Description

Today, medical imaging devices are well developed and routinely used to detect and follow up diseases. However, images generated by these devices require interpretation by well-trained professionals; computerized analysis has not been widely adopted due to the lack of reliable automated software. In this project, I will develop new computer programs for analyzing skin images and computer tomography (CT) images that can be used for screening and detecting malignant melanoma and lung cancer. Collaborating with oncologists and other scientists, I will investigate a fast and accurate algorithm to detect human skin pixels from skin images. This is the prerequisite step for many automatic analyses of skin images. Measuring image features of digital dermoscopy, which uses polarized light to reveal structures deep under the skin, is important for construction of a computer-aided diagnostic system for melanoma. I will also carry out further studies on dermoscopic images and develop novel algorithms for lesion segmentation, shape analysis and colour analysis.

CT and bronchoscopy (a screening technique to examine the lungs by inserting a flexible tube in the major airways of the lungs) are effective tools for screening and detecting lung cancer. I plan a new project to develop computer programs for preparing operation maps for bronchoscopy using pre-operational CT scans. Furthermore, I will analyze CT images taken at different times to follow up lung cancer patients.

The expected significance of this research project is to advance knowledge of computing science in building reliable image analysis algorithms for medical applications. It will provide an excellent training arena for graduate students who are interested in studying these algorithms. Ultimately, the research will lead to building computer-aided diagnostic device to detect and follow up diseases. In addition, the techniques can be expanded for non-medical applications.

Principal Investigator

Funding Agency

 The Natural Sciences and Engineering Research Council of Canada

Funding Period