K-Means Clustering and Dense–Sparse–Dense Optimized LSTM for Prostate Cancer

چکیده

Prostate cancer is one of the most common cancers affecting men worldwide. Accurate segmentation of prostate images plays a crucial role in diagnosis and treatment planning. This study presents an optimized segmentation method that combines the k-means clustering algorithm with the elbow method to automatically determine the optimal number of clusters. The approach was applied to two datasets: near-infrared (NIR) prostate images obtained using PSMA-targeted photodynamic therapy (PDT) agents, and histological prostate cancer images. Experimental results show that the method consistently identifies four as the optimal number of clusters, producing clearer and more reliable segmentation compared to conventional methods. Statistical analysis of clustered pixels confirmed the effectiveness of the proposed technique in highlighting tumor regions. The method’s simplicity and accuracy make it a useful tool to support radiologists in prostate cancer analysis and diagnosis, although further improvements are needed to address variations in illumination, contrast, and outliers.