Grants We Funded

Abstract

Impact Statement: This grant proposal aims to improve our understanding of craniosynostosis by expanding on our prior work using machine learning methodologies to objectively quantify the head shape anomalies in this condition. During the project, we will further develop a point-of-care tool, CranioRate, to allow physicians to improve their understanding of their patients and create an accessible collection of imaging data so researchers can more easily study this challenging population of patients. With this information, surgeons will be able to better tailor their interventions to the needs of their individual patients, and researchers will improve their understanding of this complex disorder.

Project Summary: The purpose of this research grant application is to build on the advanced machine learning (ML) tool developed as part of a pilot study (R21EB026061) that objectively quantifies cranial dysmorphology, or deep phenotypes, in patients with metopic craniosynostosis (MC). Abnormal cranial suture fusion (craniosynostosis) occurs in one of every 2500 infants born in the US, resulting in disrupted regional skull growth and an increased risk of elevated intracranial pressure, neurocognitive impairment and visual disturbances including blindness. Impaired skull growth along the fused suture and subsequent growth compensation in other areas of the skull lead to predictable head shape patterns in patients with craniosynostosis; surgery is recommended early in childhood to restore normal head shape and prevent neurocognitive sequelae. In our work to date, our team has developed an ML/statistical shape analysis system utilizing computed tomography (CT) scans of patients with MC. We have demonstrated that our deep ML algorithm is as effective as expert clinician ratings in assessing severity and more effective than standard craniometric tools. We have implemented an online head shape portal (CranioRate) that automates preprocessing and analysis such that users can upload their own patient images, where the resulting data contributes to clinical patient care as well as research endeavors. To date, over 30 clinicians have contributed almost 400 MC CT scans to our portal, making our metopic craniosynostosis imaging collection the largest reported. In the proposed work, we will refine our processing pipeline and shape analysis technologies, while expanding our capabilities to encompass all forms of craniosynostosis and a wider array of imaging modalities, and improve the functionality and security of the CranioRate portal. Specific goals for the current project are to: 1) Further develop a set of robust, general morphological quantification technologies and cloud-based implementations that result in effective scientific and clinical tools; 2) Extend the current methodologies to additional forms of craniosynostosis; 3) Identify and collect pertinent radiographic data to extend the utility of our shape analysis tool and shared-access database. The results of this study will significantly improve the understanding of the phenotypic variation in patients with craniosynostosis and will pave the way for more substantial imaging-based research in this understudied population.

Biography
Dr. Jesse A Goldstein is an Associate Professor of Pediatric Plastic Surgery at the University of Pittsburgh Medical Center and Children's Hospital of Pittsburgh of UPMC s Hospital of Pittsburgh where he is the Program Director for the Pediatric Plastic and Craniofacial Surgery Fellowship, Director of the Cleft and Craniofacial Center, and Associate Program Director of the Department of Plastic Surgery Residency Program. He has been at Pitt since 2014 after completing his training at Georgetown University, the Children’s Hospital of Philadelphia, and University of Pennsylvania. Dr. Goldstein serves on the board of the Robert H. Ivy Society, Ohio Valley Society of Plastic Surgeons, and has chaired numerous committees for the American Society of Pediatric Plastic Surgery. Dr. Goldstein has authored over 70 peer-reviewed articles and 20 book chapters. He has served on several study sections for the NIDCR. His research focuses on diagnostic machine learning and imaging for craniosynostosis (NSF/NIH) as well as outcomes of craniofacial surgery and health services.