif^{(12.5MB, tif)}

S1 Manual. Schistoscope user manual.

(PDF)

pntd.0011967.s005.pdf^{(3.6MB, pdf)}

S1 Video. Video showing the Schistoscopes running in the laboratory.

(MOV)

Download video file^{(16MB, MOV)}

S1 Raw Dataset

Overall raw dataset containing data for Schistoscope validation on fresh urine samples (study A), Banked slides (study B) and quality control of banked slides.

(XLSX)

pntd.0011967.s007.xlsx^{(312.5KB, xlsx)}

Acknowledgments

We extend our appreciation to Mermoz Ndong-Essone Ondong, Danny Carrel Manfoumbi Mabicka, Moutsinga Dalia Coralline ep Lehoumbou, Elsy Myrna N’noh Dansou, Marguerite Nzame Ngome, and the coordination team, along with all the members of Immuno-Epi research group of CERMEL. We thank Bertrand Lell for IT support in the field. Our gratitude also goes to Jean-Aimé Massande Ndzokou’s and the CERMEL field team for their valuable contributions to the advancement of this project.

Data Availability

All relevant data are within the paper and its Supporting Information files.

Funding Statement

This work was funded by NWO-WOTRO Science for Global Development program, grant no. W 07.30318.009 (INSPiRED—INclusive diagnoStics for Poverty REIated parasitic Diseases in Nigeria and Gabon) to LvL. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Adenowo AF, Oyinloye BE, Ogunyinka BI, Kappo AP. Impact of human schistosomiasis in sub-Saharan Africa. Braz J Infect Dis. 2015;19(2):196–205. Epub 20150127. doi: 10.1016/j.bjid.2014.11.004 ; PubMed Central PMCID: PMC9425372. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.World Health Organisation. Schistosomiasis and soil-transmitted helminthiases: treating millions of people, despite the pandemic.: World Health Organisation. https://www.who.int/news/item/08-12-2021-schistosomiasis-and-soil-transmitted-helminthiases-treating-millions-of-people-despite-the-pandemic. Access 06/23; 2020 [cited 2023 23/06]. Available from: https://www.who.int/news/item/08-12-2021-schistosomiasis-and-soil-transmitted-helminthiases-treating-millions-of-people-despite-the-pandemic. [Google Scholar]
3.World Health Organisation. Schistosomiasis: World Health Organisation. https://www.who.int/news-room/fact-sheets/detail/schistosomiasis. Access 01/24; 2024. [cited 2024 24/01]. Available from: https://www.who.int/news-room/fact-sheets/detail/schistosomiasis. [Google Scholar]
4.World Health Organisation. New guideline for the control and elimination of human schistosomiasis.: World Health Organisation. https://www.who.int/publications/i/item/9789240041608; 2022. [cited 2022 22/06]. Available from: https://www.who.int/publications/i/item/9789240041608. [Google Scholar]
5.World Health Organisation. Strengthening diagnostic capacity: World Health Organisation. https://www.eliminateschisto.org/news-events/news/world-health-assembly-resolution-on-strengthening-diagnostics-capacity. Access 07/2023; 2023 [cited 2023 17/07]. Available from: https://www.eliminateschisto.org/news-events/news/world-health-assembly-resolution-on-strengthening-diagnostics-capacity. [Google Scholar]
6.World Health Organisation. Public consultation: Target Product Profiles for diagnostic tests to meet Schistosomiasis and soil-transmitted Helminth programme needs: World Health Organization. https://www.who.int/news-room/articles-detail/public-consultation-target-product-profiles-for-diagnostic-tests-to-meet-schistosomiasis-and-soil-transmitted-helminth-programme-needs. Access 06/2023; 2021 [cited 2023 21/06]. Available from: https://www.who.int/news-room/articles-detail/public-consultation-target-product-profiles-for-diagnostic-tests-to-meet-schistosomiasis-and-soil-transmitted-helminth-programme-needs. [Google Scholar]
7.Corstjens PL, de Dood CJ, Knopp S, Clements MN, Ortu G, Umulisa I, et al. Circulating Anodic Antigen (CAA): a highly sensitive diagnostic biomarker to detect active Schistosoma infections—improvement and use during SCORE. The American journal of tropical medicine and hygiene. 2020;103(1 Suppl):50. doi: 10.4269/ajtmh.19-0819 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Obeng BB, Aryeetey YA, de Dood CJ, Amoah AS, Larbi IA, Deelder AM, et al. Application of a circulating-cathodic-antigen (CCA) strip test and real-time PCR, in comparison with microscopy, for the detection of Schistosoma haematobium in urine samples from Ghana. Ann Trop Med Parasitol. 2008;102(7):625–33. doi: 10.1179/136485908x337490 . [DOI] [PubMed] [Google Scholar]
9.Ward P, Dahlberg P, Lagatie O, Larsson J, Tynong A, Vlaminck J, et al. Affordable artificial intelligence-based digital pathology for neglected tropical diseases: A proof-of-concept for the detection of soil-transmitted helminths and Schistosoma mansoni eggs in Kato-Katz stool thick smears. PLoS Negl Trop Dis. 2022;16(6):e0010500. Epub 20220617. doi: 10.1371/journal.pntd.0010500 ; PubMed Central PMCID: PMC9258839. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Coulibaly JT, Silue KD, Armstrong M, de León Derby MD, D’Ambrosio MV, Fletcher DA, et al. High Sensitivity of Mobile Phone Microscopy Screening for Schistosoma haematobium in Azaguié, Côte d’Ivoire. The American journal of tropical medicine and hygiene. 2023;108(1):41. doi: 10.4269/ajtmh.22-0527 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Meulah B, Bengtson M, Van Lieshout L, Hokke CH, Kreidenweiss A, Diehl J-C, et al. A review on innovative optical devices for the diagnosis of human soil-transmitted helminthiasis and schistosomiasis: From research and development to commercialisation. Parasitology. 2022:1–41. doi: 10.1017/S0031182022001664 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Meulah B, Oyibo P, Bengtson M, Agbana T, Lontchi RAL, Adegnika AA, et al. Performance evaluation of the Schistoscope 5.0 for (semi-) automated digital detection and quantification of Schistosoma haematobium eggs in urine: A field-based study in Nigeria. Am J Trop Med Hyg. 2022;107(5):1047–54. doi: 10.4269/ajtmh.22-0276 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Agrebi S, Larbi A. Use of artificial intelligence in infectious diseases. Artificial intelligence in precision health: Elsevier; 2020. p. 415–38. [Google Scholar]
14.Brownstein JS, Rader B, Astley CM, Tian H. Advances in Artificial Intelligence for infectious-disease surveillance. New England Journal of Medicine. 2023;388(17):1597–607. doi: 10.1056/NEJMra2119215 [DOI] [PubMed] [Google Scholar]
15.Shenoy S, Rajan AK, Rashid M, Chandran VP, Poojari PG, Kunhikatta V, et al. Artificial intelligence in differentiating tropical infections: A step ahead. PLoS Neglected Tropical Diseases. 2022;16(6):e0010455. doi: 10.1371/journal.pntd.0010455 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Prieto-Egido I, González-Escalada A, García-Giganto V, Martínez-Fernández A. Design of New Procedures for Diagnosing Prevalent Diseases Using a Low-Cost Telemicroscopy System. Telemed J E Health. 2016;22(11):952–9. Epub 20160420. doi: 10.1089/tmj.2015.0208 . [DOI] [PubMed] [Google Scholar]
17.Dacal E, Bermejo-Peláez D, Lin L, Álamo E, Cuadrado D, Martínez Á, et al. Mobile microscopy and telemedicine platform assisted by deep learning for the quantification of Trichuris trichiura infection. PLoS Negl Trop Dis. 2021;15(9):e0009677. Epub 20210907. doi: 10.1371/journal.pntd.0009677 ; PubMed Central PMCID: PMC8448303. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Onasanya A, Bengtson M, de Goeje L, Van Engelen J, Diehl JC, van Lieshout L. Developing inclusive digital health diagnostic for schistosomiasis: a need for guidance via target product profiles. Frontiers in Parasitology. 2023;2. doi: 10.3389/fpara.2023.1255848 [DOI] [Google Scholar]
19.Oyibo P, Jujjavarapu S, Meulah B, Agbana T, Braakman I, van Diepen A, et al. Schistoscope: An Automated Microscope with Artificial Intelligence for Detection of Schistosoma haematobium Eggs in Resource-Limited Settings. Micromachines (Basel). 2022;13(5). Epub 20220419. doi: 10.3390/mi13050643 ; PubMed Central PMCID: PMC9146062. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Bengtson M, Onasanya A, Oyibo P, Meulah B, Samenjo KT, Braakman I, et al., editors. A usability study of an innovative optical device for the diagnosis of schistosomiasis in Nigeria. 2022 IEEE Global Humanitarian Technology Conference (GHTC); 2022: IEEE. [Google Scholar]
21.Oyibo P, Meulah B, Bengtson M, Lisette van L, Oyibo W, Diehle J-C, et al. Two-stage automated diagnosis framework for urogenital schistosomiasis in microscopy images from low-resource settings. Journal of Medical Imaging. 2023;10(4):044005. doi: 10.1117/1.JMI.10.4.044005 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Dejon-Agobé JC, Honkpehedji YJ, Zinsou JF, Edoa JR, Adégbitè BR, Mangaboula A, et al. Epidemiology of schistosomiasis and soil-transmitted helminth coinfections among schoolchildren living in Lambaréné, Gabon. The American Journal of Tropical Medicine and Hygiene. 2020;103(1):325. doi: 10.4269/ajtmh.19-0835 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Machin D, Campbell MJ, Tan SB, Tan SH. Sample size tables for clinical studies: John Wiley & Sons; 2011. [Google Scholar]
24.Kenguele HM, Adegnika AA, Nkoma AM, Ateba-Ngoa U, Mbong M, Zinsou J, et al. Impact of short-time urine freezing on the sensitivity of an established Schistosoma real-time PCR assay. Am J Trop Med Hyg. 2014;90(6):1153–5. Epub 20140407. doi: 10.4269/ajtmh.14-0005 ; PubMed Central PMCID: PMC4047745. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Wiegand RE, Secor WE, Fleming FM, French MD, King CH, Deol AK, et al. Associations between infection intensity categories and morbidity prevalence in school-age children are much stronger for Schistosoma haematobium than for S. mansoni. PLoS Negl Trop Dis. 2021;15(5):e0009444. Epub 20210525. doi: 10.1371/journal.pntd.0009444 ; PubMed Central PMCID: PMC8183985. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Organization WH. Prevention and control of schistosomiasis and soil-transmitted helminthiasis: report of a WHO expert committee: World Health Organization; 2002. [PubMed] [Google Scholar]
27.Tilahun B, Gashu KD, Mekonnen ZA, Endehabtu BF, Angaw DA. Mapping the role of digital health technologies in the case detection, management, and treatment outcomes of neglected tropical diseases: a scoping review. Tropical medicine and health. 2021;49:1–10. doi: 10.1186/s41182-021-00307-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Mustafa U-k Kreppel KS, Brinkel J, Sauli E. Digital Technologies to Enhance Infectious Disease Surveillance in Tanzania: A Scoping Review. Healthcare. 2023;11(4):470. doi: 10.3390/healthcare11040470 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Onasanya A, Bengtson M, Agbana T, Oladunni O, van Engelen J, Oladepo O, et al. Towards Inclusive Diagnostics for Neglected Tropical Diseases: User Experience of a New Digital Diagnostic Device in Low-Income Settings. Tropical Medicine and Infectious Disease. 2023;8(3):176. doi: 10.3390/tropicalmed8030176 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Vinkeles Melchers NV, van Dam GJ, Shaproski D, Kahama AI, Brienen EA, Vennervald BJ, et al. Diagnostic performance of Schistosoma real-time PCR in urine samples from Kenyan children infected with Schistosoma haematobium: day-to-day variation and follow-up after praziquantel treatment. PLoS Negl Trop Dis. 2014;8(4):e2807. Epub 20140417. doi: 10.1371/journal.pntd.0002807 ; PubMed Central PMCID: PMC3990496. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Keller D, Rothen J, Dangy JP, Saner C, Daubenberger C, Allan F, et al. Performance of a real-time PCR approach for diagnosing Schistosoma haematobium infections of different intensity in urine samples from Zanzibar. Infect Dis Poverty. 2020;9(1):128. Epub 20200904. doi: 10.1186/s40249-020-00726-y ; PubMed Central PMCID: PMC7487541. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Kura K, Hardwick RJ, Truscott JE, Toor J, Hollingsworth TD, Anderson RM. The impact of mass drug administration on Schistosoma haematobium infection: what is required to achieve morbidity control and elimination? Parasites & Vectors. 2020;13(1):1–10. doi: 10.1186/s13071-020-04409-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Hoekstra PT, van Dam GJ, van Lieshout L. Context-Specific Procedures for the Diagnosis of Human Schistosomiasis–A Mini Review. Frontiers in Tropical Diseases. 2021;2. doi: 10.3389/fitd.2021.722438 [DOI] [Google Scholar]
34.World Health Organization. Prevention and control of schistosomiasis and soil-transmitted helminthiasis: report of a WHO expert committee: World Health Organization. https://apps.who.int/iris/handle/10665/42588; 2002. [PubMed] [Google Scholar]

PLoS Negl Trop Dis. doi: 10.1371/journal.pntd.0011967.r001

Decision Letter 0

Gabriel Rinaldi, Aaron R Jex

2 Jan 2024

Dear Mr Meulah,

Thank you very much for submitting your manuscript "Extended laboratory validation of the performance of an artificial intelligence-based digital microscope (Schistoscope) in Lambaréné (Gabon) for automated detection of Schistosoma haematobium eggs in urine" for consideration at PLOS Neglected Tropical Diseases. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

In this study Meulah and colleagues optimise a quantitative methodology to accurately diagnose schistosomiasis. Even though the study is timely and within the scope of the journal, several issues raised by the three reviewers and the editor need first to be thoroughly addressed before considering it suitable for publication.

• The Title is probably too long, it could be shorter and punchier. Please consider something along these lines:

“Artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine.”

• Line 139: “The Schistoscope was assumed to have a sensitivity and specificity non-inferior to conventional microscopy, which were realistically assumed to be 80% and 90%, respectively based on field expert estimates using real-time PCR and UCP-LF CAA.” Provide a reference to back up the sensitivity and specificity of conventional microscopy.

• Line 247: Please, provide reference to support the selection criteria for levels of infection intensity.

• Results headings – consider writing headings that are more descriptive and highlight the main/ key findings described in the section

• PCR; is the positive control the same as the internal control? What kind of normalization is used during the qPCR analysis (ratio unknown: control ? What were the Cq values of the NTC? Did the authors run a standard curve along with controls?). All these technical points are extremely important not only because this study focuses on describing a new diagnostic methodology, but for the sake of transparency and reproducibility.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Gabriel Rinaldi, M.D., Ph.D.

Academic Editor

PLOS Neglected Tropical Diseases

Aaron Jex

Section Editor

PLOS Neglected Tropical Diseases

***********************

• The Title is probably too long, it could be shorter and punchier. Please consider something along these lines:

“Artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine.”

• Line 247: Please, provide reference to support the selection criteria for levels of infection intensity.

• Results headings – consider writing headings that are more descriptive and highlight the main/ key findings described in the section

Reviewer's Responses to Questions

Key Review Criteria Required for Acceptance?

As you describe the new analyses required for acceptance, please consider the following:

Methods

-Are the objectives of the study clearly articulated with a clear testable hypothesis stated?

-Is the study design appropriate to address the stated objectives?

-Is the population clearly described and appropriate for the hypothesis being tested?

-Is the sample size sufficient to ensure adequate power to address the hypothesis being tested?

-Were correct statistical analysis used to support conclusions?

-Are there concerns about ethical or regulatory requirements being met?

Reviewer #1: Yes

Reviewer #2: Minor additions of detail are needed

• The legend for figure 1 could be more descriptive

• Line 152 – you say Whatman membranes were used to capture the eggs but how was this done e.g. using a syringe and filter holder ?

• Line 213-224 – some more detail is needed

o What is DEC

o What was the internal control amounts used (DNA, primers and probe)

o What was the Schisto positive control

o Why was 40 cycles used as the cut off

o What was the CT value of the internal control

o How has the internal control been validated as a multiplex with the ITS primers?

• Line 243 – what is CRS ? write in full before using the acronym

Reviewer #3: The objectives of the study are clearly articulated with a clear testable hypothesis, and correct statistical analysis were used to support conclusions stated. However, the study design is not quite clear. To me, it is not necessary to have the B1 and B2 components; Just having fresh samples and banked samples should be enough to avoid unnecessary complexity. The B1 component can be deleted.

Also, the sample size calculation seems not appropriate, it seems like the formula for a cross sectional survey was used for sample size calculations. I didn’t attempt to calculate the sample size but I do think that a different formula integrating the minimal difference expected between both tests (Schistoscope vs microscopy and Schistoscope vs DNA- or Antigen-based assays) should be used (please see Page 10, Lines 245-247 for the hypothesis).

Regarding the ethical regulatory requirements (please see Page 7, Lines 135-137), this is not a clinical trial, and I am wondering why this study was registered on clinicaltrials.gov. If this study just took advantage of an existing study, it is worth to mention it. Also, the authors declared that they provided treatments to all infected individuals. Knowing that the test and treat procedures are sometimes complicated, especially when the test cannot be completed in a few minutes, I am curious to know how the authors managed to find and treat all those positive patients. This can be helpful for other studies.

--------------------

Results

-Does the analysis presented match the analysis plan?

-Are the results clearly and completely presented?

-Are the figures (Tables, Images) of sufficient quality for clarity?

Reviewer #1: Yes

Reviewer #2: A minor addition to the results would be beneficial

• Did you find other discrepancies that are challenging to interpret – eg egg positive but CAA neg and DNA neg.

Reviewer #3: The results are clearly presented and match the analysis plan. However, one of the main assets of this technology/device is the ability to be used for the diagnosis of schistosomiasis on banked samples. However, the design developed by the authors do not fully enable to better capture that asset. Indeed, it would have been interesting to have the results of conventional microscopy for all the samples, not only for a subset of samples used as quality control. Even if quality control should be considered, I am quite curious to have the results of the quality control to be able to figure out to what extent was the variation between the two readings, and whether it was unidirectional or random. This should be clearly presented and discussed.

Also, the issue of artefact raised at page 20, Lines 350-358 can be important for conventional microscopy as well. As such, it would have been worth to analyze all the banked slides, or present and discuss the results of quality control.

Page 9, Lines 191-193. It is unclear how the discrepancies between conventional microscopy readings between the two readers was done? Was a third reader involved as adjudicator or a third reading was done by the same two microscopists? This should be stated clearly.

--------------------

Conclusions

-Are the conclusions supported by the data presented?

-Are the limitations of analysis clearly described?

-Do the authors discuss how these data can be helpful to advance our understanding of the topic under study?

-Is public health relevance addressed?

Reviewer #1: Yes

Reviewer #2: The conclusions are well supported and the limitations are clearly described. I think the thoughts around the egg number cutoff to categorise intensity rather than actual egg counts is a good way forward.

other minor comment are:

• Line 381 – could the difference also relate to the fact you did not bead beat the eggs so the DNA may not have been released ?

• Mention the potential for screening stool?

Reviewer #3: The conclusions are globally supported by the data presented. However, there are a number of issues that need to be addressed:

First and foremost, the main conclusion or finding of the manuscript is not convincing enough. In fact, the authors found that Schistoscope is non inferior to conventional microscopy which is already a poor diagnostic test especially when it comes to talk about transmission interruption or elimination. It is well known and widely accepted that when the parasitic loads are low, the sensitivity of microscopy is low and more sensitive tools are therefore needed. This should be raised as a limitation of the study/manuscript. I would like to acknowledge that Schistoscope has a good potential, and the tool can still be improved. Indeed, I had the opportunity to review the manuscript presenting the early versions of the device which was clearly inferior to microscopy. The current version of the device, even if non-inferior to conventional microscopy, is significantly less performant than DNA- and antigen-based assays (please see figures at page 16), and it might be worth to clearly indicate the potential applications of Schistoscope.

It is unclear to me how the authors try to justify the difference between Schistoscope and PCR or CAA (Page 20, Lines 372-375). There is not just a matter of target but sensitivity and specificity of the techniques. This should be clearly explained.

The authors mentioned the limitations (time of Schistoscope operation) of the study (Page 22, Lines 420-433) that raised another concern or issue. It would have been worth to compare the time of operations of Schistoscope to that of conventional microscopy in the same circumstances. This is important and aligns with the applications of Schistoscope mentioned above.

--------------------

Editorial and Data Presentation Modifications?

Use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity. If the only modifications needed are minor and/or editorial, you may wish to recommend “Minor Revision” or “Accept”.

Reviewer #1: Minor Revision

Reviewer #2: (No Response)

Reviewer #3: This manuscript needs some language editing or correction of typos. For example, please consider “as previously described” not “described previously” (page 9 line 209); “internal transcribed spacer” not “internally transcribed spacer” (page 9 line 210); “ct” not “cq” (page 9 line 233) …

Authors should avoid some of abbreviations or explain them at their first used. For example “UCP-LF-CAA” and “CRS” in the abstract …

Some of the references used in the manuscript are not appropriately presented. For example, WHO should be numbered as all other references …

--------------------

Summary and General Comments

Use this section to provide overall comments, discuss strengths/weaknesses of the study, novelty, significance, general execution and scholarship. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. If requesting major revision, please articulate the new experiments that are needed.

Reviewer #1: See attached file

Reviewer #2: This study fully evaluates the sensitivity and specificity of the Schisoscope for eggs detection in urine. This is a large and detailed study that warrants publication. The methods are sound and the data interpretation has been well executed. I look forward to further developments of the Schistoscope as its limiations are addressed.

Reviewer #3: The manuscript submitted by Meulah and colleagues entitled “Extended laboratory validation of the performance of an artificial intelligence-based digital microscope (Schistoscope) in Lambaréné (Gabon) for automated detection of Schistosoma haematobium eggs in urine” aimed to validate the performance of a novel AI-based diagnostic tool, the Schistoscope, for the detection and quantification of Schistosoma haematobium eggs in urine using conventional microscopy and a composite diagnosis (real-time PCR and a lateral flow assay) as a reference. Although the accurate diagnosis of schistosomiasis is of high interest and an up-to-date issue in the framework of transmission interruption/elimination (in its 2021-2030 roadmap, one of the critical actions recommended by the WHO is the development of diagnostic tests, including standardized point-of-care diagnostic), the paper lacks clarity, especially in its design.

--------------------

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Jean T. Coulibaly

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Attachment

Submitted filename: Reviewer Comment (30 11 2023).docx

pntd.0011967.s008.docx^{(13.6KB, docx)}

PLoS Negl Trop Dis. 2024 Feb 23;18(2):e0011967. doi: 10.1371/journal.pntd.0011967.r002

Author response to Decision Letter 0

26 Jan 2024

Attachment

Submitted filename: Response to reviewers comments.docx

pntd.0011967.s009.docx^{(51.7KB, docx)}

PLoS Negl Trop Dis. doi: 10.1371/journal.pntd.0011967.r003

Decision Letter 1

Gabriel Rinaldi, Aaron R Jex

5 Feb 2024

Dear Mr Meulah,

We are pleased to inform you that your manuscript 'Validation of artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine in Gabon' has been provisionally accepted for publication in PLOS Neglected Tropical Diseases.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Neglected Tropical Diseases.

Best regards,

Gabriel Rinaldi, M.D., Ph.D.

Academic Editor

PLOS Neglected Tropical Diseases

Aaron Jex

Section Editor

PLOS Neglected Tropical Diseases

***********************************************************

PLoS Negl Trop Dis. doi: 10.1371/journal.pntd.0011967.r004

Acceptance letter

Gabriel Rinaldi, Aaron R Jex

19 Feb 2024

Dear Mr Meulah,

We are delighted to inform you that your manuscript, "Validation of artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine in Gabon," has been formally accepted for publication in PLOS Neglected Tropical Diseases.

We have now passed your article onto the PLOS Production Department who will complete the rest of the publication process. All authors will receive a confirmation email upon publication.

The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any scientific or type-setting errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript. Note: Proofs for Front Matter articles (Editorial, Viewpoint, Symposium, Review, etc...) are generated on a different schedule and may not be made available as quickly.

Soon after your final files are uploaded, the early version of your manuscript will be published online unless you opted out of this process. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers.

Thank you again for supporting open-access publishing; we are looking forward to publishing your work in PLOS Neglected Tropical Diseases.

Best regards,

Shaden Kamhawi

co-Editor-in-Chief

PLOS Neglected Tropical Diseases

Paul Brindley

co-Editor-in-Chief

PLOS Neglected Tropical Diseases

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Checklist. STARD-2015-Checklist.

(DOCX)

pntd.0011967.s001.docx^{(34.2KB, docx)}

S1 Fig

Agreement between the Schistoscope and microscopy per category of infection intensity in study A and B.

(TIF)

pntd.0011967.s002.tif^{(1,022.9KB, tif)}

S2 Fig

Correlation between S. haematobium egg counts measured by the conventional microscopy and Ct-values determined by by real-time PCR (a, c) and urine CAA concentration.

(TIF)

pntd.0011967.s003.tif^{(1.7MB, tif)}

S3 Fig. Image showing crystal incorrectly detected as an egg by the Schistoscope.

(TIF)

pntd.0011967.s004.tif^{(12.5MB, tif)}

S1 Manual. Schistoscope user manual.

(PDF)

pntd.0011967.s005.pdf^{(3.6MB, pdf)}

S1 Video. Video showing the Schistoscopes running in the laboratory.

(MOV)

Download video file^{(16MB, MOV)}

S1 Raw Dataset

Overall raw dataset containing data for Schistoscope validation on fresh urine samples (study A), Banked slides (study B) and quality control of banked slides.

(XLSX)

pntd.0011967.s007.xlsx^{(312.5KB, xlsx)}

Attachment

Submitted filename: Reviewer Comment (30 11 2023).docx

pntd.0011967.s008.docx^{(13.6KB, docx)}

Attachment

Submitted filename: Response to reviewers comments.docx

pntd.0011967.s009.docx^{(51.7KB, docx)}

Data Availability Statement

All relevant data are within the paper and its Supporting Information files.

[pntd.0011967.ref001] 1.Adenowo AF, Oyinloye BE, Ogunyinka BI, Kappo AP. Impact of human schistosomiasis in sub-Saharan Africa. Braz J Infect Dis. 2015;19(2):196–205. Epub 20150127. doi: 10.1016/j.bjid.2014.11.004 ; PubMed Central PMCID: PMC9425372. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref002] 2.World Health Organisation. Schistosomiasis and soil-transmitted helminthiases: treating millions of people, despite the pandemic.: World Health Organisation. https://www.who.int/news/item/08-12-2021-schistosomiasis-and-soil-transmitted-helminthiases-treating-millions-of-people-despite-the-pandemic. Access 06/23; 2020 [cited 2023 23/06]. Available from: https://www.who.int/news/item/08-12-2021-schistosomiasis-and-soil-transmitted-helminthiases-treating-millions-of-people-despite-the-pandemic. [Google Scholar]

[pntd.0011967.ref003] 3.World Health Organisation. Schistosomiasis: World Health Organisation. https://www.who.int/news-room/fact-sheets/detail/schistosomiasis. Access 01/24; 2024. [cited 2024 24/01]. Available from: https://www.who.int/news-room/fact-sheets/detail/schistosomiasis. [Google Scholar]

[pntd.0011967.ref004] 4.World Health Organisation. New guideline for the control and elimination of human schistosomiasis.: World Health Organisation. https://www.who.int/publications/i/item/9789240041608; 2022. [cited 2022 22/06]. Available from: https://www.who.int/publications/i/item/9789240041608. [Google Scholar]

[pntd.0011967.ref005] 5.World Health Organisation. Strengthening diagnostic capacity: World Health Organisation. https://www.eliminateschisto.org/news-events/news/world-health-assembly-resolution-on-strengthening-diagnostics-capacity. Access 07/2023; 2023 [cited 2023 17/07]. Available from: https://www.eliminateschisto.org/news-events/news/world-health-assembly-resolution-on-strengthening-diagnostics-capacity. [Google Scholar]

[pntd.0011967.ref006] 6.World Health Organisation. Public consultation: Target Product Profiles for diagnostic tests to meet Schistosomiasis and soil-transmitted Helminth programme needs: World Health Organization. https://www.who.int/news-room/articles-detail/public-consultation-target-product-profiles-for-diagnostic-tests-to-meet-schistosomiasis-and-soil-transmitted-helminth-programme-needs. Access 06/2023; 2021 [cited 2023 21/06]. Available from: https://www.who.int/news-room/articles-detail/public-consultation-target-product-profiles-for-diagnostic-tests-to-meet-schistosomiasis-and-soil-transmitted-helminth-programme-needs. [Google Scholar]

[pntd.0011967.ref007] 7.Corstjens PL, de Dood CJ, Knopp S, Clements MN, Ortu G, Umulisa I, et al. Circulating Anodic Antigen (CAA): a highly sensitive diagnostic biomarker to detect active Schistosoma infections—improvement and use during SCORE. The American journal of tropical medicine and hygiene. 2020;103(1 Suppl):50. doi: 10.4269/ajtmh.19-0819 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref008] 8.Obeng BB, Aryeetey YA, de Dood CJ, Amoah AS, Larbi IA, Deelder AM, et al. Application of a circulating-cathodic-antigen (CCA) strip test and real-time PCR, in comparison with microscopy, for the detection of Schistosoma haematobium in urine samples from Ghana. Ann Trop Med Parasitol. 2008;102(7):625–33. doi: 10.1179/136485908x337490 . [DOI] [PubMed] [Google Scholar]

[pntd.0011967.ref009] 9.Ward P, Dahlberg P, Lagatie O, Larsson J, Tynong A, Vlaminck J, et al. Affordable artificial intelligence-based digital pathology for neglected tropical diseases: A proof-of-concept for the detection of soil-transmitted helminths and Schistosoma mansoni eggs in Kato-Katz stool thick smears. PLoS Negl Trop Dis. 2022;16(6):e0010500. Epub 20220617. doi: 10.1371/journal.pntd.0010500 ; PubMed Central PMCID: PMC9258839. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref010] 10.Coulibaly JT, Silue KD, Armstrong M, de León Derby MD, D’Ambrosio MV, Fletcher DA, et al. High Sensitivity of Mobile Phone Microscopy Screening for Schistosoma haematobium in Azaguié, Côte d’Ivoire. The American journal of tropical medicine and hygiene. 2023;108(1):41. doi: 10.4269/ajtmh.22-0527 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref011] 11.Meulah B, Bengtson M, Van Lieshout L, Hokke CH, Kreidenweiss A, Diehl J-C, et al. A review on innovative optical devices for the diagnosis of human soil-transmitted helminthiasis and schistosomiasis: From research and development to commercialisation. Parasitology. 2022:1–41. doi: 10.1017/S0031182022001664 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref012] 12.Meulah B, Oyibo P, Bengtson M, Agbana T, Lontchi RAL, Adegnika AA, et al. Performance evaluation of the Schistoscope 5.0 for (semi-) automated digital detection and quantification of Schistosoma haematobium eggs in urine: A field-based study in Nigeria. Am J Trop Med Hyg. 2022;107(5):1047–54. doi: 10.4269/ajtmh.22-0276 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref013] 13.Agrebi S, Larbi A. Use of artificial intelligence in infectious diseases. Artificial intelligence in precision health: Elsevier; 2020. p. 415–38. [Google Scholar]

[pntd.0011967.ref014] 14.Brownstein JS, Rader B, Astley CM, Tian H. Advances in Artificial Intelligence for infectious-disease surveillance. New England Journal of Medicine. 2023;388(17):1597–607. doi: 10.1056/NEJMra2119215 [DOI] [PubMed] [Google Scholar]

[pntd.0011967.ref015] 15.Shenoy S, Rajan AK, Rashid M, Chandran VP, Poojari PG, Kunhikatta V, et al. Artificial intelligence in differentiating tropical infections: A step ahead. PLoS Neglected Tropical Diseases. 2022;16(6):e0010455. doi: 10.1371/journal.pntd.0010455 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref016] 16.Prieto-Egido I, González-Escalada A, García-Giganto V, Martínez-Fernández A. Design of New Procedures for Diagnosing Prevalent Diseases Using a Low-Cost Telemicroscopy System. Telemed J E Health. 2016;22(11):952–9. Epub 20160420. doi: 10.1089/tmj.2015.0208 . [DOI] [PubMed] [Google Scholar]

[pntd.0011967.ref017] 17.Dacal E, Bermejo-Peláez D, Lin L, Álamo E, Cuadrado D, Martínez Á, et al. Mobile microscopy and telemedicine platform assisted by deep learning for the quantification of Trichuris trichiura infection. PLoS Negl Trop Dis. 2021;15(9):e0009677. Epub 20210907. doi: 10.1371/journal.pntd.0009677 ; PubMed Central PMCID: PMC8448303. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref018] 18.Onasanya A, Bengtson M, de Goeje L, Van Engelen J, Diehl JC, van Lieshout L. Developing inclusive digital health diagnostic for schistosomiasis: a need for guidance via target product profiles. Frontiers in Parasitology. 2023;2. doi: 10.3389/fpara.2023.1255848 [DOI] [Google Scholar]

[pntd.0011967.ref019] 19.Oyibo P, Jujjavarapu S, Meulah B, Agbana T, Braakman I, van Diepen A, et al. Schistoscope: An Automated Microscope with Artificial Intelligence for Detection of Schistosoma haematobium Eggs in Resource-Limited Settings. Micromachines (Basel). 2022;13(5). Epub 20220419. doi: 10.3390/mi13050643 ; PubMed Central PMCID: PMC9146062. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref020] 20.Bengtson M, Onasanya A, Oyibo P, Meulah B, Samenjo KT, Braakman I, et al., editors. A usability study of an innovative optical device for the diagnosis of schistosomiasis in Nigeria. 2022 IEEE Global Humanitarian Technology Conference (GHTC); 2022: IEEE. [Google Scholar]

[pntd.0011967.ref021] 21.Oyibo P, Meulah B, Bengtson M, Lisette van L, Oyibo W, Diehle J-C, et al. Two-stage automated diagnosis framework for urogenital schistosomiasis in microscopy images from low-resource settings. Journal of Medical Imaging. 2023;10(4):044005. doi: 10.1117/1.JMI.10.4.044005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref022] 22.Dejon-Agobé JC, Honkpehedji YJ, Zinsou JF, Edoa JR, Adégbitè BR, Mangaboula A, et al. Epidemiology of schistosomiasis and soil-transmitted helminth coinfections among schoolchildren living in Lambaréné, Gabon. The American Journal of Tropical Medicine and Hygiene. 2020;103(1):325. doi: 10.4269/ajtmh.19-0835 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref023] 23.Machin D, Campbell MJ, Tan SB, Tan SH. Sample size tables for clinical studies: John Wiley & Sons; 2011. [Google Scholar]

[pntd.0011967.ref024] 24.Kenguele HM, Adegnika AA, Nkoma AM, Ateba-Ngoa U, Mbong M, Zinsou J, et al. Impact of short-time urine freezing on the sensitivity of an established Schistosoma real-time PCR assay. Am J Trop Med Hyg. 2014;90(6):1153–5. Epub 20140407. doi: 10.4269/ajtmh.14-0005 ; PubMed Central PMCID: PMC4047745. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref025] 25.Wiegand RE, Secor WE, Fleming FM, French MD, King CH, Deol AK, et al. Associations between infection intensity categories and morbidity prevalence in school-age children are much stronger for Schistosoma haematobium than for S. mansoni. PLoS Negl Trop Dis. 2021;15(5):e0009444. Epub 20210525. doi: 10.1371/journal.pntd.0009444 ; PubMed Central PMCID: PMC8183985. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref026] 26.Organization WH. Prevention and control of schistosomiasis and soil-transmitted helminthiasis: report of a WHO expert committee: World Health Organization; 2002. [PubMed] [Google Scholar]

[pntd.0011967.ref027] 27.Tilahun B, Gashu KD, Mekonnen ZA, Endehabtu BF, Angaw DA. Mapping the role of digital health technologies in the case detection, management, and treatment outcomes of neglected tropical diseases: a scoping review. Tropical medicine and health. 2021;49:1–10. doi: 10.1186/s41182-021-00307-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref028] 28.Mustafa U-k Kreppel KS, Brinkel J, Sauli E. Digital Technologies to Enhance Infectious Disease Surveillance in Tanzania: A Scoping Review. Healthcare. 2023;11(4):470. doi: 10.3390/healthcare11040470 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref029] 29.Onasanya A, Bengtson M, Agbana T, Oladunni O, van Engelen J, Oladepo O, et al. Towards Inclusive Diagnostics for Neglected Tropical Diseases: User Experience of a New Digital Diagnostic Device in Low-Income Settings. Tropical Medicine and Infectious Disease. 2023;8(3):176. doi: 10.3390/tropicalmed8030176 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref030] 30.Vinkeles Melchers NV, van Dam GJ, Shaproski D, Kahama AI, Brienen EA, Vennervald BJ, et al. Diagnostic performance of Schistosoma real-time PCR in urine samples from Kenyan children infected with Schistosoma haematobium: day-to-day variation and follow-up after praziquantel treatment. PLoS Negl Trop Dis. 2014;8(4):e2807. Epub 20140417. doi: 10.1371/journal.pntd.0002807 ; PubMed Central PMCID: PMC3990496. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref031] 31.Keller D, Rothen J, Dangy JP, Saner C, Daubenberger C, Allan F, et al. Performance of a real-time PCR approach for diagnosing Schistosoma haematobium infections of different intensity in urine samples from Zanzibar. Infect Dis Poverty. 2020;9(1):128. Epub 20200904. doi: 10.1186/s40249-020-00726-y ; PubMed Central PMCID: PMC7487541. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref032] 32.Kura K, Hardwick RJ, Truscott JE, Toor J, Hollingsworth TD, Anderson RM. The impact of mass drug administration on Schistosoma haematobium infection: what is required to achieve morbidity control and elimination? Parasites & Vectors. 2020;13(1):1–10. doi: 10.1186/s13071-020-04409-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pntd.0011967.ref033] 33.Hoekstra PT, van Dam GJ, van Lieshout L. Context-Specific Procedures for the Diagnosis of Human Schistosomiasis–A Mini Review. Frontiers in Tropical Diseases. 2021;2. doi: 10.3389/fitd.2021.722438 [DOI] [Google Scholar]

[pntd.0011967.ref034] 34.World Health Organization. Prevention and control of schistosomiasis and soil-transmitted helminthiasis: report of a WHO expert committee: World Health Organization. https://apps.who.int/iris/handle/10665/42588; 2002. [PubMed] [Google Scholar]

PERMALINK

Validation of artificial intelligence-based digital microscopy for automated detection of Schistosoma haematobium eggs in urine in Gabon

Brice Meulah

Prosper Oyibo

Pytsje T Hoekstra

Paul Alvyn Nguema Moure

Moustapha Nzamba Maloum

Romeo Aime Laclong-Lontchi

Yabo Josiane Honkpehedji

Michel Bengtson

Cornelis Hokke

Paul L A M Corstjens

Temitope Agbana

Jan Carel Diehl

Ayola Akim Adegnika

Lisette van Lieshout

Roles

Abstract

Introduction

Methods

Results

Conclusion

Trial registration

Author summary

1. Introduction

2. Methods

2.1. Ethics statement

2.2. Study design

Fig 1. Comprehensive flow chart detailing the methodical sequence of urine sample collection, processing by the Schistoscope, conventional microscopy, real-time PCR and UCP-LF CAA and data analysis.

2.3. Sample size calculations

2.4. Sample collection and processing

2.5. Diagnostic methods

2.5.1. The Schistoscope

Fig 2.

2.5.2.Conventional microscopy

2.5.3. Nucleic acid extraction and real-time PCR

2.5.4. UCP-LF CAA

2.6. Statistical analyses

3. Results

3.1. Study A: Diagnostic performance of the Schistoscope on freshly prepared samples

Table 1. Diagnostic outcomes of the Schistoscope in comparison to conventional microscopy, real-time PCR and UCP-LF CAA in study A and B.

Table 2. Diagnostic performance and pairwise level of agreement by Cohen’s Kappa statistics between the Schistoscope and conventional microscopy and the composite reference for the detection of S. haematobium infection in study A and B.

Fig 3.

Fig 4.

3.2. Study B: Diagnostic performance of the Schistoscope on banked samples

4. Discussion

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Gabriel Rinaldi

Aaron R Jex

Roles

Author response to Decision Letter 0

Decision Letter 1

Gabriel Rinaldi

Aaron R Jex

Roles

Acceptance letter

Gabriel Rinaldi

Aaron R Jex

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases