Convolutional Neural Network-Based Automatic Classification of Colorectal and Prostate Tumor Biopsies Using...

Area of Science:

• Medical Imaging
• Computational Pathology
• Artificial Intelligence in Oncology

Background:

• Colorectal and prostate cancers are leading male cancers globally.
• Manual histological analysis of biopsy samples is time-consuming and prone to observer variability.
• Current diagnostic methods for these cancers can impact reliability and efficiency.

Purpose of the Study:

• To develop an automated computerized system for colorectal and prostate tumor diagnosis.
• To enhance diagnostic accuracy and reduce the time associated with manual pathological analysis.
• To leverage deep learning for improved cancer detection from biopsy images.

Main Methods:

• A novel convolutional neural network (CNN) architecture was proposed.
• The CNN model was designed for classifying colorectal and prostate tumors using multispectral biopsy images.
• Key modifications included removing the last convolutional block and halving filters per layer.

Main Results:

• The proposed CNN achieved high accuracy: 99.8% for prostate and 99.5% for colorectal datasets.
• The system outperformed pre-trained CNNs and other classification methods.
• It eliminated the need for preprocessing and utilized a single CNN model for the entire task.

Conclusions:

• The developed CNN architecture demonstrated superior performance in classifying colorectal and prostate tumor images.
• The system offers a more efficient and reliable alternative to manual pathological review.
• The proposed CNN architecture is computationally efficient and requires no image preprocessing, making it ideal for clinical application.