Label Super Resolution with Inter-Instance Loss
This work addresses the high cost of pixel-level ground truth collection in medical imaging by enhancing label super resolution methods, though it is incremental as it builds on existing LSR approaches.
The authors tackled the problem of semantic segmentation with low-resolution supervision by proposing a novel loss function that models inter-instance variance, improving label super resolution for high-resolution images like gigapixel pathology slides, with experimental validation on infiltrating breast cancer region segmentation.
For the task of semantic segmentation, high-resolution (pixel-level) ground truth is very expensive to collect, especially for high resolution images such as gigapixel pathology images. On the other hand, collecting low resolution labels (labels for a block of pixels) for these high resolution images is much more cost efficient. Conventional methods trained on these low-resolution labels are only capable of giving low-resolution predictions. The existing state-of-the-art label super resolution (LSR) method is capable of predicting high resolution labels, using only low-resolution supervision, given the joint distribution between low resolution and high resolution labels. However, it does not consider the inter-instance variance which is crucial in the ideal mathematical formulation. In this work, we propose a novel loss function modeling the inter-instance variance. We test our method on a real world application: infiltrating breast cancer region segmentation in histopathology slides. Experimental results show the effectiveness of our method.