Precision medicine approach to Osteoarthritis: Automatic cartilage segmentation
Under the mentorship of Dr. Stephanie Jo at the Hospital of the University of Pennsylvania during the summer of 2021, I performed data extraction on knee MRI images of patients experiencing Osteoarthritis. This made up half of the Jo lab's broad project goal of using MRI findings of Osteoarthritis and associating them with SNP genetic biomarkers. The completion of the project in its entirety would validate the use of these SNP biomarkers as measurable predictors of OA risk to allow for precise surveillance, to manage treatment decisions, and to facilitate clinical interventions for Osteoarthritis.
More particularly, I was responsible for implementing a machine learning Python package, DOSMA, to automatically segment femoral, tibial, and patellar cartilage volume instead of using a time-intensive manual method. Cartilage volume measurements were made at each participant's baseline and and 24-month MRI followup. Eventually, percent volume loss between both these time periods will be calculated and this value will be used to characterize Osteoarthritis severity. The tool of using distinct SNPs to indicate Osteoarthritis risk will be developed by linking participant MRI findings with their SNP genotype to eventually construct a biological trend across the entire study sample.