Segmenting Images like MR Brain, Breast and Scintigraphy Thyroid Gland using Fuzzy C Means Based Morphological Reconstruction Filters
R. Sumathi1, M. Venkatesulu2
1R. Sumathi, Department of Computer Science and Engineering, Kalasalingam Academy of Research and Education, (Tamil Nadu), India.
2M. Venkatesulu, Department of Information Technology, Kalasalingam Academy of Research and Education, (Tamil Nadu), India.
Manuscript received on 11 December 2019 | Revised Manuscript received on 22 December 2019 | Manuscript Published on 30 December 2019 | PP: 1011-1016 | Volume-9 Issue-2S2 December 2019 | Retrieval Number: B11061292S219/2019©BEIESP | DOI: 10.35940/ijitee.B1106.1292S219
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Our study has introduced a new modified methodology using Fuzzy C Means clustering with morphological reconstruction filters to segment the abnormal parts in multimodal images such as MR brain, MR breast and scintigraphy thyroid gland.MR scanning is helpful to analyze the internal behavior of the tumor, whereas scintigraphy scanning is used to analyze the shape and location of the gland and also prevent the cancerous stage. We have used samples from public dataset like Harvard brain dataset for the brain, RIDER for breast and TCGA – THCA for the thyroid gland. In the first step, we preprocessed the image by applying the median filter which removes the noisy information present in the given input image. In the second step, Fuzzy C Means clustering was used to segment the boundary of the abnormal part in the multimodal images. In the last step, morphological reconstruction filters are used to segment the accurate shape and location of the abnormal part in all the three multimodal images. The performance and the efficiency of the segmentation were computed using the measures such as entropy, eccentricity, MSE, PSNR, sensitivity, specificity, accuracy and computational time. The results from our modified method show an accurate segmentation for all multimodalities images within 4ms and its accuracy rate is nearly 95% for all types of images when compared with existing techniques such as K-means and GA with K- Means. A new modified method using Fuzzy C means clustering with morphological reconstruction filters was applied to segment the abnormal part accurately with minimum duration in all multimodal images.
Keywords: Image Segmentation, Median Filter, Fuzzy C Means Clustering, Performance Measures.
Scope of the Article: Image Security