Related Experiment Video
Updated: Jun 4, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
Prostate cancer characterization on MR images using fractal features
1Inserm, U703, Université Nord de France, 152 rue du Docteur Yersin, 59120 Loos, CHRU Lille, France.
Medical Physics
|March 3, 2011
Summary
This study introduces a novel method for detecting prostate cancer using fractal and multifractal analysis of MRI scans. The advanced technique offers improved accuracy and robustness compared to traditional texture analysis methods.
Area of Science:
- Medical Imaging
- Computational Pathology
- Radiomics
Background:
- Prostate cancer detection relies heavily on accurate image analysis.
- T2-weighted MRI is a key modality for visualizing prostate anatomy and pathology.
- Texture analysis of medical images can reveal subtle patterns indicative of disease.
Purpose of the Study:
- To develop and evaluate a computerized method for detecting prostate cancer on T2-weighted MRI.
- To assess the efficacy of combining fractal and multifractal features for image textural analysis.
Main Methods:
- Fractal and multifractal features were extracted for textural analysis of T2-weighted MRI.
- Fractal dimension was computed using the Variance method.
- Multifractal spectrum estimation employed an adaptation of a multifractional Brownian motion model.
- Nonlinear supervised classification using Support Vector Machine (SVM) and AdaBoost algorithms was performed.
Main Results:
- Experiments were conducted on MRI images from 17 patients, with histological data serving as ground truth.
- The SVM classifier achieved 83% sensitivity and 91% specificity.
- The AdaBoost classifier demonstrated 85% sensitivity and 93% specificity.
Conclusions:
- The proposed model combining fractal and multifractal features outperformed classical texture analysis methods (Haralick, wavelet, Gabor).
- The method exhibited enhanced robustness against variations in signal intensity.
- The approach, initially applied to T2 images, holds potential for extension to multispectral MRI analysis.
