Non-Invasive Phosphorus-31 Magnetic Resonance Spectral Characterization of Breast Tissue Anomalies Using Pattern Recognition and Artificial Intelligence
Final rept. 30 Jul 2005-19 Jul 2006
ARKANSAS UNIV AT LITTLE ROCK
Pagination or Media Count:
It is highly desirable to develop a non-invasive and pattern recognition technique that can detect and reliably interpret images or spectral data from small volumes of the breast. Due to the pervasive nature of breast cancer in society today, and the consequent need of a highly accurate, early diagnostic tool, this is a very timely proposal that could have a significant impact on womens health. Patient ROtating Delivery of Excitation Off-resonance RODEO MRI data has been obtained from Dr. Diana Lindquist at the University of Arkansas for Medical Sciences. These patients, which flagged suspicious regions in breast tissue, have undergone needle biopsies from these suspect regions for pathological examination. With the patients permission, Dr. Lindquist obtained P-31 MR scans of the flagged suspect tissue and healthy tissue in the same session. Access to data from 6 patients were obtained and made available for analysis in this study. We proposed to use a combination of pattern recognition techniques, including Artificial Neural Networks ANN, to develop in vivo methods that use breast P-31 MR scans suspicious and nonsuspicious regions to diagnose potential malignant tissue. The MR scan data will be paired with the known biopsy results to create a supervised training set. Unfortunately two events occurred to prevent us from completing this study 1-3.
- Anatomy and Physiology
- Medicine and Medical Research
- Atomic and Molecular Physics and Spectroscopy