Katarzyna Bozek

Slide-Level Active Learning Reduces Annotation Burden in H&E images

Mahsa Vali, Zhilong Weng, Noémie Moreaua, Yuri Tolkach, Katarzyna Bozek

Deep learning-based segmentation of histopathology whole-slide images (WSIs) requires large amounts of pixel-level annotations, which are costly and time-consuming to obtain. Active learning (AL) has been proposed to reduce this effort, but existing methods exhibit three key limi…Read more