Related Experiment Video
Updated: Jan 11, 2026
Quantitative Multispectral Analysis Following Fluorescent Tissue Transplant for Visualization of Cell Origins, Types, and Interactions
Published on: September 22, 2013
PMI estimation with cross-species transfer learning and visual information generated by pathomics foundation model
Guoshuai An1,2, Shuwei Jing1,2, Zihe Cheng1,2
1School of Forensic Medicine, Shanxi Medical University, No. 98, University Street, Yuci District, Jinzhong, 030600, Shanxi, China.
Background:
Estimating the postmortem interval (PMI) is a key task in forensic science. Deep learning-based pathology image analysis offers a promising approach, but existing pathomics methods face two major challenges: limited translatability from animal to human samples and insufficient model interpretability.
Methods:
We propose a PMI estimation framework based on a pathomics foundation model with a two-stage cross-species transfer learning strategy. In the first stage, the model is fine-tuned on porcine liver whole-slide images (WSIs); in the second, it is further fine-tuned with a small amount of human data to achieve effective knowledge transfer. To improve interpretability, model predictions are visualized at the whole-slide level using probability maps, class maps, and classification proportion histograms.
Results:
Sixteen porcine and twenty-three human samples were used to evaluate four deep learning models-ResNet50, DenseNet121, SongCi, and UNI-for PMI estimation. The Vision Transformer-based UNI model achieved the best performance, with 91.63% accuracy in porcine data. After transfer learning with limited human samples, accuracy increased to 78.95%, representing a more than 50% improvement compared to the untuned model. The visualization framework further enhanced interpretability and traceability of the model's outputs.
Conclusion:
This study demonstrates that combining animal data priors with a fine-tuning strategy using minimal human data and whole-slide visualization enables cross-species PMI estimation. The proposed framework addresses data scarcity, enhances model transparency, and provides a practical and interpretable AI-based tool for forensic pathology.
Insights
This study introduces a novel AI framework for estimating postmortem interval (PMI) using cross-species transfer learning. The method enhances accuracy and interpretability in forensic pathology by combining animal data with limited human samples.
Area of Science:
- Forensic Science
- Artificial Intelligence
- Pathology
Background:
- Estimating postmortem interval (PMI) is crucial in forensic science.
- Deep learning in pathology image analysis shows promise for PMI estimation.
- Existing methods struggle with animal-to-human sample translation and model interpretability.
Purpose of the Study:
- To develop a robust postmortem interval estimation framework.
- To address challenges in cross-species data translation and model interpretability in forensic pathology.
- To create an interpretable AI tool for forensic analysis.
Main Methods:
- A pathomics foundation model with a two-stage cross-species transfer learning strategy was proposed.
- The model was fine-tuned on porcine liver whole-slide images (WSIs) followed by human data.
- Visualization techniques including probability maps and classification proportion histograms were used for interpretability.
Main Results:
- The Vision Transformer-based UNI model achieved 91.63% accuracy on porcine data.
- Transfer learning with limited human data improved accuracy to 78.95%, a >50% increase.
- The visualization framework enhanced model interpretability and traceability.
Conclusions:
- Combining animal data priors with fine-tuning on minimal human data enables cross-species PMI estimation.
- The framework effectively addresses data scarcity and enhances AI model transparency in forensic pathology.
- A practical and interpretable AI-based tool for forensic pathology was developed.
