Similarity-based uncertainty scores for computer-aided diagnosis
In person: 23 February 2022 • 5:30 PM - 7:00 PM PST
There is a significant amount of research in applying deep learning to medical image classification. The NIH/NCI Lung Image Database Consortium (LIDC) data set allows these techniques to be tested and applied on lung nodule data. It incorporates multiple nodule ratings, including the degree of spiculation, a visual characteristic. Our ultimate motivation is to improve resource allocation during this process. We aim to flag ambiguous cases that may require more time or more opinions from radiologists. We propose to show a correlation between radiologist semantic disagreement on spiculation ratings and cases with a high level of uncertainty based on our novel methodology. We found that the nodule images which fell under the highest 33% of our uncertainty scores captured more than 50% of the data with low and no radiologist agreement on spiculation. Our results flag more images in the spiculated category, suggesting that more important disagreements had been captured.
Whitman College (United States)
Claire Weissman is a junior computer science major at Whitman College in Walla Walla, Washington. She is interested in interdisciplinary computer science especially when related to the medical field. Claire has participated in two REU programs and has spent this summer doing research at Depaul University's REU program. Her work this summer lies in the Computer Aided Diagnosis field and seeks to assist radiologists in diagnosing "hard" cases in lung nodule CT scans. Claire explored this problem using neural networks and T-SNE visualization.