Related Experiment Video
Updated: Jan 7, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
741
Generating crossmodal gene expression from cancer histopathology improves multimodal AI predictions
Samiran Dey1, Christopher R S Banerji2,3, Partha Basuchowdhuri1
1School of Mathematical & Computational Sciences, Indian Association for the Cultivation of Science, Kolkata, India.
Nature Communications
|December 31, 2025
Summary
Artificial intelligence can now predict cancer grading and survival risk using only digital pathology images. This novel approach synthesizes gene expression data, offering a practical alternative to costly transcriptomic tests in healthcare.
Area of Science:
- Oncology
- Computational Biology
- Artificial Intelligence
Background:
- Multimodal artificial intelligence (AI) integrating digital pathology and transcriptomic data shows promise for cancer diagnosis and prognosis.
- Clinical implementation is hindered as transcriptomic data is rarely available in routine healthcare settings, with histopathology remaining the standard.
- Synthesizing transcriptomic data from digital pathology offers a potential solution for practical AI-driven cancer analysis.
Purpose of the Study:
- To develop and validate an AI model that synthesizes transcriptomic data from digital histopathology images for cancer grading and survival risk prediction.
- To assess the accuracy and reliability of predictions made using synthesized transcriptomic data compared to real transcriptomic data.
- To demonstrate the clinical feasibility of AI-based multimodal fusion without requiring actual transcriptomic measurements.
Main Methods:
- Utilized two public multimodal cancer datasets (TCGA, CPTAC) across four cohorts: glioma-glioblastoma, renal, uterine, and breast.
- Developed PathGen, a diffusion-based crossmodal generative AI model, to synthesize gene expression data from whole slide images (WSIs).
- Evaluated the model's performance in predicting cancer grading and patient survival risk, ensuring certainty and interpretability.
Main Results:
- Incorporating synthesized transcriptomic data with WSIs significantly improved cancer grading and risk estimation (p < 0.05).
- Predictions using synthesized features were statistically comparable to those using real transcriptomic data (p > 0.05) across all cohorts.
- PathGen achieved state-of-the-art performance in jointly predicting cancer grading and survival risk with high accuracy and certainty.
Conclusions:
- AI-driven synthesis of gene expression from digital pathology is a viable and accurate method for cancer grading and survival risk prediction.
- PathGen offers a practical, cost-effective approach to multimodal AI in oncology, overcoming the limitations of transcriptomic data acquisition.
- The model provides interpretable predictions and certainty guarantees, paving the way for clinical integration.
Related Concept Videos
02:43Mouse Models of Cancer Study
6.3K
Mice have long served as models for studying human biology and pathology because of their phylogenetic and physiological similarity with humans. They are also easy to maintain and breed in the laboratory, and hence, many inbred strains are now available for research. Studies on mice have contributed immeasurably to our understanding of cancer biology.
The development of transgenic, knockout, and knock-in mice has led to an exponential increase in their use as model organisms in research,...
The development of transgenic, knockout, and knock-in mice has led to an exponential increase in their use as model organisms in research,...
6.3K
02:53Adaptive Mechanisms in Cancer Cells
6.9K
Cancer cells accumulate genetic changes at an abnormally rapid rate due to the defects in the DNA repair mechanisms. From an evolutionary perspective, such genetic instability is advantageous for cancer development. Mutant cell lines accumulate a series of beneficial mutations that contribute to their progression into cancer.
Some of the advantages that cancer cells have on normal cells include - enhanced ability to divide without terminally differentiating, induce new blood vessel formation,...
Some of the advantages that cancer cells have on normal cells include - enhanced ability to divide without terminally differentiating, induce new blood vessel formation,...
6.9K