Related Experiment Video
Updated: May 30, 2026
07:41
Performing Data Mining And Integrative Analysis Of Biomarker in Breast Cancer Using Multiple Publicly Accessible Databases
Published on: May 17, 2019
Computerized three-class classification of MRI-based prognostic markers for breast cancer
Neha Bhooshan1, Maryellen Giger, Darrin Edwards
1Department of Radiology, The University of Chicago, Chicago, IL 60637, USA. bhooshan@uchicago.edu
Physics in Medicine and Biology
|August 24, 2011
Summary
Computerized analysis using Bayesian artificial neural networks (BANNs) can help classify breast tumor grades from DCE-MRI scans. This artificial intelligence approach shows promise for improving prognostic classification of breast lesions.
Area of Science:
- Radiology and Medical Imaging
- Artificial Intelligence in Medicine
- Oncology
Background:
- Accurate breast tumor grading is crucial for prognostic classification and treatment planning.
- Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) provides valuable information for lesion characterization.
- Computer-Aided Diagnosis (CADx) systems can potentially enhance the accuracy and efficiency of tumor grading.
Purpose of the Study:
- To investigate the efficacy of a three-class Bayesian artificial neural network (BANN) for characterizing breast tumor grades (1, 2, and 3) using DCE-MRI.
- To evaluate the performance of BANNs in feature selection and classification for prognostic purposes.
- To explore the potential of DCE-MRI CADx in distinguishing tumor grades for prognostic classification.
Main Methods:
- A dataset of 26 grade 1, 86 grade 2, and 58 grade 3 invasive ductal carcinoma (IDC) lesions from DCE-MRI was analyzed.
- Automated lesion segmentation and extraction of kinetic and morphological features were performed.
- Three-class BANNs were employed for stepwise feature selection and classification with leave-one-lesion-out cross-validation.
- Receiver Operating Characteristic (ROC) analysis was used to assess classification performance.
Main Results:
- The BANN model achieved Area Under the Curve (AUC) values of 0.80 ± 0.05 for grade 1 vs. grade 3, 0.78 ± 0.05 for grade 1 vs. grade 2, and 0.62 ± 0.05 for grade 2 vs. grade 3.
- The system demonstrated effective feature selection and classification capabilities for distinguishing between different tumor grades.
- High accuracy was observed in differentiating between higher-grade tumors (grade 3) and lower-grade tumors (grade 1 and 2).
Conclusions:
- Three-class BANN feature selection and classification show potential for application in Computer-Aided Diagnosis (CADx) systems.
- DCE-MRI CADx can be expanded from diagnostic to prognostic classification by effectively distinguishing tumor grades.
- This AI-driven approach offers a promising tool for improving the prognostic assessment of breast cancer patients.
