A Novel Approach of Intuitive K-means Clustering for Renal Calculi Detection in Ultrasound Images

• Published in: International Journal on Electrical Engineering and Informatics Vol. 10; no. 1; pp. 126 - 139
• Main Authors: Upadhyay, Pawan Kumar, Sharma, Arun, Chandra, Satish
• Format: Journal Article
• Language: English
• Published: Bandung School of Electrical Engineering and Informatics, Bandung Institute of Techonolgy, Indonesia 01.03.2018
• Subjects: Calculi; Cluster analysis; Clustering; Computing time; Elongation; Image detection; Image management; Image segmentation; Medical imaging; Retrieval; Signal to noise ratio; Target recognition; Ultrasonic imaging; Ultrasonic methods; Ultrasonic testing; Urology; Vector quantization; India
• ISSN: 2085-6830; 2087-5886
• DOI: 10.15676/ijeei.2018.10.1.9

Medical images are too fuzzy for discrete boundaries. This paper describes a fuzzy rule based seed point optimization in K-mean clustering method with a application in image segmentation. Prior information about the subject helps to elongate the cluster class and able to identify the target seed point smoothly for the detection of renal calculi oftenly called as a kidney stone. Kidney is a source organ for urology disorder which can be protected by efficient kidney stone detection technique in ultrasound images. Proposed method of clustering reduces the number of iterations for elobrating the region of interest in entitled images. This approach promising to give a more accurate solution for ultrasound images and it also enhances the image retrieval as compared to classical clustering methods. The experimental results justify the effectiveness of proposed approach by reducing the computational time without effecting the segmentation quality which can be validated by peak signal to noise ratio value. Results are validated on 150 Ultrasound image samples having six classes of renal calculi.