Tumor segmentation is a fundamental task in histopathological image analysis. Creating accurate pixel-wise annotations for such segmentation tasks in a fully-supervised training framework requires significant effort. To reduce the burden of manual annotation, we propose a novel weakly supervised segmentation framework based on sparse patch annotation, i.e., only small portions of patches in an image are labeled as ‘tumor’ or ‘normal’. The framework consists of a patch-wise segmentation model called PSeger, and an innovative semi-supervised algorithm. PSeger has two branches for patch classification and image classification, respectively. This two-branch structure enables the model to learn more general features and thus reduce the risk of overfitting when learning sparsely annotated data. We incorporate the idea of consistency learning and self-training into the semi-supervised training strategy to take advantage of the unlabeled images. Trained on the BCSS dataset with only 25% of the images labeled (five patches for each labeled image), our proposed method achieved competitive performance compared to the fully supervised pixel-wise segmentation models. Experiments demonstrate that the proposed solution has the potential to reduce the burden of labeling histopathological images.

Background: Glomerular lesion recognition is one of the most crucial steps in the diagnosis of kidney disease. Deep learning, which relies on large numbers of pathology images, assists pathologists to access glomerular lesions more efficiently, objectively and accurately. However, due to different pathological development of glomeruli, complicated lesion patterns, and limited resolution of pathology images, there is annotation noise in datasets, making the deep learning model under- or over-fit. Methods: In this paper, we propose a novel noisy label learning model for lesion recognition in glomerular datasets with annotation noise. The model integrates uncertainty-based noisy label discriminator, contrastive learning, and consistency regularization to achieve high signal-to-noise supervision, pathology feature extraction, and utilization of pathology images. Results: We constructed large-scale glomerular datasets from 870 kidney disease cases using different stainings including Periodic acid–Schiff (PAS), Masson Trichrome (MT) and Periodic Schiff-Methenamine (PASM). Intensive experiments demonstrated the superiority of the proposed model for glomerular lesion recognition compared to other methods, as 25% of the lesions had f ₁ - score above 85%, 43.75% had f ₁ - score above 80%, and 75% had f ₁ - score at or above 70%. Additionally, further experiments demonstrate the effectiveness of each module. Conclusions: The noisy label learning model proposed is able to recognize the most glomerular lesions, with the annotation noise discrimination and large amounts of pathology images utilization, laying the foundation for the development of computer-aided evaluation system for renal pathology.