A
team lead by
Prof. Sang Hyun Park (First Author: Philip Chikontwe) has recently
published an article in
Medical image analysis journal (IF: 13.82). They show that
a deep learning-based approach for weakly supervised segmentation only using
image-level labels in Histopathology slides has clinical relevance. They address
the scenario where fast diagnosis is required in memory constrained settings
when datasets are limited and/or contain weak annotations without infected
areas highlighted.
Overall,
the research introduces algorithms and strategies to solve accurate biomedical
data segmentation, robust feature extraction and analysis. The findings
presented will appeal to the research community applying machine/deep learning
for histopathology analysis including other imaging modalities.
