TY  - JOUR
AU  - Serel, Tekin Ahmet
AU  - Ertunç, Onur
AU  - Kıran, Merve Meryem
AU  - Şengül, Esra
AU  - Hakkoymaz, Zeynep Ruken
AU  - Aydın, Eda Uysal
PY  - 2026
DA  - 2026/09/07
TI  - Predicting Tumor Mutational Burden in Prostate Cancer Using Deep Learning on Histopathological Images
JO  - OBM Genetics
SP  - 357
VL  - 10
IS  - 03
AB  - Tumor mutational burden (TMB) is a critical biomarker associated with the response to immunotherapy in prostate cancer. The heterogeneity of tumors can complicate the prediction of TMB, making reliable detection essential for effective treatment planning. Recent advancements in deep learning (DL) have facilitated the analysis of histopathological images, enabling better predictions of TMB. This study utilized the Cancer Genome Atlas (TCGA) cohort of prostate cancer patients, comprising 580 H&E-stained whole slide images (WSIs) and corresponding mutation data. We classified TMB into high (TMB-H) and low (TMB-L) groups based on a threshold of 0.9. Pre-trained deep learning models, specifically Inception V3 and VGG16, were employed to analyze the WSIs. The models were trained using a fivefold cross-validation approach, and various data preprocessing and augmentation techniques were applied to enhance model performance. The Inception V3 model demonstrated superior predictive performance, achieving a training accuracy of 0.77 and validation accuracy of 0.81, compared to the VGG16 model, which had a training accuracy of 0.68 and validation accuracy of 0.77. Both models effectively classified patients into TMB-H and TMB-L categories, with the Inception V3 model showing lower validation loss, indicating better generalization to unseen data. Our findings suggest that deep learning models, particularly Inception V3, can effectively predict TMB status in patients using histopathological images. This research highlights the potential for integrating DL techniques in clinical settings to enhance personalized treatment strategies for patients with varying TMB.
SN  - 2577-5790
UR  - https://doi.org/10.21926/obm.genet.2603357
DO  - 10.21926/obm.genet.2603357
ID  - Serel2026
ER  -