Buruli ulcer (BU) is a devastating flesh-eating bacterial infection that, each year, affects thousands of people, mostly children under the age of 15 years. If detected early BU can be easily treated and cured. In the last few years, automated dermoscopy systems have shown promise for accurate detection of skin lesions in the early stages. The objective of this proposal is to develop software tools to analyze suspicious skin lesions to automatically detect BU at an early stage. We propose to 1) collect a well-defined set of BU images using surface illumination with light of different frequencies, 2) develop algorithms for accurate lesion segmentation 3) identify critical features that correlate with surface and subsurface lesion characteristics, such as morphology, color, texture, and subsurface tissue hardening and thickening, 4) develop new classifiers based on implicit mapping with kernels, 5) test the performance of the new algorithms against manual image segmentation and lesion labeling by two domain experts independently. In addition, the proposed project addresses technological and socioeconomic issues, namely (a) the feasibility of deploying sophisticated diagnostic tools on general-purpose devices, such as cell phones which have limited computing power and memory, and employ fixed-point arithmetic, and (b) whether the use of non-descript technology, such as ordinary camera cell phones protects patient privacy, and thus it helps individuals to overcome the social stigma associated with BU so that that they will seek treatment. The proposed project is an important step toward the development of automated tools for identifying key structural features associated with BU. As of yet, no system accomplishes this, and no system has incorporated deep tissue characteristics to identify these structures. If successful, this software will be available as a diagnostic adjunct for deployment on a broad scale. The use of general-purpose devices like cell phones will help improve not only access to healthcare, but also the quality of life of patients, while also minimizing the cost of health care delivery.
Buruli Ulcer (BU) is a devastating and potentially debilitating tropical disease that affects people in more than 30 countries. BU is listed on the WHO list of neglected disease and is treatable and preventable. The development and application of low-cost, ubiquitous, non-invasive technologies that can be used privately by the end user of the local health care provider is critical for (1) increasing access to healthcare and (2) addressing the stigma associated with hospital visits. The success of the project can lower the cost of treatment for this disease and also improve the quality of life of patients.
| Zouridakis, George; Wadhawan, Tarun; Situ, Ning et al. (2015) Melanoma and other skin lesion detection using smart handheld devices. Methods Mol Biol 1256:459-96 |
| Hu, Rui; Queen, Courtney M; Zouridakis, George (2012) Lesion border detection in Buruli ulcer images. Conf Proc IEEE Eng Med Biol Soc 2012:5380-3 |
| Wadhawan, Tarun; Situ, Ning; Rui, Hu et al. (2011) Implementation of the 7-point checklist for melanoma detection on smart handheld devices. Conf Proc IEEE Eng Med Biol Soc 2011:3180-3 |
| Wadhawan, Tarun; Situ, Ning; Lancaster, Keith et al. (2011) SkinScan©: A PORTABLE LIBRARY FOR MELANOMA DETECTION ON HANDHELD DEVICES. Proc IEEE Int Symp Biomed Imaging 2011:133-136 |
| Situ, Ning; Yuan, Xiaojing; Zouridakis, George (2011) Assisting Main Task Learning by Heterogeneous Auxiliary Tasks with Applications to Skin Cancer Screening. J Mach Learn Res 15:688-697 |
| Situ, Ning; Wadhawan, Tarun; Hu, Rui et al. (2011) EVALUATING SAMPLING STRATEGIES OF DERMOSCOPIC INTEREST POINTS. Proc IEEE Int Symp Biomed Imaging 2011:109-112 |
