Nuclear Segmentation and Classification: On Color & Compression Generalization
This addresses a critical issue for clinical pathology applications where algorithms must handle real-world variations, though it is incremental as it builds on existing models and benchmarks.
The paper tackled the problem of poor generalization of nuclear segmentation and classification models to out-of-distribution data, particularly color shifts, by evaluating top models from the CoNIC challenge, finding that they are robust to compression but suffer significant performance drops with color variations, and that neural style transfer improves test performance more consistently than stain normalization.
Since the introduction of digital and computational pathology as a field, one of the major problems in the clinical application of algorithms has been the struggle to generalize well to examples outside the distribution of the training data. Existing work to address this in both pathology and natural images has focused almost exclusively on classification tasks. We explore and evaluate the robustness of the 7 best performing nuclear segmentation and classification models from the largest computational pathology challenge for this problem to date, the CoNIC challenge. We demonstrate that existing state-of-the-art (SoTA) models are robust towards compression artifacts but suffer substantial performance reduction when subjected to shifts in the color domain. We find that using stain normalization to address the domain shift problem can be detrimental to the model performance. On the other hand, neural style transfer is more consistent in improving test performance when presented with large color variations in the wild.