AUTHOR=Luo Chenhua , Yang Jiyan , Liu Zhengzheng , Jing Di 

TITLE=Predicting the recurrence and overall survival of patients with glioma based on histopathological images using deep learning

JOURNAL=Frontiers in Neurology

VOLUME=Volume 14 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2023.1100933

DOI=10.3389/fneur.2023.1100933

ISSN=1664-2295

ABSTRACT=Background: To predict the recurrence and overall survival of glioma patients based on the representative biopsy tissue, the deep learning (DL) model can be applied to optimize personalized medicine.
Methods: Our study retrospectively collected 162 glioma patients and randomly divided them into a training cohort (n=113) and a validation cohort (n=49) to build DL model. The HE-stained slide was segmented into a size of 180×180 without overlapping. The patch-level features were extracted by the pre-trained ResNet50 to predict the recurrence and overall survival. Additionally, a “light-strategy” was introduced where low-size digital biopsy images with clinical information were inputted into the DL model to ensure minimum memory occupation. 
Results: Our study extracted 512 histopathological features from the HE-stained slides of each glioma patient. 36 and 18 features were identified as significantly related to DFS and OS (P<0.05) using the univariate Cox proportional-hazards model. Pathomics signature showed a C-index of 0.630 and 0.652 for DFS and OS prediction, respectively. The time-dependent Receiver Operating Characteristic (ROC) curves along with nomograms were used to assess the diagnostic accuracy at a fixed time point. In the validation cohort (n=49), AUC in the 1-year and 2-year DFS were 0.955 and 0.904, respectively, and the 2-year, 3-year, and 5-year OS were 0.969, 0.955, and 0.960, respectively. We stratified the patients into low- and high-risk groups using the median hazard score (0.083 for DFS, and -0.177 for OS) and showed significant differences between these groups (P<0.001).
Conclusion: Our results demonstrated that the DL model based on the HE-stained slides showed the predictability of recurrence and survival in glioma patients. The results can be used to assist oncologists in selecting the optimal treatment strategy in clinical practice.