Related Experiment Video
Updated: Jun 26, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Automatic classification for pathological prostate images based on fractal analysis.
Po-Whei Huang1, Cheng-Hsiung Lee
1Department of Computer Science and Engineering,National Chung Hsing University, Taichung 40227, Taiwan. powhei.huang@msa.hinet.net
IEEE Transactions on Medical Imaging
|January 24, 2009
Summary
This study introduces a computer-aided system for automatic Gleason grading of prostate cancer images. Fractal dimension features achieve high accuracy, improving objective pathological image analysis for better prognosis and treatment.
Area of Science:
- Digital pathology
- Medical image analysis
- Computational oncology
Background:
- Accurate grading of prostatic carcinoma is crucial for patient prognosis and treatment planning.
- Manual histological grading of prostate tissues using the Gleason grading system is subjective and time-consuming.
Purpose of the Study:
- To develop an automated computer-aided system for grading pathological prostate images using the Gleason grading system.
- To evaluate novel fractal dimension-based feature extraction methods for analyzing image intensity and texture complexity.
Main Methods:
- Proposed two feature extraction methods utilizing fractal dimension to quantify variations in intensity and texture.
- Employed Bayesian, k-Nearest Neighbors (k-NN), and Support Vector Machine (SVM) classifiers for image classification.
- Utilized leave-one-out and k-fold cross-validation for performance estimation, achieving Correct Classification Rates (CCR).
Main Results:
- Achieved high CCRs: 91.2% (Bayesian), 93.7% (k-NN), and 93.7% (SVM) on 205 pathological prostate images.
- Optimizing fractal-based features with sequential floating forward selection improved CCRs to 94.6% (Bayesian), 94.2% (k-NN), and 94.6% (SVM).
- The proposed fractal-based feature set demonstrated superior performance compared to multiwavelets, Gabor filters, and GLCM, with a smaller feature size and strong discriminating capability.
Conclusions:
- The developed computer-aided system effectively automates Gleason grading of prostatic carcinoma.
- Fractal dimension-based features offer a robust and efficient method for analyzing pathological prostate images.
- The automated system has the potential to enhance objectivity and efficiency in prostate cancer grading, aiding clinical decision-making.
