کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
485216 | 703318 | 2016 | 6 صفحه PDF | دانلود رایگان |
Magnetic resonance imaging (MRI) provides detailed information about the soft tissues of the brain to identify tumors, cancer or any abnormal conditions in brain. Usually these images contain a significant amount of noise caused by operators or some external factors. De-noising is essential before detection of any abnormal areas from these images. There are several de-noising methods. In this paper, wavelet denoising techniques are considered for denoising the MRI image. The quality of the image is assessed using Peak Signal to Noise Ratio (PSNR), Squared Error Mean (SEM) and Absolute Mean Error (AME). These de-noised images are very useful for the segmentation process to easily extract the abnormal area from the image. Segmentation plays a crucial role for detection of any abnormal areas, tumors and irregularities in brain images. For this process two unsupervised algorithms are proposed for the detection of abnormal area from de-noised image. Both Expectation Maximization and K-means segmentation algorithims are used to identify the abnormal tissues in a given MRI. The performance of both methods are analyzed with suitable parameters like Entropy, Area and Perimeter. These parameters show the performance of the algorithms for the segmentation process and hence a suitable algorithm can be applied.
Journal: Procedia Computer Science - Volume 87, 2016, Pages 204–209