Project

Computer Aided Diagnosis (CAD)

IM-AIM (IMage and Artificial Intelligence in Medicine) is software for cancer diagnosis. Originally, it is developed for Nasopharyngeal Carcinoma (NPC) diagnosis and treatment. This cancer is one of head and neck cancers. It is frequently found in Southeast Asia, particularly South China, Hong Kong, and Taiwan. In Thailand, Nasopharyngeal Carcinoma is the 5th most common cancer in male (2010). The primary tumor normally originates in the nasopharynx which is inaccessible. Therefore, NPC is difficult to detect in the early stage.

Medical images take an important role in NPC diagnosis and treatment from which useful tumor information, such as tumor size, location, and extent of invasion can be obtained. Moreover, 3D model of the tumor can be constructed from the images. In Thailand, CT scan is widely used because it is commonly available and cheaper than MR imaging. However, differentiating between the tumor and normal soft tissues in a CT image can be difficult. Identifying the tumor region in each CT slice is a task which demands experience and expertise of radiologists and otolaryngologists

IM-AIM allows physicians to retrieve medical records of patients, delineate tumor regions from medical images by three options, i.e., manual, semi-automatic, and automatic segmentation. 3D model of the tumor can be constructed and displayed on the virtual skull. Physicians thus can precisely determine the tumor location and plan their treatments. In addition, IM-AIM can predict risk of NPC recurrence for individual patient with the accuracy above 70%. Follow-up time can be appropriately scheduled for each patient. Videoconferencing is another feature of IM-AIM that supports tele-consulting between a physician and an expert.

IM-AIM has been extended to support bone tumor diagnosis from MR images. Recently, it has been tested and verified by radiologists and otolaryngologists at Ramathibodi hospital, Thailand.