pandani.ai

Accurate molecular profiling and prognostication from routine histopathology slides could transform precision oncology. We are developing a ViT-based MIL framework for combined predictions of 32 solid tumour types, TP53 biomarker detection, and survival prediction directly from Whole Slide Images (WSIs).

Our model can simultaneously infer tumour taxonomy, TP53 mutation status, and gene expression directly from WSIs, with performance comparable to conventional genomic assays, while prognostic risk remains limited. This integrated, slide-level approach lays the groundwork for scalable computational pathology—opening a pathway to precision oncology that is accessible even in resource-constrained settings.

Schematic overview

Accurate quantification of the Ki-67 proliferation index is essential for breast cancer prognosis and treatment planning. Current automated methods, including classical and deep learning approaches based on cell detection or segmentation, often face challenges due to densely packed nuclei, morphological variability, and inter-laboratory differences. Since Hematoxylin and Eosin (H&E) staining is routinely performed, accurately estimating Ki-67 from these slides could save resources by eliminating the need for additional immunohistochemical staining.

Our approach precisely quantifies Ki-67 expression and automates hotspot detection directly from H&E-stained WSIs, providing a scalable tool for digital pathology workflows. The cross-modality model potentially quantifies molecular expression from morphological features using H&E-stained WSIs.

Read the paper

Digital pathology has significantly advanced cancer diagnosis by enabling high–resolution visualisation and assessment of tissue specimens. However, the manual analysis of these images remains labour–intensive and susceptible to human error, resulting in inconsistencies in diagnosis and treatment decisions.

We developed and externally validated a self–supervised, multi–resolution model for classifying the most common subtypes of skin cancer from whole slide images (WSIs). Our model demonstrates robust performance across diverse skin cancer subtypes, indicating a strong potential to assist anatomical pathologists in automatically detecting, highlighting, and classifying skin cancer subtypes directly from WSIs, and a subsequent capability to improve the effectiveness of clinical decision making. For non–melanoma cancer, our model can be deployed in regions with limited access to experienced dermatopathologists and a high incidence of the disease, particularly in low-resource settings.

Read the paper

Prostate cancer is the second most frequent carcinoma among men and a leading cause of morbidity and mortality. The gold standard for diagnosing and grading prostate cancer depends on a histopathological examination of prostate tissue biopsies, where the architectural pattern of the tissue and cellular morphology are assessed to assign a Gleason pattern score.

However, manual grading is labour–intensive, subjective, and susceptible to inter– and intra–observer variability, leading to diagnostic inconsistencies and compromised outcomes. These challenges highlight the need for developing more precise, objective, and reproducible cutting–edge tools in digital pathology.

Read the paper