The Spectrum of Severity in 368 Patients with Metopic Craniosynostosis: An Update to the CranioRate™ Machine Learning Algorithm
Anne E. Glenney*1, Joseph Mocharnuk1, Griffin Bins2, Erin Anstadt1, Lucas Dvoracek1, Wenzheng Tao3, Ross Whitaker3, Lisa David2, Christopher Runyan2, Michael Golinko4, Michael Alperovich5, Jesse Taylor6, Jordan Swanson6, Jesse Goldstein1
1Plastic Surgery, UPMC, Pittsburgh, PA; 2Plastic Surgery, Wake Forest, Winston Salem, NC; 3Computer Science, University of Utah, Salt Lake City, UT; 4Plastic Surgery, Vanderbilt, Nashville, TN; 5Plastic Surgery, Yale, New Haven, CT; 6Plastic Surgery, University of Pennsylvania, Philadelphia, PA
CranioRate™ is a publicly available, point-of-care analysis tool which utilizes machine learning to quantify morphologic severity in patients with metopic craniosynostosis. Here, we present a detailed examination of a large cohort of imaging samples to understand the spectrum of severity and to quantify drivers of clinically appreciable severity in metopic craniosynostosis.
The CranioRate™ machine learning algorithm provides two objective, holistic metrics for quantifying severity in metopic craniosynostosis: Metopic Severity Score (MSS) and Cranial Morphology Deviation (CMD). De-identified CTs from normal and metopic patients from multiple institutions across the U.S. were compiled and analyzed using descriptive statistics, demographic associations, and regression analyses.
A total of 460 CT scans (92 normal patients, 368 metopic patients) from five institutions across the U.S. were uploaded to CranioRate™. Average age at CT was 0.75 ± 0.51 years, and 74.0% of patients were male. Among normal controls, average MSS was 0.00 ± 1.04, and average CMD was 85.23 ± 19.32. Among metopic patients, MSS averaged 5.02 ± 2.41, and CMD averaged 192.20 ± 44.62. Both MSS and CMD were significantly different between control and metopic patients (p<0.0001); no severity differences were noted between centers (p>0.05). There was a positive correlation between severity and earlier age at CT (MSS: r= 0.0043±0.00037, p<0.0001; CMD: 0.2676±0.02667, p<0.0001). Regression analysis identified the central frontal bone, lateral orbit, and supraorbital rim as the regions most associated with severity differences (p<0.05).
Our results are the first to objectively derive the regions of the skull most associated with phenotypic severity in metopic craniosynostosis and to establish a temporal relationship between severity and patient presentation. As we collate more scans from across the U.S., we hope to approximate the full spectrum of severity among the broader population of metopic patients.
Back to 2023 Abstracts