The Challenge of Small Intestine Length Measurement: A Systematic Review of Imaging Techniques

Abstract

Background

Short bowel syndrome is the most common cause of intestinal failure, with morbidity and mortality linked to remanent small intestine length. There is no current standard for non-invasive bowel length measurement.

Materials and Methods

The literature was systematically searched for articles describing measurement of small intestine length from radiographic studies. Inclusion required reporting intestinal length as an outcome and use of diagnostic imaging for length assessment compared to a ground truth. Two reviewers independently screened studies for inclusion, extracted data, and assessed study quality.

Results

Eleven studies met inclusion criteria and reported small intestinal length measurement using four imaging modalities: barium follow-through (BaFT), ultrasound (US), computed tomography (CT), and magnetic resonance (MR). Five BaFT studies reported variable correlations with intraoperative measurements (r = 0.43–0.93); most (3/5) reported underestimation of length. US studies (n = 2) did not correlate with ground truths. Two CT studies reported moderate-to-strong correlations with pathologic (r = 0.76) and intraoperative measurements (r = 0.99). Five studies of MR showed moderate-to-strong correlations with intraoperative or post-mortem measurements (r = 0.70–0.90). Vascular imaging software was used in two studies, and a segmentation algorithm was used for measurements in one.

Conclusions

Non-invasive measurement of small intestine length is challenging. Three-dimensional imaging modalities reduce the risk of length underestimation common with two-dimensional techniques; however, they also require longer times to perform length measurements. Automated segmentation has been trialed for MRE, but this method does not translate directly to standard diagnostic imaging. While three-dimensional images are most accurate for length measurement, they are limited in their ability to measure intestinal dysmotility, which is an important functional measure in patients with intestinal failure. Future work should validate automated segmentation and measurement software using standard diagnostic imaging protocols.

INTRODUCTION

Short bowel syndrome (SBS) is the most common cause of intestinal failure, generally acquired due to foreshortening of the small intestine secondary to surgical resection. Following resection, intestinal adaptation can lead to bowel dilation—a process which can exceed physiologic size and proceed to a pathologic state in which the segment loses coordinated motility, interfering with enteral feeding tolerance or contributing to bacterial overgrowth (1). In addition, it is known that shorter bowel lengths are associated with the inability to wean from total parenteral nutrition (TPN) and maintain enteral autonomy (2). Thus, the capacity for absorption after bowel resection is thought to be largely dependent upon the length of bowel remaining (3). In addition, the morbidity and mortality of patients with SBS has been directly linked to the remaining length of small intestine (4). Unfortunately, those who have undergone emergency surgery may have no record of remnant bowel length available and it is unreasonable to perform a second surgery simply to measure the length (5).

The ability to measure bowel length prior to resection would allow for estimation of disease severity and the patient’s ability to eventually attain enteral autonomy (2, 6). In addition, accurate estimation of the diameter and length of dilated, nonfunctional segments has the potential to assist with decision making (e.g., instituting therapy for bacterial overgrowth) and timing of future surgical interventions (e.g., potential autologous intestinal lengthening) (5). Finally, the assessment of bowel length in animals would be beneficial for the ability to study potential interventions in populations unable to attain enteral autonomy.

The current gold-standard for bowel measurement in humans is intra-operative evaluation performed by an experienced surgeon (5). However, even measurements performed in the operating room are notably unreliable, as the result can be influenced by many factors including: adequate exposure, stretch applied during measurement, the measurement tool used, and the individual performing the measurement (7). In addition, by the time of this assessment, the decision to operate has already been made, with the patient under anesthesia and partial surgical intervention performed. In animals, ex vivo measurement is the only way to accurately assess intestinal length. This requires sacrifice of the animal, making it impossible to longitudinally follow adaptation over time or to serially assess pre- and post-intervention lengths, making the study of any relevant interventions for enteral autonomy very challenging.

While bowel dilation and dysmotility are of primary interest in surgical planning, preoperative length determination would be useful for conditions in which resection of bowel is being debated, and may be especially important for informed consent regarding prognostic outcomes in those requiring multiple resections of the small intestine (8). One option by which to assess the bowel non-invasively is to do so via radiologic study. Unfortunately, measurement performed via imaging has the potential to be confounded by several anatomic, physiologic, and geometric factors. Muscular contraction, for example, has been shown to affect measured length of bowel in cadaver studies, highlighting the ability of bowel contraction to alter dimensions on static images and result in discrepancy when comparing unanesthetized measurements to anesthetized intra-operative measurements (9, 10). In addition, adhesions between bowel loops are generally taken down prior to intra-operative measurements, which can also increase the difference between radiographic and operative findings (11). Further, there are a variety of imaging modalities to consider, each of which have associated strengths and limitations.

Clear evidence is lacking regarding the ideal imaging modality and length measurement technique for performing non-invasive small intestine length measurements in vivo for both humans and animals. The purpose of this systematic review was to summarize the evidence for the validity and reliability of small intestine length measurements in both humans and animals based on diagnostic imaging techniques. Availability of a standardized, accurate method for non-invasive bowel length measurement based on a readily available diagnostic imaging modality would enable significant research and clinical progress toward understanding and treating short bowel syndrome.

METHODS

The reporting of this study follows PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) (12). The protocol of this systematic review was submitted to the PROSPERO international prospective register of systematic reviews (ID 244242) on December 14^th, 2021.

Search Strategy

The literature search was conducted in August 2021 and updated in December 2021 by a medical librarian in the following databases: Ovid MEDLINE, Ovid MEDLINE Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Embase Classic + Embase, IEEE Xplore, Web of Science, and the Cochrane Library on Wiley. In addition, an artificial intelligence (AI) search tool, Research Rabbit (https://www.researchrabbit.ai/, Human Intelligence Technologies Incorporated ©2021), was utilized to identify additional studies for inclusion. No time filters were used in the search.

Search terms included combinations of the following: small bowel or small intestine, length or organ size, diagnostic imaging, magnetic resonance imaging, or three-dimensional imaging. The complete search strategy is reported in Appendix 1.

Eligibility Criteria

All full-length English-language studies were evaluated. Inclusion required mention of intestinal length as a primary or secondary outcome within the study of an in vivo model. Diagnostic imaging was required as a modality to assess the length. Articles were excluded if they did not describe their method of measurement, if the article did not assess intestinal length, or if results were not validated against a ground truth.

Studies were independently screened by MAC and NAW at the title and abstract level and duplicate publications were removed. After this screening, papers considered suitable for our analysis were examined at the full-text level to select articles eligible for the systematic review. Any discrepancies during the selection process were discussed and eventually resolved between the two authors.

Data Extraction and Synthesis

Articles were reviewed in full. During review, the following variables were recorded: type of study, population demographics, number of subjects, type of radiographic study, method of radiographic measurement, radiographic measurement lengths, ground truth measurement, method of ground truth measurement, ground truth measurement lengths, time between measures, statistics used and available results (variation between measures, correlation, etc.), and limitations. Effect measures, bias assessments, and certainty assessments, as reported by the original authors, were recorded as available. Quality of included studies was appraised using the guidelines set forth by the Cochrane Review Handbook in relation to reviews of diagnostic test accuracy (13); each article was assessed using the QUADAS tool (14).

RESULTS

Included Studies

The initial library search retrieved 86 unique results after duplicates had been removed. Abstracts for each of these articles were reviewed and subdivided into categories: eligible for inclusion (n = 6), excluded (n = 67), and articles of unclear relevance (n = 13). A collection containing all included articles was created in ResearchRabbit and the AI search tool suggested an additional 45 unique articles as “Similar Work.” Abstracts for each of these articles were also reviewed and subdivided into: eligible for inclusion (n = 3), excluded (n = 36), and articles of unclear relevance (n = 6).

The full texts of 27 articles (Eligible for Inclusion = 9, Unclear relevance = 19) were assessed. Of the 19 articles of initially unclear relevance, 1 was ultimately included (15) and 18 were excluded. Reasons for exclusion included: no ground truth for comparison (n = 12;(10, 16–26)), bowel length not assessed (n = 4;(27–30)), and bowel length not assessed via diagnostic imaging (n = 2; (31, 32)).

Ultimately, 11 articles met criteria for this review (Figure 1). Of these studies, imaging modalities included were barium follow through (BaFT, n = 5), ultrasound (US, n = 3), computerized tomography (CT, n = 2), and magnetic resonance (MR, n = 5). Four studies evaluated multiple imaging modalities. Each imaging modality is discussed individually below, with a summary of results available in Table 1 and detailed methodologies in Table 2.

Figure 1. Study Identification.

This flow chart shows the detailed process of article review for inclusion. From an initial review of 131 total library and ResearchRabbit abstracts, 28 articles underwent an evaluation of the full-text. Ultimately, 11 studies met criteria for final review.

Table 1 –

Study overview.

Article	Diagnostic imaging modality	Study type	Population	Sample size	Ground truth	Length measurements (cm)	Correlation (r)	Significance (P-value)
Nightingale 19913	BaFT	Retrospective Total small intestine length	Varying pathologies (e.g., Crohn’s, radiation enteritis) with small bowel anastomosis and residual small intestine length <200 cm	18	Surgical measure	<200	0.72	<0.01
Shatari 20048	BaFT	Retrospective Total small intestine length	Patients with Crohn’s disease Median age: 39 y	22	Surgical measure	-	0.70	<0.01
Lodwick 201611	BaFT	Retrospective Total small intestine length	Patients with SBS undergoing intestinal lengthening Mean age: 3.9 y	13	Surgical measure	Radiographic median: 77 (29–142) Surgical median: 69 (21–170)	0.93	<0.01
Velazco 201733	BaFT	Retrospective Total small intestine length	Children with SBS and severe intestinal failure Mean age: 0.8 y	36	Surgical measure	Radiographic mean: 45 (11138) Surgical mean: 90 (18–270)	0.43	<0.01
Cheng 20185	BaFT, CTE	Prospective, single-center Total small intestine length	Patients with SBS Median age: 43.8 y	29	Surgical measure	BaFT mean: 51 (6–89) CTE mean: 76 (6–168) Surgical mean: 73.1 (5–160)	BaFT: 0.71 CTE: 0.99	BaFT: < 0.01 CTE: < 0.01
Migaleddu 200915	US	Prospective, single-center Total small intestine length	Patients with Crohn’s disease Mean age: 38 y	47	Endoscopic measure	-	0.47	<0.01
Walldorf 201534	US, btMRE	Prospective, Rodents Total colonic length	Pathogen-free male balb/c mice Dextran sulfate sodium colitis (n = 36), healthy colon (n = 15)	51	Ex vivo measure	-	US: - MRE: 0.53	US: > 0.05 MRE: < 0.05
Dillman 201635	US, MRE	Prospective, single-center Length of involved segment	Children with newly diagnosed small bowel Crohn’s disease Mean age: 15.3 y	29	Additional imaging	US mean: 8.8 (1.5–16.1) MRE mean: 14.6 (3.9–25.3)	-	-
Broquet 201736	CTE or MRE	Prospective, single-centert	Patients with ileocolonic Crohn’s disease Mean age: 36 y	CTE: 19 MRE: 35	Ex vivo pathologic specimen	Radiographic means*: 20.5 (2–73 cm) and 20 (3–90) Pathologic mean: 16.5 (2–75)	0.76	<0.01
Sinha 201437	MRE	Prospective, single-center Total small intestine length	Patients with Crohn’s disease	32	Surgical measure	Radiographic mean: 411 (156–797) Pathologic mean: 414 (156–797)	0.98	<0.05
Wilson 201738	MRE	Prospective, single-center and Rodents Total small intestine length	Human: Patients who underwent routine MRE for clinical purposes Rodents: 10 C57BL/6 mice, age 10–12 wk	Human: 15 Rodent: 10	Human: None Rodent:Ex vivo measure	Human radiographic mean: 436 (203–912) Rodent radiographic mean: 36 (29–43) Rodent ex-vivo: 38 (34–43)	0.63	0.02

BaFT = Barium follow-through; CTE = Computed Tomographic Enterography; US = Ultrasound; btMRE = benchtop magnetic resonance enterography.

^*Measurements performed by two different radiologists.

Table 2 –

Study methods.

Article	Bowel prep	Imaging protocol	Radiographic measurement method	Ground truth (detailed)	Time between measures	Risk of bias (QUADAS)
Nightingale 19913	-	Over-couch films, prone* Tube-film distance: 100 cm	Opisometer, single radiologist; Performed 3 times, nearest 5 cm; mean recorded	Surgical measure	Median: 18 (2–118) mo	4
Shatari 20048	-	Over-couch films, prone Tube-film distance: 100 cm	Opisometer, gastrointestinal surgeon; Performed 3 times, nearest 5 cm; mean recorded	Surgical measure (50- cm suture + ruler length)		5
Lodwick 201611	-	Radiopaque ruler in field Short fluoroscopic acquisition to delineate overlapping bowel loops	Radiopaque ruler in field, single radiologist	Surgical measure	-	7
Velazco 201733	-	-	Opisometer, single radiologist + surgical trainee; mean recorded	Surgical measure	Mean: 10 (7–20) d	3
Cheng 20185	BaFT: 8–12 h fast, 1.5 L barium suspension 2 h prior, cisapride + aerogenic powder 30 min prior CTE: 8–12 h fast, 1L 5% gastrograffin 15 min prior, 10 mg IM anisodamine (if length > 150 cm)	BaFT: Images q10–15 min CTE: Collimation: 0.6 mm, Table speed: 14 mm/rotation, Pitch 1.4, 120 kVp, 230 mAs	BaFT:-CTE: Vascular imaging software; Segments tagged and virtually reconstructed	Surgical measure (20-cm suture + ruler length, nearest cm)	Mean: 36.8 (6–67) mo	7
Migaleddu 200915	-	CE-US: Multiband high-frequency transducer after injection of sulfur hexafluoride-filled microbubbles	3 expert radiologists	Endoscopic measure	-	3
Walldorf 201534	US: None MRE: 10 μL iron oxide-ink solution injected endoscopically one day prior; rectal contrast prior to exam	US: mice anesthetized MRE: T1-weighted signal, Spin echo time: 9.8 ms, field-of-view: 25 × 25, slice thickness: 3 mm; mice anesthetized	-	Ex vivo measure	0 d	2
Dillman 201635	-	US: None MRE: Axial + coronal single-shot fast spin-echo +/− fat saturation, post-contrast T1-weighted 3D spoiled GRE fat-saturation images	-	Compared to additional imaging during same episode of care	0 d	4
Broquet 201636	CTE: 1L water via NJ tube, 1.5–2L 3% mannitol 45–60 min prior + anti-peristaltic drug† MRE: 1 L 3% mannitol over 50–55 minutes along with anti-peristaltic drug†	CTE: Triple-phase helical acquisition prior to 100 mL iodine contrast (injection rate: 2.5–4 mL/s), followed by portal (60–90 s) and tardive (3 min) phases. MRE: True FISP, half-Fourier acquired single-shot fast spin-echo, T2-weighted turbo spin-echo, and gadolinium-enhanced fat-suppressed spoiled gradient-echo sequences	Two-dimensional multiplanar reconstruction of diseased bowel	Ex vivo pathologic specimen (operative resection, pinned to cork board without traction, 10-cm string length)	<2 mo	4
Sinha 201437	1.2–1.3 L 3% mannitol over 50–55 min, 10 mg metoclopramide in two aliquots 25 min apart, 200 mL oral contrast just prior to exam	1.5 T scanner, gradient field strength: 30–45 mT/m, slew rate: 125–200 T/m/s, phased-array coils; FISP images	Vascular imaging software, segments tagged and automated centerline path	Surgical measure (ruler, nearest cm)	<6 wk	2
Wilson 201738	Human: 1–3 bottles barium sulfate suspension (460 mL) 45 min after 4 h fasting, single dose IM glucagon injection immediately prior to scan Rodent: 100 mL water, 10 g polyethylene glycol, and 200 μL gadolinium-based contrast agent offered ad libitum 24 h prior	Human: 1.5 T system, phased-array coils, axial + coronal images, true FISP. IV contrast given in single dose of gadoversetamide. Rodent: 4.7 T system, 2.5 cm quadrature birdcage coil, standard gradient echo multi-slice sequences (Tr/Te = 225/3.75 ms; NS = ~30; slice thickness 1 mm; matrix: 128 × 128)	Human: manually segmented Rodent: Images processed and segmented using custom-designed algorithm	Human: none Rodent: Ex vivo measure (free hanging for 2 min, laid flat, ruler)	0 d	2

BaFT = barium follow-through; CTE = computed tomographic enterography; US = ultrasound; btMRE = bench top magnetic resonance enterography; CE-US = contrast-enhanced ultrasound; 3D = three-dimensional; FISP = fast Imaging with steady state precession; GRE = gradient recalled echo; IM = intramuscular.

^*Except in patients with stomas.

^†Anti-peristaltic drug is not named.

Four studies were retrospective (36%), while seven were reported as prospective (63%). Risk of bias was analyzed in eleven domains (Figure 2). The most common risks for bias were in the domains of patient population (representative spectrum) and lack of blinding for index and reference test results. No studies were excluded due to risk of bias estimates.

Figure 2. Risk of Bias.

Risk of bias was analyzed in with the Quality Assessment for Diagnostics Studies (QUADAS) tool in eleven domains, as recommended by the Cochrane Review Handbook for reviews of diagnostic test accuracy. Each domain is represented by score-associated color: high risk (red), moderate risk (yellow), low risk (green); and by and symbol: clearly indicated (+), unclear in text (?), or not performed (−).

Barium Follow Through (BaFT)

There were five individual studies, including a total of 118 patients, that assessed small intestine length using Barium Follow Through (BaFT) (3, 5, 8, 11, 33). Four were retrospective, while one was prospective. Two of them assessed patients with Crohn’s disease, while three included only patients with SBS.

Three of the five studies used a curved measurement tool, or opisometer, to measure radiographic small intestine length; a fourth used a radiopaque ruler during the time of image acquisition. All five studies compared radiographic small intestine length to intraoperative measurements and reported correlations ranging from weak to strong (Pearson’s r range: 0.43 – 0.93). Three of the five studies explicitly noted that radiographic lengths were underestimated compared to intraoperative measurements (5, 8, 33).

Underestimation of radiographic length was exacerbated by longer lengths of small intestine (5, 8). Specifically, Shatari et al., report that radiographic underestimate was < 20 cm with 90% confidence when bowel lengths were shorter than 250 cm (8). Furthermore, Lodwick et al., studied a group of 13 subjects planned to undergo a lengthening procedure for short bowel syndrome (intraoperative small intestine length median: 69 cm, range: 21–170 cm) and found a strong correlation between the intraoperative and radiographic measurements of small intestine length (median 71 cm, range: 29–142 cm, r = 0.93) (11). In contrast, Cheng et al., studied 29 patients with SBS (intraoperative small intestine length mean ± standard deviation [SD]: 73.1 ± 37.2 cm). They found only a moderate correlation between the intraoperative and radiographic BaFT measurements (mean ± SD: 27.14 ± 8.41 cm, r = 0.81) (5). Similarly, Velazco et al., noted that radiographic assessment underestimated intestinal length in 82% of patients in their retrospective series which included patients with longer small intestine lengths (intraoperative small intestine length median: 90 cm, range: 18–270 cm). They found a weak correlation between the intraoperative and radiographic small intestine length measurements (median 45 cm, range: 11–138 cm, r = 0.43) (33).

Bias / Reproducibility

All five studies evaluating BaFT were found to have populations not considered generalizable toward the average population. Three included subjects with SBS, which would bias measurements toward increased accuracy given shorter lengths. These three studies had the highest correlations. Each study was also found to have higher risk of bias in relation to blinding of measurements. Lodwick et al. reported a lack of blinding in their study as the radiologist performing measurements had access to operative information (11), while the other BaFT studies did not explicitly report blinding (3, 5, 8, 33). Finally, two BaFT studies had measurements performed by radiologists with extensive, non-generalizable experience.

Ultrasound (US)

There were three individual studies, including a total of 127patients, which assessed bowel length using ultrasound (US) (15, 34, 35). All three studies were prospective; one assessed total small intestinal length (15), one assessed large bowel length in rodents (34), and one assessed the length of the involved segment in newly diagnosed Crohn’s disease (35). None of the US studies provided their methodology for performing radiographic or ground truth measurements. Statistical assessment and length reporting varied greatly.

Migaleddu et al. compared US measurements to endoscopic measurements in 47 patients with Crohn’s disease. Radiographic measurements were performed by three radiologists and averaged. Direct length measurements were not reported from either US nor endoscopy, but the authors reported a weak correlation (r = 0.47, p = 0.0009) (15). Walldorf et al. assessed 51 rodents who underwent both US and MRE prior to sacrifice for ex vivo measurements. Length measurements were not provided from radiographic analysis, but were reported to correlate poorly with post mortem lengths (range: 38 – 103.2 cm) (34). Dillman et al. compared US length measurements (mean: 8.8 cm, range: 1.6 – 16.1 cm) to MRE length measurements (mean: 14.6 cm, range: 3.9 – 25.3 cm) in 29 patients; pairwise univariate comparisons between individual radiologists varied widely for length of involvement (p-value range: 0.0009 – 0.60), and agreement was considered slight to moderate (ICC: 0.37 – 0.42). The authors concluded that US was a weak method of assessment for length of bowel (35).

Bias and Reproducibility

Of the three US studies, one evaluated total small intestinal length. Walldorf et al. assessed rodent anatomy, and in particular, measured colonic length, which maintains more consistent anatomy and likely biases toward higher accuracy. Dillman et al. assessed involved segment length in those with Crohn’s disease, which likely biases toward greater accuracy given shorter lengths. Despite these biases, neither study reported strong correlations with their ground truth assessments. None of the studies provided detailed explanations of methodology, nor did they provide raw data of bowel length for comparison to the average population.

Computerized Tomography (CT)

There were two individual studies, including a total of 83 patients, which assessed bowel length using Computerized Tomography (CT) (5, 36). Both studies were prospective, single-center studies. One study evaluated total small intestinal length (5) while the other assessed length of an involved segment, both in patients with Crohn’s disease. Imaging parameters and protocols varied (Table 2).

Cheng et al. used a vascular imaging software on CT enterography (CTE) images of 39 patients to create a virtual reconstruction of their small intestines for radiographic measurement (mean: 76 cm, range: 6 – 168 cm), then compared these results to surgical measurements performed with 20-cm lengths of suture and a ruler (mean 73.1 cm (5 – 160 cm). There was a strong correlation between the two measurements (r = 0.99, p < 0.0001) (5). Broquet et al. created a two-dimensional multiplanar reconstruction of the diseased bowel from CTE or magnetic resonance enterography (MRE) in 54 patients with Crohn’s disease to measure the length of the involved segment (mean: 20 cm, range: 2 – 90 cm). Radiographic measurements were compared to pathology from the operative resection. Pathologic specimens were measured while pinned to a cork board without traction and using a string of 10 cm length (mean: 16.5 cm, range: 2 – 75 cm), with a moderate correlation between radiographic and pathologic measurements (r = 0.76, p < 0.001). Univariate analysis revealed that length > 20 cm was a risk factor for radiographic measurement error > 5 cm (noted in 18/28 patients) (36).

Bias and Reproducibility

Both of the CT studies evaluated populations not fully generalizable to the average, healthy adult, as Cheng assessed those with short bowel syndrome and Broquet assessed only the involved segment of those with Crohn’s disease. As previously mentioned, shorter lengths are likely to increase accuracy with radiographic measurement. Additionally, Cheng et al. was noted to have an average of 36.8 months between imaging and ground truth measurements, which could impact the correlation between results, though this would likely reduce accuracy.

Magnetic Resonance Imaging (MRI)

There were five individual studies, including a total of 190 patients, that assessed bowel length using Magnetic Resonance Imaging (MRI) or Enterography (MRE) (34–38). Each of these were prospective studies, with three consisting of patients from a single center. However, one study measured colonic length in rodents, two studied length of an involved segment, and two assessed total small intestinal length. Methodology and imaging protocols may be reviewed in more detail in Table 2.

Walldorf et al., as discussed above in Ultrasound, assessed rodent colonic length in 51 mice using benchtop MRI which was subsequently compared to ex vivo measurement and found a moderate correlation (r = 0.53) (34). Dillman et al. and Broquet et al. both assessed patients with Crohn’s disease and measured the length of the involved segment. Dillman compared MRE measurements (mean: 14.6 cm, range: 3.9 – 25.3 cm) to US measurements (mean: 8.8 cm, range: 1.5 – 16.1 cm) performed during the same episode of care (correlation not significant) (35), while Broquet’s population subsequently underwent resection of that segment and compared the imaging measure (mean: 20 cm, range: 2 – 75 cm) to that of the pathologic specimen (mean: 16.5 cm, range: 2 – 75 cm). Broquet’s radiographic and pathologic measurements were strongly correlated (r = 0.76, p < 0.01) (36).

Sinha et al. and Wilson et al. measured the entirety of small intestinal length. Sinha et al. assessed 32 patients undergoing elective laparotomy for Crohn’s disease and used intraoperative measurement as their ground truth. Vascular imaging software was applied with segments tagged and an automated centerline path used for radiographic measurements (mean: 411, range: 156–797 cm), while surgical measure was performed using lengths of a ruler (mean 414 cm, range 156 – 797 cm). Their results correlated very well (r = 0.98, p < 0.05), but the authors noted that longer lengths were associated with significantly longer time requirements for radiographic measurements, making the process impractical (37).

Wilson et al. measured full small intestinal length in 15 humans and 10 rodents, also using software. Wilson’s group designed a custom computer algorithm to perform the measurements, with the aim of significantly reducing the time required for radiographic measurements. The software was designed using human data (mean 436 cm, range 203 – 912 cm) and subsequently applied to rodents (mean: 36 cm, range: 29 – 43 cm). Rodent data was compared to post-mortem, ex vivo measurements as a ground truth (mean: 38 cm, range: 34 – 43 cm) with moderate correlation (r = 0.63, p = 0.02) (38).

Bias and Reproducibility

Sinha et al. was the only study to truly compare full small bowel intestinal length to a ground truth in this group; however, their population was specific to Crohn’s disease and did not include the average person with healthy bowel. Other studies were limited by measuring only segment length, measuring colon alone, or having ground truth only for rodent data. Dillman et al. was additionally lacking a standardized measure of ground truth, given that ultrasound results were found to be unreliable.

Generalizability is a concern in several of these studies given their specific populations. Wilson et al., in particular, utilized an imaging protocol specific to mice that included a significant amount of contrast and bowel distension that is unlikely to be tolerated in humans. Additionally, 1 mm slice thickness was used for MRI, which is not standard for clinical diagnostic imaging; thus, it is unclear whether this algorithm would easily generalize to standard clinical images.

DISCUSSION

This study represents the first systematic review of the literature for assessing small intestine length via diagnostic imaging in humans and animals. While many studies have used radiographic images to assess small intestine dysmotility and segment diameters, achieving a time-efficient, accurate, and non-invasive measurement of small intestine length remains a significant challenge. While accurate evaluation of dysmotility and segmental diameter(s) may be more feasible with current diagnostic imaging techniques, those measurements are outside the scope of this review. Several imaging modalities have been assessed, but a vast majority of reports lack validation against a ground truth and therefore cannot assess measurement accuracy. Furthermore, the only universally accepted ground truth, operative measurement, is invasive, requires either surgical intervention or animal sacrifice, and remains a somewhat unreliable measurement due to high variability.

Of the works that met inclusion criteria for this review, measurement methodology, measurement results, and statistical assessment varied greatly, with several studies not reporting raw length data nor providing any assessment of agreement or bias. While most identified as prospective studies, the majority included components of imaging which had been collected days-to-months prior and were compared with ground truths that were collected prospectively. Finally, the majority of studies were performed in humans (n = 9), while one was performed in both humans and mice and one was performed only in rodents.

Barium Follow Through

BaFT is a well-described technique for assessment of small bowel characteristics and functionality at a basic level (8). Measurements of bowel lengths have been reported with use of an opisometer, which is a small, hand-held mechanical device typically used for measuring non-linear distances on maps. The typical use involves running a small wheel over the long axis of the bowel and the distance traveled is indicated on a dial with a pointer (8). However, given the two-dimensional nature of fluoroscopic images, there are several factors that lead to inaccuracy of measurements including: peristalsis, overlapping bowel segments, and a slight magnification effect which occurs due to separation of bowel and film from the X-ray source (8). Shatari et al. noted that lengths > 250 cm led to greater amounts of intestinal overlap and greater technical difficulty in measurement, exacerbating this inaccuracy. However, the authors also remark that an exact length measurement is less important when patients have longer intestinal lengths (8).

One option that may reduce the risk of inaccuracies is to have all imaging and measurements performed by a dedicated radiologist, using a radiopaque ruler in the field to minimize magnification errors, then acquiring intermittent short fluoroscopic acquisitions to delineate overlapping loops of bowel, as performed by Lodwick et al. (11). However, this method was noted to be time consuming and required the manual labor of an individual with extensive and very specific training unlikely to be available elsewhere, reducing the generalizability of these results.

In general, BaFT was reported to underestimate bowel length in three of five studies (5, 8, 33). Of the remaining two BaFT studies, individual length measures were either not specified (3) or were only applicable to short bowel lengths (11). With the current data as described and resources currently available at most medical institutions, BaFT is unlikely to be a reliable method for measurement of small intestinal length in any population.

Ultrasound

Given cost, expense, and ionizing radiation required for other methods of imaging, small bowel ultrasound has been largely explored in countries outside of the United States (18). Without need for any sort of preparation, ultrasound is able to provide high resolution, two-dimensional images of bowel layers, vascularization, and associated pathology (18). However, given the limited field of view that ultrasound provides, studies have only assessed shorter segments, such as stricture length or involved inflammatory length in Crohn’s disease, as opposed to the full length of the bowel (15, 18, 24).

Even in assessment of shorter segments of bowel, comparisons to endoscopic measurements (15), post-mortem measurements (34), or to US performed during the same time frame (35) did not show clinically acceptable agreement. None of the three included US studies recommended the use of ultrasound to measure bowel length.

Computerized Tomography

Computerized Tomography images are well-known to provide quick, high-quality diagnostic images in both humans and in veterinary capacities (17). CT is broadly used for preoperative planning when reconstruction of bony or soft tissue structures are required. Image segmentation and 3D surface reconstruction have been well-studied and commercial artificial intelligence and segmentation programs are available on the market. When time permits, enterography is performed with CT imaging, using oral contrast material in the small bowel to provide better visualization of the bowel and associated pathology. CT can have better special resolution than MRI (36) and can be performed much more quickly. In addition, as a three-dimensional modality, it lacks the two-dimensional inaccuracy of overlapping segments which leads to underestimation of length in methods such as BaFT.

In the two included CT studies, radiographic length measurements were found to correlate well with intraoperative and pathologic measurements, particularly when overall bowel lengths were shorter. When additional imaging software was applied, correlation was as high as 0.99 in association with good agreement, thus proving it an accurate measure. However, CT requires significant exposure to ionizing radiation. Radiation exposure has been linked to increased cancer risk; in particular, a study of nearly 200,000 patients found nearly triple the risk of leukemia and brain cancer in children who underwent CT scans prior to ten years of age(39). Thus, CT is not ideal for young patients who may require serial imaging due to their underlying pathology.

Magnetic Resonance

Magnetic resonance imaging (MRI) and enterography (MRE) are considered ideal techniques for small bowel assessment given safety, reproducibility, and ability to perform repeated assessment without radiation exposure (40, 41). Diffusion-weighted imaging provides quantitative functional evaluation without intravenous contrast, which makes it suitable even in patients with impaired renal function, and MRE or MR enteroclysis are frequently used for assessment of inflammatory bowel disease. As in CT, three-dimensional, planar methodology reduces the risk of length underestimates associated with overlapping segments (37). Given the young ages of many patients with SBS, the avoidance of ionizing radiation makes MRI preferable to CT (40). Unfortunately, MR images are expensive and more time-consuming to obtain, requiring stillness of the subject and possible sedation in younger patients (18).

Of the five MRI studies included in this review, four were compared to pathologic or intra-operative measures which are well-accepted as ground truth measurements of bowel length. Of these, three assessed small-intestinal length with correlations ≥ r = 0.70 (36–38). One study that assessed colonic lengths reported only a moderate correlation (34), while another evaluated agreement between MRE and US, which was expectedly poor (35).

Two studies used computer software to ease the manual labor of measurements from MR images (37, 38). However, Sinha et al. reported that due to the sheer number of images compared to two-dimensional files, the lack of dedicated segmentation software made the process far more time-consuming (37). Dedicated semi-automatic segmentation software was subsequently trialed by Wilson et al. in 2017 (38). However, the sample size in this study was limited to 10 mice and a 4.7 T system with 1-mm slice thickness was used. Typically, diagnostic MRIs are performed on 1.5–3 T systems with slice thickness of 3–6 mm, which may limit the generalizability of the automated segmentation and measurement software.

Conclusions

While ex vivo or surgical measurements are considered the gold standard of intestinal length measurement, the inability to perform these measurements without operative intervention or sacrifice of study animals makes prognostic assessment of nutritional capabilities and the long-term study of intestinal length in response to any intervention extremely challenging. For this reason, there is significant need for a method to non-invasively measure small intestine length. This review highlights the opportunities and challenges associated with radiographic assessment of small intestine length.

While two-dimensional imaging modalities largely underestimate intestinal length compared to intra-operative measurements, three-dimensional imaging modalities have their own benefits and limitations. CT imaging is quick to perform, but requires ionizing radiation, while MR imaging is more time consuming, but avoids radiation exposure. Both three-dimensional, planar methods reduce the risk of bowel overlap and the subsequent likelihood of length underestimation. However, given the increased number of images, both modalities require far more time to perform measurements. While automated segmentation has been trialed for MRE and validated in a mouse model, this method does not likely translate directly to standard diagnostic imaging in human subjects. In addition, while three-dimensional imaging techniques offer the strength of accurate length measurements, their usefulness for the evaluation of patients with intestinal failure is limited in other capacities, including the important functional measure of dysmotility.

In all, three-dimensional imaging, including CT or MR, appear to be the most reliable methods for bowel length measurement in all populations, with the caveat that MR is preferable in young patients for avoidance of radiation. Future work to validate automated segmentation and measurement software in humans using standard diagnostic imaging is required.