Improving Clinical Diagnosis Through Change Detection in Mammography
Abstract
Temporal change of mass lesions overtime is a key piece of information in computer-aided diagnosis of breast!- cancer and treatment monitoring. For a specific patient, change detection is a critical step to select lesion candidates for follow-up diagnosis performed by either clinicians or computers. The purpose of the project is to develop an automatic change detection method to quantitatively extract the clinically important changes of suspicious lesions, upgrade the existing CAD system, and thus improve the clinical diagnosis of breast cancer. "1 particular, we have developed (1) PAR/mPAR/MLP/TPS based hybrid registration software to align sequential mammograms involving non-rigid deformation; (2) site model based change detection scheme to detect new lesions and/or select lesion candidates for further computer analysis; (3) feature extraction algorithm to obtain discriminative imagery features of true masses against mass-like normal tissues; and (4) neural network based decision support system(s) for mass detection. Image registration algorithm effectively recovers non-rigid deformation between mammograms. Site model based change detection scheme automatically detects the subtle changes and prioritizes lesion candidates with significant changes. The combined change-triggered and appearance-triggered initial lesion candidate selection increases the sensitivity of existing CAD system. The neural network based classifiers further improve the specificity of mass detection. The performance was initially 0.78-0.80 for the areas Az under the ROC curves using the conventional neural network, and later improved to Az values of 0.84-0.89 when using the newly developed multiple circular path neural networks.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2003
- Accession Number
- ADA416617
Entities
People
- Yue-joseph Wang
Organizations
- The Catholic University of America