Back
October 2, 2026
•
3 minute read

Quantifying Infant and Pediatric Brain Development from Routine Clinical Imaging

Blogs
Pediatric
Population Brain Charting

Share Article

Table of Contents

New research from Centile Bio introduces two complementary tools for quantifying pediatric brain and skull growth from routine clinical imaging.

Clinicians may use MRI to investigate seizures or developmental concerns, or CT to assess an injury. Those scans, acquired to answer an immediate clinical question, also contain quantitative biomarkers of brain and skull development useful to characterize underlying pathology. 

Most neuroimaging quantification solutions face particular challenges in early brain development, when the brain and skull change rapidly, and clinical imaging varies across scanners, modalities, and protocols. 

Children who could benefit the most from precise measurement have often been the most challenging to assess.

Two new preprints from the Centile Bio team and collaborators address this exact problem. ClinSeg measures brain structures in pediatric MRI, including newborn scans, and CranioTrace measures head circumference and cranial shape from MRI and CT. Both tools work on existing clinical scans with no additional imaging and score measurements against population reference charts beginning at birth. Both tools are integrated into our platform.

Why the first months of life are so challenging to measure

During infancy, the brain is actively forming myelin, which insulates nerve fibers. This process causes gray and white matter to change their appearance on MRI: initially, their contrast is reversed compared to adults, then they briefly appear similar, and eventually they adopt the adult pattern. Clinical scans add further variation because they are acquired using different scanners, settings, and protocols. As a result, studies based on hospital imaging archives often exclude infants younger than six months.

Measuring head circumference also presents challenges. Tape readings can vary among examiners. Obtaining accurate measurements may be difficult in sedated children or those with craniofacial differences, and sometimes these data are missing from the medical record. A tape measure records head size but not shape.

ClinSeg: bringing infants into quantitative brain measurement

ClinSeg is a brain segmentation model designed for the practical challenges of clinical pediatric MRI, where orientation, resolution, and image contrast often vary. The team trained the model on expert-segmented infant scans as well as synthetic scans. These synthetic scans replicate different stages of myelination and a broad spectrum of anatomical features, including structural abnormalities observed in clinical imaging.

The model performed best where clinical need is greatest: the first year of life.

- Infant accuracy: On infant scans with expert manualsegmentations, ClinSeg's results were more closely aligned with experts thanthose of two widely used tools. In older children and adolescents, itsperformance was similar to these tools.

- Newborn scans: In an analysis of 78,508 clinical brain MRIs,none of the newborn scans passed automated quality control using a widely usedtool. With ClinSeg, nearly all newborn scanning sessions met quality criteria.

- Early infancy: Thousands of additional scans from babiesaged one to six months could be measured using this approach.

The availability of these newly measurable scans enabled the construction of brain growth charts from 11,699 individuals, spanning birth to age 21, using a single model across the pediatric period. The charts reproduce well-established developmental patterns:

- Sensory regions reach peak volume earliest, around age four.

- Regions involved in higher-order thinking reach their peak later, with some peaking around age 13.

CranioTrace: the tape measure, brought into the scan

CranioTrace automatically measures head circumference and cranial shape from brain MRI and CT, incorporating built-in quality checks and requiring no manual steps. The team constructed sex-specific reference charts from 9,685 scans across 25 cohorts, covering birth to age 20. The charts use the same statistical framework as the World Health Organization's head circumference standards.

- Agreement withtape: Imaging-based measurements agreed closely with tape, with averagedifferences of less than 0.2 cm across four cohorts.

- Consistency: Repeat scans acquired during the same session yielded nearly identical results.

- Quality review: 97.3% of 1,000 manually reviewed scans met quality standards.

CranioTrace is not intended to replace the tape measure. Instead, it provides a consistent measurement from scans a child has already undergone. This is especially valuable when a tape reading is missing or difficult to obtain, or when a clinician wishes to compare measurements with earlier imaging. Because CranioTrace also works on CT, it can be used for children whose only brain imaging was performed during an emergency visit.

Two heads with the same circumference may have distinctly different shapes. CranioTrace addresses this by assessing eight measures of cranial shape and asymmetry, each charted across development. The charts also illustrate the rapid pace of early growth. At birth, head circumference increases by about 0.8 millimeters per day, and by age two, this rate slows to less than 0.1 millimeters per day.

The intersection of brain and skull development

Both studies tested their methods in children with genetic conditions that affect brain and head development.

With CranioTrace, head circumference differed from the population reference in the expected direction for each condition. It was larger on average in NF1 and 16p11.2 deletion and smaller in 22q11.2 deletion and 16p11.2 duplication. Children with NF1 also demonstrated a distinct cranial shape pattern beyond what head size alone would predict.

The team also built condition-specific growth charts, even for small groups. For a child with a genetic condition, comparing against children with the same condition may offer more useful context than comparing against the general population alone.

ClinSeg was used to revisit 22q11.2 deletion syndrome, a condition well studied in older children but rarely in infants. The model enabled measurement in 15 infants under age one who previously could not be assessed. The findings confirmed brain differences reported in earlier studies and suggested that the thicker cortex characteristic of the condition appears to develop after infancy.

Together, these two studies contribute to a more comprehensive understanding of early development, with both skull and brain measured from the same scans and compared to population references using consistent methods.

What this means for our partners

For clinicians and radiologists, these methods provide consistent, age-referenced measurements to complement their expert interpretation. For researchers, archived infant and pediatric scans can now be analyzed without additional imaging. For drug developers studying rare pediatric conditions, where cohorts are small and distributed across multiple sites, these population-referenced brain and cranial measurements support patient characterization and stratification using existing imaging.

About the research

Both studies are available as preprints prior to peer review, providing the research community with early access to the methods and findings.

The studies were co-authored by members of the Centile Bio team and scientific advisory board, including Jakob Seidlitz, Aaron Alexander-Bloch, Philip Mattisson, Anna Zapaishchykova, Joshua Bernstock, and Richard Bethlehem.

References

1. Levitis E, et al. ClinSeg: Robust Brain Segmentation for Clinically Acquired Pediatric MRI. medRxiv. 2026. doi:10.64898/2026.08.28.26361643
2. Mattisson P, et al. Charting neuroimaging-based head circumference and cranial shape across development. bioRxiv. 2026. doi:10.64898/2026.08.29.747918
3. World Health Organization. WHO Child Growth Standards: Head Circumference-for-Age, Arm Circumference-for-Age, Triceps Skinfold-for-Age and Subscapular Skinfold-for-Age: Methods and Development. Geneva: World Health Organization; 2007. https://www.who.int/publications/i/item/9789241547185

Learn more about how our platform brings accurate population-referenced brain quantification to pediatric imaging.

Explore a partnership