This work presents segmentation of image for the detection of renal cell carcinoma (RCC) using k-means clustering unsupervised algorithm based on the color features. We used color images of renal cell carcinoma for detecting the affected area. RCC is a kidney cancer. This work has used a database containing 36 pathological images to detect affected the area of renal cell carcinoma. Finding RCC affected areas with pathological way is a very time-consuming process, hence the proposed work helps in this aspect to recognize it more rapidly. The input images are not subjected to any kind of preprocessing techniques, hence the originality of these images are maintained. The process of segmentation is carried out in dual phase. Initially based on their color and spatial features the pixels were clustered, later specific numbers of regions were merged by clustered blocks. This leads to amplification of efficiency in computation; as a result, we can ignore the feature extraction for each pixel in the image segmentation of renal cell carcinoma. Since our concentration is on separation of affected region of RCC image from an overall tissue image, image segmentation based on k-mean clustering is used, which is the feasible effective solution for detecting an affected area of RCC. We have evaluated the proposed approach of its segmentation with factors like time complexity and iteration on affected area of renal cell carcinoma by taking it as a case study with an adequate number of the image in our database. The outcome of this process is promising as the computational time consumed is incredibly less to find the area of affected region. Even after comparing it with other methods using K-Mean Clustering image segmentation successfully segmented with success rate of 83.33%.
Volume 9 | 05-Special Issue
Pages: 144-149