Cranial normality indices in Ugandan infants based on smartphone photogrammetry: key parameters for the study of cranial deformation

Abstract

Background

Early identification of cranial deformities in infants is essential for timely intervention and preventing long-term complications. In Uganda, particularly in rural areas, access to specialised paediatric healthcare remains limited, making it difficult to diagnose and manage cranial abnormalities early. This study takes a unique approach by using the PhotoMeDAS system. This non-invasive technology allows accurate cranial measurements through QR-based imaging to define normative cranial index ranges for Ugandan infants. By analysing key cranial parameters associated with head shape, symmetry, and overall growth patterns, this study aims to establish reference values that can serve as a baseline for clinical assessments.

Methods

A cross-sectional study was conducted in healthcare centres in Lugazi, Uganda, collecting cranial measurements from 195 infants aged 0–2 years old. The data was collected using the PhotoMeDAS system, a non-invasive technology that allows accurate cranial measurements through QR-based imaging. The primary cranial variables analysed included Head Circumference (HC), Cephalic Index (CI), Cranial Vault Asymmetry Index (CVAI), Pseudovolume (PV), and Frontal Angle (FA). HC was measured to establish general growth patterns, while the CI provided insights into head shape classification (dolichocephalic, mesocephalic, or brachycephalic). The CVAI assessed cranial symmetry, detecting potential cases of plagiocephaly, while the PV evaluated the proportionality of cranial growth relative to the perimeter. Finally, the FA was considered to assess anterior cranial development and forehead morphology.

For the management and optimisation of processes, Microsoft Visio was used to create flow diagrams that facilitated the organisation of the paediatric campaign. SPSS and Excel were employed for statistical analysis, evaluating the relationship between cranial indices and variables such as sex, age, and ethnicity. Normality tests (Kolmogorov-Smirnov), chi-square tests, and ANOVA were applied to identify distribution patterns and significant variations among subgroups. In addition, Python was used to process cranial images obtained during the paediatric research.

Results

This study provides the first normative dataset for cranial indexes in Ugandan infants, highlighting significant variations in head morphology according to age and sex. The CI revealed a predominantly mesocephalic distribution, with a small proportion of dolichocephalic and brachycephalic cases. The CVAI showed a low prevalence of significant asymmetry, suggesting minimal cases of plagiocephaly in the sample. The PV indicated proportional cranial growth trends, with expected increases in head size over time. The FA exhibited slight variations across age groups, reflecting dynamic changes in forehead development during early childhood. These findings have the potential to impact healthcare practices significantly, guiding early diagnosis and intervention strategies for cranial deformities in Ugandan infants.

Conclusion

This study fills a crucial gap in paediatric cranial morphology research by establishing the first cranial index reference values for Ugandan infants. These findings provide essential data for healthcare professionals to facilitate early diagnosis, guide preventive strategies, and improve clinical assessments in low-resource settings where specialised care is scarce. Further research is needed to investigate genetic and environmental influences on cranial development and compare these normative values with populations from different geographical and ethnic backgrounds. This emphasis on the need for further research suggests the potential for future developments in the field.

Introduction

Cranial deformations in newborns represent an emerging issue in global paediatric health, particularly in countries with limited access to specialised services [1, 2]. These conditions can be congenital or acquired during the first months of life due to prolonged positioning in the same posture or inadequate paediatric follow-up. The most common children with definitive non-synostotic head deformations include positional plagiocephaly (abnormal CVAI), positional brachycephaly/dolichocephaly (abnormal CI), and combined positional plagiocephaly and brachycephaly [3], all of which may have severe developmental repercussions if not detected and treated on time [4].

In 1992, the American Academy of Pediatrics (AAP) recommended placing infants in the supine position for sleep to reduce the incidence of sudden infant death syndrome (SIDS). While this measure has significantly lowered SIDS rates, it has also contributed to an increase in the prevalence of positional cranial deformations due to constant pressure on the developing posterior skull [5, 6]. The increase in the prevalence of positional skull deformities worldwide has been widely documented following the AAP’s “Back to Sleep” campaign [7], which promoted supine sleeping to reduce the risk of SIDS. Unlike in Western countries, where this campaign is a key etiological factor, in Uganda and other African nations, cultural practices of child rearing and carrying are different [8]. Therefore, the high prevalence associated with supine positioning is less relevant, while the lack of early detection strategies due to a shortage of healthcare professionals remains the main local challenge.

In Uganda, where the birth rate is approximately 36 births per 1,000 inhabitants and 76% of the population lives in rural areas, early identification of cranial deformations is a challenge. According to recent data, 26% of children under five in the country experience growth delays, which may be associated with a higher risk of cranial morphology alterations [9–12]. The lack of training in infant care and the shortage of skilled professionals contribute to these conditions going unnoticed, increasing the risk of long-term functional and aesthetic complications [13].

To address this issue, it is essential to implement prevention and early detection strategies at community level. Educating parents and caregivers on the importance of varying the infant’s posture, encouraging tummy time during wakefulness, and recognising early signs of cranial deformation can significantly reduce the incidence of these conditions [14]. Additionally, introducing measurement and monitoring tools, such as coded caps for evaluating cranial indices, can enhance detection and treatment capabilities in resource-limited settings [15, 16].

This study aims to establish reference values for standard cranial indices in the paediatric population of Lugazi (Buikwe District of the Central Region of Uganda). Through the collection and analysis of morphological data, the goal is to generate a knowledge base that facilitates the implementation of child health programs tailored to Uganda’s healthcare needs, contributing to the prevention and early detection of cranial deformations in infants.

Background

Evaluation of infant cranial deformation: previous methods and their evolution

The analysis of cranial morphology in the paediatric population has evolved significantly, allowing the assessment of essential parameters such as HC, CI, CVAI, PV, and FA. Over time, different methodological approaches have been developed, including manual anthropometry, medical imaging techniques, and three-dimensional (3D) scanning systems. Each method has provided advantages and challenges in studying infant cranial deformation, influencing the accuracy and applicability of tools to help diagnostics in various clinical and epidemiological contexts.

Manual anthropometric methods have been widely used due to their accessibility and ease of application. These measurements are performed using tools such as measuring tapes and sliding callipers (craniometers), allowing data collection on skull symmetry and proportions. HC is a key indicator of infant cranial development [17, 18]. CI evaluates the relationship between skull width and length, providing information on overall cranial morphology [19–21]. CVAI is used to quantify cranial asymmetry, particularly in cases of plagiocephaly [18, 20, 21]. Despite their usefulness, these methods may present limitations regarding accuracy and inter-observer reproducibility [22].

The development of new 3D digital techniques has enabled the calculation of additional values such as PV [23], which provides a relative volumetric analysis of cranial deformation, allowing for a more detailed assessment of skull flattening severity [24]. Additionally, FA measures the degree of inclination of the frontal skull region, helping to evaluate its influence on overall morphology [20, 25]. These indices have improved the precision of anthropometric evaluation, providing more detailed criteria for classifying cranial anomalies.

With advancements in medical technology, computed tomography (CT) and magnetic resonance imaging (MRI) have allowed for a more detailed analysis of cranial morphology, providing high-resolution images that facilitate the identification of structural anomalies. However, these techniques present significant drawbacks, such as high costs, the need for specialised equipment, and, in the case of CT, radiation exposure, limiting their applicability in large-scale epidemiological studies [26].

However, the development of 3D scanning systems has revolutionised the evaluation of infant cranial deformation, enabling fast, precise, and non-contact measurements. Technologies such as photogrammetry and laser scanners have proven highly effective for assessing cranial asymmetry without the risks associated with radiation or the high costs of conventional medical imaging techniques [27]. These approaches have facilitated standardised data collection in population studies, opening new possibilities for epidemiological analysis and planning early intervention strategies in resource-limited settings [24].

This study will use an innovative anthropometric method of mobile phone-based 3D scanning technology to evaluate infant cranial deformation in Lugazi, Uganda. The analysis will focus on generating reference data for selected indices, aiming to contribute to the development of prevention and early detection strategies in paediatric health.

The PhotoMeDAS (Photogrammetric Medical Deformation Assessment Solutions) system, a tool for head 3D scanning using smartphones, was used to acquire measurements (Quispe-Enriquez et al., 2023). This solution has already been patented in Spain, United States, and Canada by the Universitat Politècnica de València and Hospital Universitari i Politècnic La Fe.

The PhotoMeDAS system has undergone a rigorous verification process, and its accuracy has been compared to that of highly accurate reference methods and other 3D scanning systems, confirming that it offers accurate metric performance for helping to diagnose cranial deformities. In a low-resource setting, where the need for cost-effective, portable, and rapid diagnostic tools is crucial, PhotoMeDAS presents itself as an innovative alternative that enables objective and simple 3D cranial measurements using only a smartphone, without radiation exposure, taking a few minutes to complete the 3D scanning and processing. In [28] the repeatability of the system was established by scanning a mannequin 54 times under various conditions, verifying the results against a Coordinate Measuring Machine (CMM), the gold standard in micrometric metrology. The findings of this study established that the metric uncertainty of PhotoMeDAS is around 1 mm, and it was estimated that this would only affect the final values of the derived anthropometric indices by around 1%, demonstrating high reliability. Compared to other standard scanners in the market [23], analysed the PhotoMeDAS data and compared them to a reference handheld 3D scanner (ACADEMIA 50), using both mannequins and real patients. After scanning five 3D models, average differences in head morphology of 0.39 ± 1.02 mm were obtained, and differences in the measurement of key distances between markers were around 1.25 mm. These results confirm that PhotoMeDAS is reliable, despite using a purely mobile platform (a handheld smartphone). Additionally [29], confirmed that the metric quality of the final result is independent of lighting conditions, which reinforces its applicability in various clinical settings, although good lighting is recommended for acquisition.

Additionally, the feasibility of implementing accessible 3D scanning technologies in resource-limited communities will be explored, promoting improved early diagnosis and intervention for cranial deformities in international development cooperation contexts.

Methods

The study adhered to the STROBE guidelines for comprehensive reporting of observational research.

Study design and setting

This observational study was conducted in Lugazi, Uganda, focusing on paediatric cranial morphology to establish infant normality indexes. The research was carried out in collaboration with local medical centres and the Kitega Community Centre. Data was collected in a public hospital, Kawolo Hospital, and Living Water medical clinic.

The study followed a structured methodology consisting of five key phases: feasibility study, data collection and preparation, implementation of 3D scanning techniques, data analysis and evaluation, and development of reports and outreach. These phases ensured a rigorous and systematic approach to obtaining reliable infant cranial morphology data.

Data and data sources

Data were collected from paediatric patients attending medical centres in Lugazi. The study involved cranial measurements using 3D photogrammetry. The primary variables assessed included HC, CI, CVAI, PV, and FA. All these data were obtained using the PhotoMeDAS application. Secondary variables such as sex, ethnicity, and age groups were also recorded for analysis.

To ensure data reliability, all measurements followed standardised protocols. The collected data were anonymised and securely stored for further statistical analysis.

Variables

The study focused on five primary cranial morphology variables to assess the normal cranial development of infants:

• HC (cm): This variable identifies the largest circumference of the infant’s head. This automated process ensures consistency across all samples. Head circumference is a key indicator of overall cranial growth and is widely used in paediatric assessments to identify conditions such as microcephaly or macrocephaly.

• CI: This index provides insight into cranial shape variations, distinguishing between dolichocephalic (long and narrow), mesocephalic (average proportion), and brachycephalic (short and broad) head shapes. This index is crucial in anthropometric studies and cranial deformity assessments.

• CVAI: This variable quantified the degree of asymmetry in the cranial vault and was computed as a higher CVAI value indicating more pronounced asymmetry, which is relevant for the early detection of plagiocephaly and other cranial deformities. CVAI is a key marker in evaluating the impact of external pressures on skull development.

Where a is front-right (FR) to posterior-left and b is front-left (FL) to posterior-right (Fig. 1).

Fig. 1.

Views and measurements used to automatically compute the HC, PV, AI, FAI, PAI, CI, CVAI, OCLR and GI after the 3D scanning

• PV: This variable estimates cranial volume based on external cranial dimensions, offering a practical alternative to volumetric radiometric and 3D scanning imaging techniques. It considers the volume above the HC, making it helpful in assessing overall cranial growth trends and detecting potential abnormalities.

• FA: These variable measures the inclination of the frontal bone relative to predefined cranial reference points. It is advantageous for evaluating conditions such as trigonocephaly, where a reduced frontal angle may indicate premature metopic suture fusion. This measurement variable helps assess the proportionality and structural development of the anterior cranial vault.

Figures 1 and 2 illustrate the anthropometric landmarks used for calculating cranial indices. The PhotoMeDAS solution evaluates a comprehensive set of cranial morphology variables, including: HC; PV; Distance between preauricular points - Eurion (d); Maximum width on the cranial perimeter (d’); Arc (A); Arc (B); Arc (F); Maximum anteroposterior distance (c); Anteroposterior distance (c’); Oblique diagonals (a and b) and their frontal and posterior components (a_f, a_p, b_f, b_p); Distance between a_f and b_f (g). In addition, CI; CVAI; Asymmetry Index (AI); Posterior Asymmetry Index (PAI); Frontal Asymmetry Index (FAI); Perimeter30; Oblique Cranial Length Coefficient (OCLR); Towering Index (TI); Metopic Index (MI); FA; Trigonoceplaly Index (TrI); and the Global Index (GI) are reported and monitored on the full anthropometric report. The formulation of each index can be found in [24, 30].

Fig. 2.

Views and measurements used to automatically compute the CVAI, MI, TI, OCLR, TrI, and FA after the 3D scanning

These variables collectively provide a comprehensive assessment of cranial morphology, facilitating the identification of standard developmental patterns and potential abnormalities.

Additionally, demographic factors such as sex, ethnicity, and categorised age groups were recorded to facilitate stratified analysis. Sex was coded as binary (0 for females, 1 for males), and age was categorised into predefined intervals to assess developmental trends across infancy.

Data processing and analysis

Data preparation

All collected data were digitised and structured within a secure database to facilitate analysis. Quality control measures included duplicate data verification, outlier detection using interquartile range (IQR) methods, and standardisation of measurement protocols to ensure consistency. A comprehensive descriptive and statistical analysis of the selected variables was conducted according to sex and five age groups. The chosen age ranges were 0–2, 3–6, 7–12, 13–20, and > 20 months, considering that variable growth follows an exponential pattern in the early months of life, necessitating smaller age intervals. The overall age distribution can be seen in (Fig. 3).

Fig. 3.

Scatter plot of the variable HC (cranial_perimeter) according to age (months)

Descriptive statistical analysis

Descriptive statistics were calculated for each cranial morphology variable, including the mean, median, standard deviation, and IQR, to summarise the central tendency and dispersion of the data. Box plots were generated to visually inspect the distributions and identify potential anomalies across sex and age groups. Additionally, normality tests, specifically the Shapiro-Wilk and Kolmogorov-Smirnov tests, were performed to determine whether the data followed a normal distribution, subsequently influencing the selection of appropriate inferential statistical tests.

Inferential statistical analysis

Chi-square tests assessed associations between categorical variables, in particular examining the relationship between sex and ethnicity with cranial morphology indices. P-values were calculated to determine the statistical significance of observed differences between subgroups. Additionally, correlation analyses were conducted to explore relationships between the different cranial morphology variables, offering insight into potential interdependencies.

Python implementation

To complement the analysis performed in SPSS, Python was utilised for advanced statistical modelling and data visualisation. Data manipulation and cleaning were performed using Pandas and NumPy, ensuring structured and efficient data handling. Visualisation techniques with Matplotlib and Seaborn were implemented to generate detailed graphical representations of cranial morphology distributions. Additionally, statistical analysis was conducted using SciPy, applying regression models and correlation tests to validate the robustness of the findings.

Results

Results of the descriptive population analysis

Before presenting the descriptive results of the morphometric variables, it is relevant to analyse the characteristics of the studied population, highlighting that most of the measured babies belong to the Baganda ethnic group.

The descriptive analysis of the data on the cranial measurements of Ugandan infants is based on the distribution of patients by sex, age, and ethnic group. Specific trends and patterns in the studied population can be identified from the tables and graphs. The data reflects a total sample of 195 infants, of whom 101 are girls and 94 are boys, representing 51.79% and 48.21% of the total, respectively. This indicates a relatively balanced distribution between both sexes, although with a slight predominance of females in the sample.

Regarding age, the babies were classified into five age groups in months and by sex, with 0 representing ‘F’ (female) and 1 representing ‘M’ (male): 0–2 months, 3–6 months, 7–12 months, 13–20 months, and over 20 months. The data distribution shows that the highest concentration of measurements is found in the youngest groups, particularly in babies aged 0–2 months and 3–6 months. This pattern may be because cranial measurements are primarily conducted in the first months of life when skull growth is more pronounced and requires closer monitoring.

The comparison of data by sex within each age group reveals that the proportion of males and females is relatively balanced in most groups. However, a slight variability in the distribution between males and females is observed in certain specific age ranges, such as 7–12 months and 13–20 months. This could be related to external factors such as the availability of infants for the study or differences in growth patterns. In the first months of life, there are more measurements, which gradually decrease as age increases.

The following graph shows all measured patients distributed by age range and ethnic group, demonstrating that the majority ethnic group is the Baganda. This is expected, as the measurements were conducted in hospitals in the central region of Uganda, specifically in the Lugazi area, where the Baganda population is predominant, followed by the Basoga ethnic group (Fig. 4).

Fig. 4.

Bar chart of patients measured according to age range and sex

Elimination of minor and extreme outliers

In the box plots generated in SPSS, atypical values were identified and classified as mild outliers (between 1.5 and 3 times the IQR) and extreme outliers (more than 3 times the IQR). As this study aims to establish normality indices for cranial measurements, these values could distort data distribution and bias the results.

Therefore, a data cleaning process was carried out, first removing extreme outliers and then mild outliers after assessing their impact. For the CVAI, values outside the ± 2.5 range were excluded to ensure that the indices reflect the expected cranial morphology (Fig. 5).

Fig. 5.

Data removal decision tree for outliers

After the filtering process presented in Fig. 5, the statistical analysis was conducted with a final sample of 145 subjects (74 females and 71 males), allowing for a more accurate and representative evaluation of the studied variables.

The definition of the reference cohort for normocephaly is based on a strict threshold for the Cranial Vault Asymmetry Index (CVAI ≤ 2.5%). This was a deliberate and rigorous methodological choice, driven by the need to reduce inherent population variability and isolate the most symmetrical cranial morphology possible, an essential requirement for generating normative data.

While commonly accepted clinical criteria, such as the Children’s Healthcare of Atlanta (CH-OA) scale, use a CVAI threshold of < 3.5% to classify asymmetry as ‘normal’ or ‘mild’ (indicating that it does not require intervention with cranial orthoses) [31, 32]. This threshold of 3.5% represents, in essence, the upper limit of clinically irrelevant asymmetry. In databases of untreated infants, a CVAI of 3.5% has been consistently reported around the 75th percentile (P75) in the general population, implying a degree of flattening greater than 75% of infants considered ‘normal’ by default [33].

This study focuses on maximizing the internal validity of the reference cohort. On the one hand, by isolating pure symmetry by applying the strictest threshold of CVAI ≤ 2.5%, individuals with mild asymmetry (2.5%< CVAI ≤ 3.5%) are eliminated, which, although not pathological, represent a significant deviation from ideal symmetry (CVAI = 0).

On the other hand, by reducing the Variability for Normative Studies, the exclusion of heads with mild asymmetries in the range of 2.5% to 3.5% ensures that the mean and standard deviation calculated for normocephaly are based on a highly homogeneous sample. This approach of selecting the most symmetrical subset, even by adjusting the cut-off points below what is clinically relevant, is a defensible practice for improving specificity and validity in the creation of normative databases, as demonstrated by the practice of other researchers of adjusting diagnostic thresholds based on the cohort and the purpose of the study [34, 35].

Therefore, this methodological approach is not used for clinical treatment decisions, but rather as a statistical selection tool to define a highly symmetrical subset of the Ugandan population, ideal for creating reference values and growth curves with minimal variability in asymmetry.

This justification can be seen in the following whisker plots (Figs. 6 and 7).

Fig. 6.

Box plot of the CVAI with outliers, in women

Fig. 7.

Box plot of the CVAI with outliers, in women

Results of the descriptive analysis of variables

After analysing the variables HC, CI, CVAI, PV, and FA, significant differences by sex were observed in several.

First, the normalised values of cranial morphology variables for the total sample (N = 195) are presented. Next, the normalised values of the same variables are detailed by age range for females (Table 1) and males (Table 2), allowing for a more specific comparison of the evolution of these measurements throughout growth.

Table 1.

Normalised values of the variables HC, CI, CVAI, PV and FA according to age group in women

Age group (months)	N	Variables
		HC (SD)	CI (SD)	CVAI (SD)	PV (SD)	FA (SD)
0–2	18	366.56 (21.85)	76.89 (4.3)	0.18 (0.58)	408.78 (51.5)	135.57 (5.83)
3–6	20	418.55 (26.25)	74.3 (2.98)	0.12 (0.38)	592.6 (130.55)	135.02 (4.98)
7–12	15	460.67 (17.78)	73.6 (3.48)	0 (0)	786.13 (119.11)	134.29 (6.43)
13–20	15	473.53 (19.17)	72.93 (1.91)	-0.25 (0.56)	845.67 (115.22)	132.43 (4.81)
> 20	6	497 (11.83)	74.17 (1.6)	-0.1 (0.25)	951.5 (76.3)	130.7 (3.63)
Total	74	431.95 (49.07)	74.5 (3.45)	0.02 (0.45)	667.51 (211.67)	134.13 (5.47)

Table 2.

Normalised values of the variables HC, CI, CVAI, PV and FA according to age group in men

Age group (months)	N	Variables
		HC (SD)	CI (SD)	CVAI (SD)	PV (SD)	FA (SD)
0–2	17	378.88 (18.74)	76.59 (4.46)	0 (0)	421.71 (49.83)	136.99 (4.2)
3–6	16	422 (32.71)	73.19 (2.66)	0.12 (0.46)	610.31 (155.81)	137.33 (3.04)
7–12	15	468 (17.13)	76 (3.84)	0.04 (0.16)	832.93 (85.45)	135.63 (3.67)
13–20	9	494.11 (10.79)	72.78 (2.33)	0.13 (0.4)	924 (67.22)	128.08 (4.2)
> 20	14	505.36 (19.74)	73.79 (2.42)	0.08 (0.31)	999.71 (182.7)	130.66 (4.96)
Total	71	446.97 (52.77)	74.66 (3.62)	0.07 (0.3)	728.73 (248.91)	134.4 (5.21)

In general terms, HC showed a higher mean in males compared to females (X̄ females = [431.95], SD = [49.07]; X̄ males = [446.97], SD = [52.77]), indicating a difference in cranial development between sexes.

CI remained relatively homogeneous between males and females, although with slight variations (X̄ females = [74.5], SD = [3.45]; X̄ males = [74.66], SD = [3.62]), without reaching statistically significant differences.

CVAI reflected a tendency towards a higher total asymmetry value in males compared to females (X̄ females = [0.02], SD = [0.45]; X̄ males = [0.07], SD = [0.3]), suggesting specific morphological differences in cranial symmetry.

On the other hand, PV helped identify differences in the cranial structure of the analysed subjects, showing higher values and more significant variability in males (X̄ females = [667.51], SD = [211.67]; X̄ males = [728.73], SD = [248.91]).

Finally, FA presented very similar values and variability in both sexes (X̄ females = [134.13], SD = [5.47]; X̄ males = [134.4], SD = [5.21]).

These results confirmed the existence of distinct patterns in cranial morphology based on sex, which could have implications for the analysis of development and potential cranial deformations in the studied population.

Results of the statistical analysis of variables

Normality analysis of the selected variables

The Shapiro-Wilk test was used to evaluate the distribution of the selected variables (HC, CI, CVAI, PV, and FA). This test allowed for determining whether the data followed a normal distribution across different age and sex groups.

The results showed that HC and CI followed a normal distribution in most groups, while CVAI, PV, and FA exhibited non-normal distributions in some cases. Since some variables did not meet the normality assumption, non-parametric tests such as Kruskal-Wallis were used for group comparisons.

Homogeneity of variances analysis

Levene’s test was performed to determine whether the variances were homogeneous across different age and sex groups. This test is essential for adequately applying variance analysis (ANOVA) in variables with a normal distribution.

The results indicated that HC and CI met the assumption of homogeneity of variances in most cases, allowing for the use of ANOVA. However, CVAI, PV, and FA exhibited heterogeneous variances in some groups, leading to the application of Welch’s t-test, a robust version of the t-test which does not assume equal variances.

Mean homogeneity analysis

In comparing means by age ranges in females, ANOVA revealed significant differences in HC, CI, and FA among different age groups (p < 0.05). To identify which groups showed significant differences, Tukey’s post hoc test was applied, finding that:

• For HC, differences were observed between the 0–2 months age group and the 7–12 and 20 + months age groups, suggesting a clear evolution in cranial development.

• For CI, the 0-2- and 3-6-months age groups differed significantly from the 20 + months age group.

• For FA, significant differences were identified between the youngest age ranges and the 20 + months age group, reflecting the influence of cranial growth.

• For CVAI and PV, where the assumption of homogeneity of variance was not met, Welch’s t-test was applied, showing significant differences in CVAI between the 0-2- and 7-12-months age groups. At the same time, PV exhibited substantial variations in the extreme age groups.

The Kruskal-Wallis test also revealed differences in CVAI and PV, reinforcing the idea that cranial asymmetry and relative volume to perimeter change significantly with age.

In comparing means by age range in males, ANOVA showed significant differences in HC and FA. Tukey’s post hoc test indicated that:

• For HC, the 0-2- and 3-6-months age groups differed significantly from the 20 + months age group.

• For FA, differences were observed between the 0–2 months age group and the 13–20 and 20 + months age groups.

Welch’s t-test revealed differences in CI and PV, indicating that variability in these indices is more pronounced in the early months of life.

The Kruskal-Wallis test detected significant differences in CVAI, suggesting that cranial asymmetry varies with age in males, similar to what was observed in females.

In summary, the analyses indicate that cranial morphology varies significantly with age in both sexes. Variables such as HC, CI, and FA show notable differences between age groups, especially in the early years of life. Likewise, the CVAI and PV exhibit variations related to cranial development and the possible presence of asymmetries. These results highlight the importance of establishing age-specific normative references to assess cranial growth and detect potential infant developmental anomalies.

Results of the implementation with python

To complement the traditional statistical analysis, an algorithm in Python was implemented to process cranial morphology data automatically, facilitating analysis and generating detailed graphical representations. These visual representations allowed for more explicit identification of cranial structure variability between different month age and sex groups.

The first approach was the extraction of the contour from 3D projections, given a 3D point cloud:

which represents the cranial surface, projecting onto various planes corresponding to standard anatomical views (axial, sagittal, and coronal). For the axial view, for instance, the projection is defined as:

Similarly, sagittal and coronal views are obtained by discarding the or components, respectively.

Let

denote the set of projected points. The convex hull [26] is computed to obtain an ordered set of points

which constitutes the contour. The centroid of the contour is then defined by

For each angle

consider the ray

For each segment

the intersection is computed by solving

The radial function is defined as

For a group of subjects, the mean and standard deviation functions are given by

By projecting the 3D surface onto traditional anatomical planes, one obtains contours directly comparable across subjects, permitting statistical analysis of cranial shape. This technique assumes consistent acquisition protocol and anatomical alignment, which minimises the need for complex registration algorithms [27].

Based on these representations, aggregated image matrices were generated. These images were classified by age and sex groups, allowing for an observation of how cranial shapes evolved throughout growth. To assess cranial morphology variability, comparisons were made between two datasets: one including all measured values and another in which outliers were removed. The results showed that the presence of outliers increased data dispersion and significantly raised the standard deviation lines. When outliers were removed, the resulting matrices were more homogeneous and representative of the average cranial shape in each group (Figs. 8 and 9).

Fig. 8.

Mean cranial perimeter matrix (no outliers) in axial, sagittal, coronal planes by female age range

Fig. 9.

Mean cranial perimeter matrix (no outliers) in axial, sagittal, coronal planes by male age range

The growth analysis revealed a predictable pattern in cranial development, characterised by progressive expansion in all directions as age increased. Additionally, in the sagittal and coronal views, more significant variability was detected in the lower part of the skull, suggesting possible differences in the inclination of the measurement cap used for data collection.

Specific matrices for males and females were generated for the sex-based analysis, excluding outliers, to represent cranial morphology in each group accurately. Males were identified to have a larger average cranial perimeter across all age ranges, while females exhibited a slight elongation in the anteroposterior axis. Moreover, cranial expansion in males was more uniform in all directions, whereas females displayed variations in cranial proportions. In both sexes, overall variability decreased with age.

The mean HC values by age and sex reflected these growth differences:

• 0–2 months: Females: 366.56 mm ± 21.85 mm | Males: 378.88 mm ± 18.74 mm.

• 3–6 months: Females: 418.55 mm ± 26.25 mm | Males: 422 mm ± 32.71 mm.

• 7–12 months: Females: 460.67 mm ± 17.78 mm | Males: 468 mm ± 17.13 mm.

• 13–20 months: Females: 473.53 mm ± 19.17 mm | Males: 494.11 mm ± 10.79 mm.

• > 20 months: Females: 497 mm ± 11.83 mm | Males: 505.36 mm ± 19.74 mm.

These findings reinforce the analysed sample’s general growth trends and morphological differences between sexes.

Anomalous cranial morphology cases

Two anomalous cases were selected from the previously identified outliers to further illustrate the cranial morphology variability within the studied population. These cases, one male and one female, both belong to the 0–2 months age range and exhibit significant deviations from the established reference values.

On the one hand (Fig. 10), showcases a female case displaying evident cranial asymmetry compared to the mean. This asymmetry is particularly pronounced in the right posterior region, potentially suggesting right-sided posterior plagiocephaly. This type of cranial deformation is commonly linked to positional factors, such as sleeping posture, and, if left untreated, it can cause aesthetic [26, 36] and functional problems, including delays in neurocognitive and motor development [37, 38], although the debate continues as to whether this relationship is causal or correlational.

Fig. 10.

Aggregate matrix image of the HC of sample 116 (female 0–2 months) in the axial plane

On the other hand (Fig. 11), the selected male case, presents an elongated and disproportionately large cranial morphology compared to the mean reference shape (highlighted in red). This cranial configuration may indicate dolichocephaly or scaphocephaly, a condition characterised by excessive elongation of the skull; scaphocephaly is often associated with sagittal suture closure anomalies. To ensure the validity of the normative cohort, all participants were pre-screened by the hospital paediatrician. This screening aimed to exclude major pathological congenital malformations (e.g., microcephaly, hydrocephalus, or obvious clinical suspicion of craniosynostosis). Infants with visible pathological deformities at this stage were excluded from the study before cranial measurement. The observed deviation underscores the necessity of early detection, as untreated scaphocephaly can influence cranial growth and neurodevelopmental outcomes.

Fig. 11.

Aggregate matrix image of the HC of sample 28 (male 0–2 months) in the axial plane

These cases highlight the importance of early screening and preventive strategies, particularly in regions with limited access to specialised paediatric neurosurgical evaluations. By integrating photogrammetry-based screening tools and community awareness programs, the timely identification and management of cranial deformities could be significantly improved, reducing the risk of long-term complications.

Discussion

This study represents an essential step in understanding cranial morphology in the paediatric population of Lugazi, Uganda. Only full-term infants were considered in this study. By analysing HC, CI, CVAI, and other cranial indices such as PV and FA, we have established reference values that contribute to the early detection of cranial deformities and inform clinical practice and future research. Our findings provide essential baseline data for assessing cranial development in this specific population and highlight potential risk factors that may contribute to abnormal cranial growth. These findings are particularly relevant to the paediatric health initiatives conducted in Lugazi, including our collaboration with local healthcare providers.

Technical challenges

The methodological approach of this study involved smartphone-based PhotoMeDAS 3D scanning and precise measurement of various cranial indices of the patient and their statistical analysis using SPSS. One main challenge was ensuring measurement consistency and accuracy, particularly in a field setting with limited access to advanced imaging technology. The use of caps coded for scanning proved to be a reliable tool for obtaining standardised measurements. However, variability in data collection, particularly in younger infants with lower compliance during measurements, required additional validation steps to ensure data reliability. Moreover, differences in head shape patterns across ethnic groups and age categories needed a careful approach to data interpretation.

Additionally, the study was conducted in various clinical settings, including the public hospital Kawolo and the Living Water medical clinic, where logistical constraints sometimes limited the number of infants that could be assessed daily. Despite these challenges, the data collected provides a solid foundation for establishing cranial growth reference values tailored to the paediatric population in Lugazi.

Clinical relevance of the cranial indices

The results of our study align with previous research on cranial asymmetry and cephalometric indices in paediatric populations [17–19, 39–42]. The CI, a widely used parameter for classifying head shapes, showed significant variations according to age and sex, confirming that head growth patterns are dynamic and subject to individual and environmental influences. Similarly, the CVAI values in our study provided a critical reference for evaluating the prevalence of positional plagiocephaly and other cranial asymmetries in this population. Given that early intervention is crucial for managing cranial deformities, the reference values established in this study could aid healthcare professionals in diagnosing and monitoring cranial shape abnormalities more effectively. However, early detection would have required much finer time intervals to reflect the dynamics of cranial growth. The choice of an age range of 0 to 2 years in this study responded to the need to establish an anthropometric baseline during the period of maximum cranial growth and plasticity (Muñoz Cabeza & López Lobato, n.d.; Sgouros et al., 1999) where approximately 80% of head growth takes place (Dekaban, 1977). Since treatment of positional deformities is most effective before 5–6 months of age (Kluba et al., 2014b), for early detection and optimal monitoring of anomalies, much narrower age intervals are requested.

Furthermore, comparing our data with previously reported indices in other populations suggests potential genetic and environmental influences on cranial morphology, warranting further investigation.

These findings are particularly relevant to the paediatric campaign developed as part of the broader public health initiative in Lugazi, which aims to improve child health outcomes by integrating cranial measurements into routine paediatric check-ups. The collaboration with Village Health Teams in Makendu, Kisaala, and Kitega has facilitated the dissemination of this information to local healthcare workers, ensuring that these new reference values can be applied in clinical practice.

Data quality and limitations

The quality of the collected data was a key consideration in ensuring the robustness of our findings. While manual measurement techniques are employed in other studies, we adopted a more standardised and less variable approach using specialised software. Nonetheless, certain limitations must be acknowledged. First, while representative of the Lugazi paediatric population, the study sample may not capture the full diversity of cranial morphology across Uganda. Second, the lack of longitudinal data restricts our ability to track individual cranial development over time. Third, a standard limitation in surface photogrammetry is the impact of hair. However, the PhotoMeDAS system mitigates this problem by relying on elastic caps with coded targets (QRs). The software extracts cranial measurements solely from the triangulation of these codes with high precision, allowing it to automatically discard surface artefacts, such as hair, skin, or reflections. It is recognised, however, that excessively dense or long hair could introduce slight variations in the extraction of the cap dimensions on the head, which should be considered in the interpretation of field data. Nevertheless, anthropometric indices are based on differences or ratios, so the results are minimally altered. Fourth, the system is fully verified and in the process of being validated. The results will be presented to the scientific community soon.

Future research should focus on expanding the dataset with follow-up measurements to establish growth trajectories and assess potential long-term implications of early cranial shape variations.

Implications for paediatric healthcare

Establishing localised cranial reference indices is essential for improving paediatric healthcare and guiding clinical decision-making. The findings from this study emphasise the importance of routine head measurements in early childhood check-ups to detect abnormal cranial growth patterns before they become clinically significant. Additionally, raising awareness among healthcare providers and caregivers about the risk factors associated with cranial deformities, such as prolonged supine sleeping positions or limited tummy time, could contribute to preventive strategies.

This study also supports ongoing community health education efforts, as demonstrated in the sensitisation sessions held in local schools, where information on child health and development was shared with teachers and parents. By integrating cranial health education into existing outreach programs, we can ensure that these findings translate into tangible improvements in paediatric care.

The establishment of normative values for the Cephalic Index (CI) and Cranial Volumetric Asymmetry Index (CVAI) in infants is an active area of research, driven by the need for updated references following the “Back to Sleep” campaign and the consequent increase in positional cranial deformities. Comparing our findings in the Ugandan cohort with the most recent Western normative references reveals key differences that have significant clinical implications.

Contemporary studies in populations in the United States and other Western areas have worked to re-evaluate and update the cranial indices used as the norm [43]. have reviewed the Cephalic Index standards in a large, diverse population, and [44] have examined the concept of the “new normal” for the IC. These works are crucial, as is the narrative review of [49] which analyses the current applicability of the Cephalic Index.

By comparing the average IQ and CVAI values observed in the Ugandan population with these modern references [47–49], it can be confirmed that the Ugandan population exhibits slight deviations in cranial morphology, particularly towards greater dolichocephaly (average CI of 75%) compared to contemporary Western cohorts. This divergence is likely and fundamentally attributed to population-specific environmental, cultural, and genetic factors, such as infant carrying practices, prone/supine sleep time, or ethnic variations. This discrepancy highlights the limitation of applying global reference values. It strongly supports the need to establish growth curves and references specific to diverse geographic and ethnic populations, such as those of Uganda.

Future directions

Building upon the current findings, future studies should integrate additional diagnostic tools, such as low-cost smartphone-based photogrammetric 3D cranial scanning, to enhance measurements and provide more comprehensive data on cranial morphology. Expanding the research to include other regions of Uganda would also allow for comparative analyses and a deeper understanding of regional variations in cranial development. Furthermore, investigating the impact on cranial shape of nutritional status, socioeconomic factors, and cultural childcare practices could provide valuable insights into the multifactorial nature of cranial growth patterns.

Strengths and contributions of the study

Despite its limitations, this study significantly contributes to paediatric cranial research in Uganda. It establishes a critical foundation for assessing cranial morphology outside of high-income countries, which could inform clinical interventions and future public health initiatives. Integrating anthropometric and cephalometric data into routine paediatric assessments can potentially improve early diagnosis and management of cranial deformities, ultimately enhancing child health outcomes in the region. Furthermore, the collaboration with community health workers and local healthcare facilities strengthens the sustainability of these findings, ensuring their application in paediatric care beyond the duration of this study.

Conclusions

This study provides a comprehensive analysis of cranial morphology in Ugandan infants, establishing reference indices that contribute to a more precise understanding of standard cranial development patterns in this population. Integrating photogrammetric smartphone-based data collection and advanced statistical modelling has enabled the identification of key morphological parameters, facilitating early detection and potential preventive strategies for cranial deformities.

The results indicate significant variations in head circumference (HC), CI, CVAI, PV, and FA across age groups and between sexes. HC and PV showed a progressive increase with age, aligning with established paediatric growth trajectories, yet exhibiting slight deviations from global reference values, reinforcing the need for region-specific normative data. The CI classification revealed a predominance of mesocephalic morphology, though some cases displayed borderline dolichocephalic or brachycephalic traits, suggesting potential variations influenced by genetic and environmental factors. The CVAI findings highlight a small percentage of infants with moderate to severe asymmetry, reinforcing the importance of early screening and caregiver education to prevent positional plagiocephaly.

This research’s novel aspect lies in the PV and FA, which provide additional insights into cranial shape variations and potential early indicators of developmental irregularities. These parameters may help refine cranial assessment protocols, but further studies are needed to determine their clinical applicability.

Implementing interactive dashboards for data visualisation has also proven a valuable methodological innovation. These dashboards allow for real-time analysis of cranial parameters, enhancing the usability of collected data for medical and research applications. The ability to dynamically compare cranial morphology trends across demographic variables provides a new avenue for understanding developmental patterns and improving early intervention strategies.

Despite the success of this study, some challenges were encountered. The limited access to advanced imaging technologies in rural settings underscores the need for alternative, cost-effective diagnostic tools such as photogrammetric techniques. Additionally, language barriers required local translators, highlighting the necessity of culturally adapted educational materials for community-based interventions. The lack of follow-up mechanisms for detected cases also represents a limitation, emphasising the importance of integrating cranial screening programs into paediatric healthcare infrastructures.

In conclusion, this study sets a foundation for developing localised cranial growth standards in Uganda, facilitating more accurate clinical assessments and strengthening preventive healthcare initiatives. Future research should focus on the longitudinal tracking of cranial development, refining photogrammetry-based diagnostic protocols, and incorporating artificial intelligence models to enhance screening accuracy. By combining advanced technology, community-based education, and multidisciplinary collaboration, this research contributes to the long-term goal of reducing the incidence of preventable cranial deformities and improving paediatric healthcare outcomes in low-resource settings.

Abbreviations

PhotoMeDAS

Photogrammetric Medical Deformation Assessment Solutions

QR

Quick response

HC

Head circumference

CI

Cephalic index

CVAI

Cranial Vault Asymmetry

PV

Pseudovolume

FA

Frontal angle

SPSS

Statistical Package for the Social Sciences

ANOVA

Analysis of Variance

AAP

American Academy of Pediatrics

SIDS

Sudden Infant Death Syndrome

CAT

Computed Axial Tomography

RMN

Magnetic Resonance Imaging

3D

Three dimensions

IQR

Interquartile Range

Authors’ contributions

Conceptualization, D.V.-C. and J.L.L.; methodology, D.V.-C., J.L.L. and M.M.-D. ; software, M.M.-D. and M.R-F; validation, D.V.-C. and J.L.L.; formal analysis, M.M.-D. ; investigation, M.M.-D.; resources, D.V.-C., J.L.L. and M.M.-D. ; data curation, M.M.-D. and M.R-F; writing—original draft preparation, M.M.-D. ; writing—review and editing, M.M.-D., , J.L.L.; supervision, D.V.-C., J.L.L. ; project administration, D.V.-C. and J.L.L. ; funding acquisition, J.L.L. and M.M.-D. All authors have read and agreed to the published version of the manuscript.

Funding

Instituto de Salud Carlos III has funded this project under project number PI22/01416, with joint financing by the European Union and the invaluable support of the development cooperation grant Programa de Cooperación al Desarrollo (PCD) awarded by the Centro de Cooperación al Desarrollo (CCD), UPV with the economic agreement of Meridies, PCD y Adsideo − (10301 CCD). This grant supports technical science projects in developing countries, fostering a bidirectional exchange of social and scientific knowledge. This initiative aims to provide technical expertise while gaining insights from the local context, contributing to mutual learning and sustainable development.

Data availability

The datasets used and/or analysed during the current study are not publicly available. If you are interested, please write to [mmuodom@upv.edu.es](mailto: mmuodom@upv.edu.es) .

Declarations

Ethics approval and consent to participate

This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Research Committee of the Universitat Politècnica de València (protocol code no. P03-29-04-2022 on 27 July 2023). Additionally, the project was endorsed by the Centro de Cooperación al Desarrollo (CCD) of the Universitat Politècnica de València (UPV) in collaboration with the NGO Destino (Kitega Community Centre, Uganda). This partnership was essential for the planning, coordination, and execution of the study, ensuring that ethical research principles and community engagement practices were carried out.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.