RSNA: Computer-aided diagnosis (CADx) inches toward practice
Developed at the University of Chicago, the multimodality workstation prototype could better integrate image-based biomarkers in breast cancer diagnosis, prognosis and therapeutic assessment. The workstation displays mammography, ultrasound and MRI images and incorporates an array of quantitative datasets to facilitate the image interpretation process.
During a diagnostic work-up, the radiologist can access reader studies that show the probability of malignancy, similar cases and an online atlas of benign and malignant cases by clicking on a lesion on a mammography or ultrasound image. The workstation also provides multi-parametric breast MRI analysis including dynamic contrast-enhanced MRI and T2-weighted images and MRI lesion color maps based on kinetics and morphology.
The system can deliver a number of benefits, said Maryellen L. Giger, PhD, professor of radiology at the University of Chicago. Not only can the workstation help radiologists better understand tumors, it also provides assistance with complex and quantitative elements of the interpretative process, which may, in turn, boost efficiency and promote diagnostic confidence.
The multi-modality prototype workstation employs MR image-based prognostic biomarkers to help breast imagers characterize lesions by assessing invasion, lymph node metastasis and tumor grade.
Finally, an MR image-based predictive biomarker helps readers differentiate between lesions that will respond to therapy by decreasing in size and those that will not. “Radiologists can look at the probability of a good response,” explained Giger.
The University of Chicago researchers are currently discussing licensing with a few companies and considering the next steps to begin the FDA approval process and transition the project to the clinical space, shared Giger.