Related Experiment Video
Updated: Apr 6, 2026
06:08
A Cognitive Fusion-guided Prostate Biopsy Using Multiparametric Magnetic Resonance Imaging and Transrectal Ultrasound
Published on: March 21, 2025
2.0K
Automated prostate cancer detection via comprehensive multi-parametric magnetic resonance imaging texture feature
Farzad Khalvati1,2, Alexander Wong3, Masoom A Haider4,5
1Department of Medical Imaging, University of Toronto, Toronto, ON, Canada. farzad.khalvati@sri.utoronto.ca.
BMC Medical Imaging
|August 6, 2015
Summary
This study introduces advanced radiomics models using multi-parametric MRI for more accurate prostate cancer auto-detection. These new models significantly improve upon conventional methods, aiding in earlier diagnosis and better patient outcomes.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Prostate cancer is a leading cause of cancer death in North America.
- Early detection significantly improves patient survival rates.
- Current multi-parametric MRI (MP-MRI) auto-detection algorithms do not fully utilize available data.
Purpose of the Study:
- To develop a radiomics-based auto-detection method for prostate cancer using MP-MRI data.
- To enhance prostate cancer detection accuracy by leveraging comprehensive MP-MRI data.
Main Methods:
- Developed novel MP-MRI texture feature models incorporating T2-weighted MRI (T2w), diffusion-weighted imaging (DWI), computed high-b DWI (CHB-DWI), and correlated diffusion imaging (CDI).
- Calculated a comprehensive set of texture features and performed feature selection for each modality.
- Constructed optimized texture feature models by combining the best features from each modality.
Main Results:
- Evaluated model performance using leave-one-patient-out cross-validation with a support vector machine (SVM) classifier.
- Trained and tested models on 40,975 cancerous and healthy tissue samples from clinical MP-MRI datasets.
- The proposed MP-MRI texture feature models demonstrated superior cancer detection accuracy compared to conventional T2w+DWI models.
Conclusions:
- Developed comprehensive texture feature models for improved radiomics-driven prostate cancer detection via MP-MRI.
- Optimal texture feature models, utilizing extensive features and selection methods, significantly enhanced prostate cancer auto-detection accuracy over conventional approaches.
