Research Projects

Examples of Research Projects in CAD Lab

Recently, our CAD lab conducted a number of research projects or studies, which include use of the quantitative image features computed from

Mammograms to predict short-term breast cancer risk,
Breast MRI to predict patients’ response to neoadjuvant chemotherapy,
Lung CT images to predict risk of cancer recurrence of lung cancer patients after surgery,
Abdominal CT images to predict progression-free survival of ovarian cancer patients undergoing clinical trials for testing the new chemotherapy drugs,
Abdominal CT images to predict peritoneal metastasis in gastric cancer patients,
Abdominal CT images to predict the risk of metastasis of gastric cancer patients,
Brain CT or MRI images to predict prognosis of aneurysmal subarachnoid hemorrhage patients or detect the residual tumor after brain tumor surgery.

Lung nodule and disease detection and assessment of disease severity

An integrated CAD scheme and graphic user interface (GUI) tool of lung CT images has been developed to detect and segment lung nodules and other types of lung disease regions. The goal is to extract and compute new quantitative image markers.

Prediction of short-term cancer risk based on bilateral asymmetry of breast density

A CAD scheme along with a GUI tool has been developed to detect and quantify bilateral breast density and tissue patterns, and then use these features to develop a new quantitative imaging marker to predict short-term breast cacner risk.

Automated segmentation of aneurysmal subarachnoid hemorrhage regions

An interactive CAD scheme and GUI tool has been developed to segment ASH regions and quantity disease severity. The tool has been installed in two neurology departments of two medical centers to support clinical researchers of neurologists.

Classification of breast lesions using contrast-enhanced digital mammograms

A CAD scheme and GUI tool has been developed to segment breast lesions and detect image features from a new CEDM imaging modality to develop machine learning models to more accurately classify breast lesions.