Assessing Polyp Detection Delay Using Tracklets

Samuel Gussman-Toh, Nicholas Petrick, Wei-Chung Cheng, Feng Yang

Keywords: assessment, detection delay, computer aided detection, CADe, regulatory science, colonoscopy

TL;DR: This study introduces a tracklet-based approach for estimating the polyp detection delay for a colon CADe system as part of overall standalone performance testing.

Abstract: This study introduces a tracklet-based approach for estimating the polyp detection delay for a colon CADe system as part of overall standalone performance testing. We propose using detection delay as a benchmarking metric that is agnostic to processing time and dataset modifications. Our study addresses how polyps frequently come in and out of frame, creating distinct tracklets, and examines the impact this behavior can have on detection delay calculations. Lastly, we introduce detection delay curves to compare model performance over a range of sensitivities and use detection window curves to measure how sensitivity improves the longer a model observes a polyp.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 4