Computer Vision for Predicting the Efficacy of Neoadjuvant Therapy in Breast Cancer.
Daria Sitnikova1, Alexey Fayzullin1, Fedor Chistov2
1Institute for Regenerative Medicine, Sechenov University, 8-2 Trubetskaya St., 119991 Moscow, Russia.
Cancers
|June 12, 2026
Summary
Computer vision can predict breast cancer response to neoadjuvant therapy (NAT) using histopathology images. This approach identifies subtle morphological patterns to aid in selecting patients for personalized cancer treatment.
Area of Science:
- Computational pathology
- Digital pathology
- Breast cancer research
Background:
- Neoadjuvant therapy (NAT) is crucial for breast cancer treatment, but patient response varies significantly.
- Predicting pathological complete response (pCR) is vital for optimizing patient selection for NAT.
- Current methods lack precision in predicting individual patient responses to NAT.
Purpose of the Study:
- To review and summarize the application of computer vision techniques for predicting breast cancer response to NAT.
- To identify key histopathological features associated with therapy response or resistance.
- To highlight the potential of artificial intelligence (AI) in personalized breast cancer therapy.
Main Methods:
- Systematic review of studies using computer vision and machine learning on whole-slide images (WSIs).
- Analysis of hematoxylin and eosin (H&E) and immunohistochemically stained WSIs.
- Focus on morphological features of tumor cells, stroma, and tumor-infiltrating lymphocytes (TILs).
Main Results:
- Computer vision models can detect subtle morphological patterns linked to pCR.
- Therapy resistance predictors include low tumor cell density, necrosis, specific stroma types, and vascularization.
- Therapy sensitivity predictors include high nuclear staining intensity, high tumor cell density, and lymphocyte infiltration.
- Multimodal data integration enhances predictive performance.
Conclusions:
- Computer vision offers a powerful tool for predicting breast cancer response to NAT from histopathology.
- AI can identify complex morphological features beyond human pathologist capabilities.
- Further development of AI, particularly interpretable and cross-modal solutions, can improve prediction accuracy and understanding of NAT response mechanisms.
