VI Seminar #76: Explainable methods for computer aided diagnosis 

Presenter: Anuja Vats, Postdoctoral Researcher at DART and Colorlab, Department of Computer Science, NTNU.

Abstract

For AI assistance to gain meaningful clinical acceptance, it must be paired with explainability - a slippery concept that needs to be tailored to domain and application needs.

In this talk, I will present explainable methods for computer-aided diagnostic tasks, with a focus on wireless capsule endoscopy - a challenging domain due to the complexity and variability of gastrointestinal imagery. I will explore three complementary approaches to explanation: counterfactual explanations that illustrate "what-if" scenarios, uncertainty quantification methods that communicate model confidence, and briefly touch upon concept-based interpretability techniques that can align with medical reasoning patterns.

In compliance with GDPR consent requirements, presentations given in a Visual Intelligence context may be recorded with the consent of the speaker. All recordings are edited to remove all faces, names and voices of other participants. Questions and comments by the audience will hence be removed and will not appear in the recording.  With the freely given consent from the speaker, recorded presentation may be posted on the Visual Intelligence YouTube channel.