Jay Sunil Sawant
BME MS Thesis Defense Presentation
Date: 2026-04-14
Time: 4:00 PM - 6:00 PM
Location / Meeting Link: https://us04web.zoom.us/j/71806985265?pwd=ipLH15wKisKFn7ySgMJpr8QYFpk2xh.1

Committee Members:
Dr. Cassie Mitchell; Dr. Peng Qiu; Dr. Karthik Menon


Title: ISM INSIGHT: AN AUTOMATED COMPUTER VISION-DRIVEN SOFTWARE TO STREAMLINE INTERSPINOUS MOTION MEASUREMENTS FOR THE POST-OPERATIVE ASSESSMENT OF ACDF SURGERY

Abstract:
Automatic assessment of interspinous motion (ISM) from flexion and extension radiographs can improve postoperative fusion evaluation after the anterior cervical discectomy and fusion (ACDF) surgery by reducing subjectivity and manual effort. This work presents a software framework that integrates cervical spinous process segmentation, contour-based landmark extraction, and ISM calculation within a single pipeline. A novel annotated dataset was developed, and an enhanced CervSpineNet model using a pre-trained SAM ViT-B encoder with a custom residual and squeeze-and-excitation decoder was implemented for segmentation. The segmented masks were then processed to extract contours and identify consistent landmark points near the spinous process tips for distance-based ISM estimation. Experimental results on original and augmented datasets showed strong segmentation performance, while the ISM pipeline demonstrated close agreement with expert measurements, with an overall average mean difference of 0.830 mm. These results were incorporated into a web-based interface for automatic ISM calculation and postoperative fusion assessment.