Deep Clustering Activation Maps for Emphysema Subtyping
This work addresses emphysema subtyping for medical imaging analysis, offering interpretability through visualization, but it is incremental as it builds on existing clustering and segmentation methods.
The paper tackled emphysema subtyping from CT scans by proposing a deep learning clustering method that uses dense features for high-resolution visualization, achieving 43% unsupervised clustering accuracy, which outperformed a baseline at 41% and was comparable to supervised classification at 45%.
We propose a deep learning clustering method that exploits dense features from a segmentation network for emphysema subtyping from computed tomography (CT) scans. Using dense features enables high-resolution visualization of image regions corresponding to the cluster assignment via dense clustering activation maps (dCAMs). This approach provides model interpretability. We evaluated clustering results on 500 subjects from the COPDGenestudy, where radiologists manually annotated emphysema sub-types according to their visual CT assessment. We achieved a 43% unsupervised clustering accuracy, outperforming our baseline at 41% and yielding results comparable to supervised classification at 45%. The proposed method also offers a better cluster formation than the baseline, achieving0.54 in silhouette coefficient and 0.55 in David-Bouldin scores.