University of Houston

Melanoma is a cancer of the cells that produce melanin, which is the pigment that colors skin. Melanoma is one of the most curable cancers if detected early. Screening for melanoma is performed by a physician who visually inspects the skin surface where a skin lesion (a mole) is present. A physician may use magnification and a focused light source to reveal surface detail. Physicians look for specific features that are markers for melanoma such as those indicated by the ABCD rule: Asymmetry, irregular Borders, specific Color features, and Diameter. However, studies that measure the success rate of physicians at detecting melanoma by visual inspection reveal that they are incorrect 1 out of 3 times. Therefore, there is motivation for developing an assistive 'smart' device that analyzes skin-lesion images and helps physicians improve their diagnostic success rate, especially with the detection of early forms of skin cancers. Such a device should be programmable, readily available when needed, self contained, and capable of providing immediate help on demand. Thus, we envision a handheld device running sophisticated algorithms for image processing and image analysis. One can easily imagine the application of these technologies to other forms of medical imaging devices, motivating the development of a general software solution.

Miniaturization and integration of electronic components have enabled a wide variety of handheld wireless consumer devices, such as those found in communication, audio, imaging, and video applications. The primary example device is the cell phone. We are interested in the application and specialization of these technologies to medical applications. In this dissertation, we will focus on an object-oriented framework and supporting tools for developing optimized medical image analysis software for handheld devices. We will use the framework and support tools to implement an automated skin-cancer detection system, which will run on a Texas Instrument's DaVinci based platform. DaVinci technology includes software, tools, and a family of embedded SOC's (system on chip) with heterogeneous processing elements specialized for image and video processing.