Pan-Renal Cell Carcinoma classification and survival prediction from histopathology images using deep learning

Overview:

Histopathological images contain morphological markers of disease progression that have diagnostic and predictive values. In this study, we demonstrate how deep learning framework can be used for an automatic classification of Renal Cell Carcinoma (RCC) subtypes, and for identification of features that predict survival outcome from digital histopathological images. Convolutional neural networks (CNN’s) trained on whole-slide images distinguish clear cell and chromophobe RCC from normal tissue with a classification accuracy of 93.39% and 87.34%, respectively. Further, a CNN trained to distinguish clear cell, chromophobe and papillary RCC achieves a classification accuracy of 94.07%. Here, we introduced a novel support vector machine-based method that helped to break the multi-class classification task into multiple binary classification tasks which not only improved the performance of the model but also helped to deal with data imbalance. Finally, we extracted the morphological features from high probability tumor regions identified by the CNN to predict patient survival outcome of most common clear cell RCC. The generated risk index based on both tumor shape and nuclei features are significantly associated with patient survival outcome. These results highlight that deep learning can play a role in both cancer diagnosis and prognosis.

Click to view the published paper.

Related Publications: 

Tabibu, S., Vinod, P.K. and Jawahar, C.V., 2019. Pan-Renal Cell Carcinoma classification and survival prediction from histopathology images using deep learning. Scientific reports, 9(1), pp.1-9.


@article{tabibu2019pan,
title={Pan-Renal Cell Carcinoma classification and survival prediction from histopathology images using deep learning},
author={Tabibu, Sairam and Vinod, PK and Jawahar, CV},
journal={Scientific reports},
volume={9},
number={1},
pages={1--9},
year={2019},
publisher={Nature Publishing Group}
}

Dataset Download:

The whole slide images and clinical information were downloaded from TCGA data portal (https://gdc.cancer.gov/). Slides with reading and compatibility issues were removed (972 slides in total were removed from the whole dataset). We selected 1027 (KIRC), 303 (KIRP), and 254 (KICH) tumor slide images for our study. Further, corresponding 379, 47 and 83 normal tissue slide images for each subtype were selected. 512*512 sized tiles were extracted with 50% overlap ensuring multiple viewpoints within the tissue, at a magnification of 20x and 40x. For subtype classification, patches from 20x magnification were used. The patches were removed if the mean intensity of 50%-pixel values was larger than a threshold (in our case 210 for RGB channels).

Click to view dataset source.

Code:

The code we used for this  is added here