The field of digital pathology is rapidly advancing with the development of innovative deep learning techniques. Recent studies have focused on improving the accuracy and reproducibility of breast cancer classification, melanoma diagnosis, and histology image analysis. A key direction in this field is the integration of low-resolution and high-resolution features to enhance prognosis and diagnosis. Another area of research is the development of multi-stage auto-context deep learning frameworks for tissue and nuclei segmentation and classification.

Noteworthy papers in this area include: Contrasting Low and High-Resolution Features for HER2 Scoring using Deep Learning, which introduces the India Pathology Breast Cancer Dataset and develops predictive models for HER2 3-way classification. PixelCAM: Pixel Class Activation Mapping for Histology Image Classification and ROI Localization, which proposes a multi-task approach for weakly supervised object localization and introduces PixelCAM, a cost-effective foreground/background pixel-wise classifier. SCFANet: Style Distribution Constraint Feature Alignment Network For Pathological Staining Translation, which proposes a novel network for direct translation of H&E stained images into IHC stained images and achieves precise transformation of H&E-stained images into their IHC-stained counterparts. CellVTA: Enhancing Vision Foundation Models for Accurate Cell Segmentation and Classification, which improves the performance of vision foundation models for cell instance segmentation by incorporating a CNN-based adapter module. GECKO: Gigapixel Vision-Concept Contrastive Pretraining in Histopathology, which proposes a pretraining approach that aligns WSIs with a Concept Prior derived from the available WSIs and delivers clinically meaningful interpretability.