مقايسه مرزبندي تصاوير درماتوسكوپي از ضايعه ملانوما با فيلتر پلاريزه و بدون آن به كمك شبيه سازي

Comparison of Dermatoscopic Image Boundary of Melanoma by Simulated Polarized Filter

Background/ Objective:Melanoma is one of the common types of skin cancer the diagnosis of which is difficult for most dermatologists. To observe and diagnose the type of skin cancer, the non-invasive dermatoscopy is generally used. As this method depends on visual observation and evaluation, the diagnosing of melanoma, particularly in its early stage, is rather complicated and difficult. The diagnosis can be made easier if the boundary between the lesion and the surrounding non-affected area can be determined accurately. This article presents a method which can determine the boundary accurately and therefore facilitate diagnosis of the melanoma. Materials & Methods: We have shown, using simulation work, the importance of using polarized filter for determining the boundary between the lesion and surrounding area. The simulated polarized filter, with an appropriate selection of polarizer angle, is applied to the dermatoscopic image. This technique provides a new approach to preprocessing of images of cancerous lesions by increasing the contrast between the lesion and shadows. The boundary detection algorithm can then be used on the best preprocessed image to define the boundary. The resulting image can then be used by the dermatologist to diagnose possible melanoma. Results: Before using the boundary detection algorithm, applying the simulated polarized filter to the image causes reduction of noise considerably. One of the sources of noise in the image is the presence of hair. The use of polarized filter will greatly decrease interfering noise due to the presence of hair. Another source of noise is the inclusion of surrounding area within the boundary while leaving out part of the lesion under study. By appropriate selection of polarizer angle one can define the boundary more accurately. Conclusion: The simulated polarized filter provides an efficient preprocessing tool for determining the boundary between the lesions and surrounding tissue very accurately. This is very important when melanoma is being examined by the dermatologist. In this article, we have shown that using the above technique can assist the dermatologist to diagnose melanoma more accurately.