HipoMap

architecture

 HipoMap propose a novel graphical representation map of whole-slide histopathological images for slide-based analysis. HipoMap generates a fixed size of image-type map that represents comprehensive patterns of interests in a WSI. Thus, HipoMap provides a flexible solution to apply any advanced machine learning methods, suchas CNN, for slide-based analysis. The main contributions of ourproposed model, HipoMap are it can create a graphical representation map as feature extraction of WSI,which can be easily applied for various slide-based WSI problems(e.g., survival analysis). ROI is not needed for traning pretrain model. HipoMap outperformed the current state-of-the-art methods in slide-based pathological image analysis

This study was published in Scientific reports:


Citation

@article{article,
         author = {Sai, Kosaraju and Lee, Jeongyeon and Lee, Hyun and Yang, Jung Wook and Kang, Mingon},
         year = {2022},
         month = {12},
         
         title = {Deep learning-based framework for slide-based histopathological image analysis.},
         volume = {1},
         journal = {Scientific reports},
         doi = {https://doi.org/10.1038/s41598-022-23166-0}
}