D Y Kim1, J W Park. 1. Department of Information and Communication Engineering, Chungnam National University, Republic of Korea. dykim@ns.kopec.co.kr
Abstract
PURPOSE: To implement a computer-aided detection system for kidney segmentation and kidney tumor detection on abdominal computed tomography (CT) scans. MATERIAL AND METHODS: Abdominal CT images were digitized with a film digitizer, and a gray-level threshold method was used to segment the kidney. Based on texture analysis performed on sample images of kidney tumors, a portion of the kidney tumor was selected as seed region for start point of the region-growing process. The average and standard deviations were used to detect the kidney tumor. Starting at the detected seed region, the region-growing method was used to segment the kidney tumor with intensity values used as an acceptance criterion for a homogeneous test. This test was performed to merge the neighboring region as kidney tumor boundary. These methods were applied on 156 transverse images of 12 cases of kidney tumors scanned using a G.E. Hispeed CT scanner and digitized with a Lumisys LS-40 film digitizer. RESULTS: The computer-aided detection system resulted in a kidney tumor detection sensitivity of 85% and no false-positive findings. CONCLUSION: This computer-aided detection scheme was useful for kidney tumor detection and gave the characteristics of detected kidney tumors.
PURPOSE: To implement a computer-aided detection system for kidney segmentation and kidney tumor detection on abdominal computed tomography (CT) scans. MATERIAL AND METHODS: Abdominal CT images were digitized with a film digitizer, and a gray-level threshold method was used to segment the kidney. Based on texture analysis performed on sample images of kidney tumors, a portion of the kidney tumor was selected as seed region for start point of the region-growing process. The average and standard deviations were used to detect the kidney tumor. Starting at the detected seed region, the region-growing method was used to segment the kidney tumor with intensity values used as an acceptance criterion for a homogeneous test. This test was performed to merge the neighboring region as kidney tumor boundary. These methods were applied on 156 transverse images of 12 cases of kidney tumors scanned using a G.E. Hispeed CT scanner and digitized with a Lumisys LS-40 film digitizer. RESULTS: The computer-aided detection system resulted in a kidney tumor detection sensitivity of 85% and no false-positive findings. CONCLUSION: This computer-aided detection scheme was useful for kidney tumor detection and gave the characteristics of detected kidney tumors.
Authors: Marius George Linguraru; Rabindra Gautam; James Peterson; Jianhua Yao; W Marston Linehan; Ronald M Summers Journal: Proc IEEE Int Symp Biomed Imaging Date: 2009-06
Authors: Jianfei Liu; Shijun Wang; Marius George Linguraru; Jianhua Yao; Ronald M Summers Journal: Med Image Anal Date: 2014-08-15 Impact factor: 8.545
Authors: Marius George Linguraru; Shijun Wang; Furhawn Shah; Rabindra Gautam; James Peterson; W Marston Linehan; Ronald M Summers Journal: Med Phys Date: 2011-10 Impact factor: 4.071
Authors: Marius George Linguraru; Shijun Wang; Furhawn Shah; Rabindra Gautam; James Peterson; W Linehan; Ronald M Summers Journal: Conf Proc IEEE Eng Med Biol Soc Date: 2009
Authors: Marius George Linguraru; Jianhua Yao; Rabindra Gautam; James Peterson; Zhixi Li; W Marston Linehan; Ronald M Summers Journal: Pattern Recognit Date: 2009-06-01 Impact factor: 7.740