The Computer-Aided Diagnosis and Imaging division has been a leader in the development of computer-aided detection systems for both lung and breast cancer using Chest Radiography and mammography since 1992. We have worked with our industrial partner (Deus Technology LLC now Riverain Medical) to develop their computer-aided system for lung cancer on chest radiographs. The ISIS Center performed the pivotal clinical trial for FDA Pre-Market Approval (PMA). Our group is currently developing improved methods for the detection of lung nodules on both chest radiographs and chest computed tomography (CT) including temporal subtraction of chest radiographs for improving cancer detection. We expect to see a significant increase in our efforts toward computer-aided diagnosis research and plan to extend this work to other chest diseases.
We have a long-standing collaborative research program with Dr. Heang-Ping Chan at the University of Michigan and recently completed a prospective clinical trial of their system for the computer-aided detection of breast cancer with Dr. Chan. The results were published in Radiology (April 2004). We have recently received an NIH/NCI grant to develop a computer-aided diagnosis system using a mammographic library with an intelligent agent to search for clinically similar cases which can assist the radiologists in making breast cancer diagnosis.
We also developed a unified framework that integrates all predictive transform and decomposition methods including wavelet and sub-band transforms. Currently, we are developing methods to switch wavelet kernel when encountering different image patterns in a medical image (e.g., brain tissues, bones, and bone edges in a CT head image).
Having been supported by an NIH/NIBIB grant, we are working with Imperium, Inc., to develop a speckle-free transmission ultrasound imaging system based on a newly invented PE-CMOS ultrasound sensor for medical applications. The imagery generated is striking for its radiographic appearance and its lack of speckle. This state-of-the-art PE-CMOS sensor can generate projection ultrasound images with resolution as high as 250-350 microns with a dynamic range up to 70 dB when a planar transducer is operated at a 5 MHz to 7.5 MHz range.
Our group also has joint training and education programs with the Electrical Engineering and Computer Science programs of Howard University Medical Center , the Catholic University of America , and Virginia Polytechnic Institute and State University .
Division Director: Shih-Chung Benedict Lo, PhD