Deep Learning-Based Early Diagnosis of Prostate Cancer Using Deep Neural Networks
N Ananthasagar, P. K. Krishnan Namboori, P Aarshageetha · 2025
Prostate cancer remains a leading cause of cancer-related death, and therefore the need for efficient and timely diagnostic methods is crucial. This paper proposes a system using deep neural networks (DNN) that combines histopathological images and genomic data obtained from MAP2K4-related genes in an attempt to enable improved early detection. We carried out rigorous preprocessing procedures, such as normalization and standardization, to improve the data integrity using a sizable dataset with 1,166 histopathological images and 497 cases of The Cancer Genome Atlas (TCGA). The protein-protein interactions (PPI) network identified 20 genes associated with prostate cancer development. The proposed DNN model was trained using these multimodal inputs. The model achieved classification accuracy of 96.85% and mean squared error (MSE) of 0.5386, showing high efficiency in distinguishing malignant features on holdout data, with 10-fold cross-validation confirming robustness (85.12%). Our research demonstrated the synergistic potential of the convergence of genetic and histopathological image data in the direction of precision oncology, which led to the development of a scalable system with clinical applications.