Explainable AI assisted Ultrasound Thyroid Classification

Ultrasound thyroid nodule grading is a system used to categorize thyroid nodules based on their characteristics observed during ultrasound imaging, typically between 1 and 5. Classes 1 and 2 have a low suspicion for malignancy, whereas 4 and 5 have high suspicion for malignancy. This project stems from clinical audits of ultrasound thyroid nodule classification, where it was deemed a potentially viable Explainable AI (XAI) research project.

This research project is investigating the potential for the use of XAI in providing additional context to the grading of thyroid nodules, particularly for borderline classifications like class 3. This research study was a part of a clinical AI research internship offered by the Mater Hospital to a PhD candidate, working on XAI.