Bidur Khanal

Research evaluates self-supervised techniques to improve robustness against noisy labels in medical imaging.

A new method enhances classification despite noisy labels and imbalanced datasets.

DINOv2 offers promising results in imaging the left atrium for better heart care.