Abstract
Prostate cancer (PCa) is a common malignancy in men worldwide. Current diagnosis is limited by nonspecific serum markers, biopsy sampling errors, and subjective Gleason grading, highlighting the demand for objective and quantitative assessment approaches. Second harmonic generation (SHG) imaging enables label‐free characterization of collagen architecture remodeling in prostate tissue. We acquired SHG images and developed an automated framework to extract 142 collagen morphological and textural features. Using least absolute shrinkage and selection operator (LASSO) regression for feature selection and regularization, a prostate collagen index (PCI) was established. The PCI showed preliminary discrimination capability between benign prostatic hyperplasia (BPH) and PCa, with an exploratory region of Interest (ROI)‐level area under the curve (AUC) of 0.9286 and a primary case‐level AUC of 0.9688. These findings provide preliminary evidence supporting the feasibility of a LASSO‐driven, SHG‐based quantitative framework for integrating multidimensional collagen features and performing exploratory case‐level discrimination between BPH and PCa.
Funding Information
  • National Natural Science Foundation of China (62505110)
  • Chinese Academy of Medical Sciences (2025‐JKCS‐08)
  • Xiamen Municipal Bureau of Science and Technology (3502Z20224ZD1008)

This publication has 28 references indexed in Scilit: