Michael Ronald Chinander, Presenter: Nothing to Disclose

Yonglin Pu MD, PhD, Abstract Co-Author: Nothing to Disclose

Maryellen L. Giger PhD, Abstract Co-Author: Stockholder, Hologic, Inc Royalties, Hologic, Inc Royalties, General Electric Company Royalties, MEDIAN Technologies Royalties, Riverain Medical Royalties, Mitsubishi Corporation Royalties, Toshiba Corporation

Mohan S. Gundeti MD, Abstract Co-Author: Nothing to Disclose

Alexandre Rosen MD, Abstract Co-Author: Nothing to Disclose

Bill C. Penney PhD, Abstract Co-Author: Nothing to Disclose

Currently, the clinical interpretation of renal scintigraphy typically only includes analysis of the whole kidney. Thus we aim to develop and assess a new automated segmentation method involving both the whole kidney and the renal cortex and renal pelvis on renal scintigraphy scans. Ultimately, this would be the initial stage of an automated method to compute radiotracer renal transit times to be used to aid in the diagnosis of obstructive hydronephrosis.  

Thirty Tc99m MAG3 renal scintigraphy scans were retrospectively collected under an IRB approved protocol. The scans consisted of 60 1-second acquisitions from the first minute after injection of the radiotracer, followed by 15-second acquisitions for the remainder of the exam; 45-55 minutes. Segmentation of the kidneys and their cortex and pelvis, was performed using a fuzzy c-means (FCM) algorithm with six clusters. All images from the scans were used as input to the FCM. The segmentation method was assessed by comparing the whole kidney segmentation with outlines drawn by a radiologist. A total of 55 kidneys had radiologist drawn outlines in our database. The overlap ratio was computed as intersection of the FCM segmentation and radiologist outline divided by their union. The component segmentation of the renal pelvis was visually assessed as being acceptable or not.  

Performance of the automatic segmentation yielded 82 percent of the segmented kidneys with a high overlap ratio of 0.4 or higher when compared to the radiologist outlines. Qualitatively, 76 percent had acceptable segmentation of the renal pelvis component.  

The results of this study show that automated segmentation can be conducted of both the whote kidney as well as the renal cortex and renal pelvis components from renal scintigraphy scans.  

Automated segmentation of the whole kidney as well as the renal cortex and renal pelvis from renal scintigraphy scan has the potential to be used to compute radiotracer renal transit times.  

Chinander, M, Pu, Y, Giger, M, Gundeti, M, Rosen, A, Penney, B, Investigation of Fuzzy C-Means Segmentation in the Delineation of the Kidney, Pelvis, and Cortex Components on Renal Scintigraphy Scans  .  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11034617.html