Related Experiment Video
Updated: Oct 23, 2025
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
Wide & Deep neural network model for patch aggregation in CNN-based prostate cancer detection systems
L Duran-Lopez1, Juan P Dominguez-Morales1, D Gutierrez-Galan1
1Robotics and Tech. of Computers Lab., Universidad de Sevilla, 41012, Seville, Spain; Escuela Técnica Superior de Ingeniería Informática (ETSII), Universidad de Sevilla, 41012, Seville, Spain; Escuela Politécnica Superior (EPS), Universidad de Sevilla, 41011, Seville, Spain; Smart Computer Systems Research and Engineering Lab (SCORE), Research Institute of Computer Engineering (I3US), Universidad de Sevilla, 41012, Seville, Spain.
This study introduces a new AI method for prostate cancer diagnosis using digital pathology images. The system accurately classifies whole slides by aggregating patch-level predictions, aiding pathologists and speeding up screening.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Prostate cancer (PCa) is a leading cause of cancer death in men globally.
- Digital pathology and AI, particularly Convolutional Neural Networks (CNNs), are transforming cancer diagnosis.
- Analyzing gigapixel whole-slide images (WSIs) for PCa diagnosis presents computational challenges due to their size.
Purpose of the Study:
- To develop a novel patch aggregation method for slide-level prostate cancer classification.
- To leverage CNN-derived patch-level predictions for improved diagnostic accuracy.
- To create an AI tool that assists pathologists in screening prostate tissue samples.
Main Methods:
- Digitized prostate tissue WSIs were processed into smaller patches for CNN analysis.
- A custom Wide & Deep neural network was employed for patch aggregation and slide-level classification.
- Features such as malignant tissue ratio and probability histograms were utilized for classification.
Main Results:
- The proposed patch aggregation method achieved high accuracy (94.24%) and sensitivity (98.87%).
- The system effectively performs slide-level classification using aggregated patch data.
- The model demonstrated robust performance in identifying malignant tissues.
Conclusions:
- The developed AI system shows significant potential to aid pathologists in prostate cancer diagnosis.
- This approach can accelerate the screening process for prostate cancer, improving efficiency.
- The findings support the integration of advanced AI techniques in digital histopathology workflows.
