Fully automatic region of interest selection in glomerular filtration rate estimation from 99mTc-DTPA renogram

Kun Ju Lin, Jia Yann Huang, Yung Sheng Chen*

*Corresponding author for this work

Research output: Contribution to journalJournal Article peer-review

18 Scopus citations


Glomerular filtration rate (GFR) is a common accepted standard estimation of renal function. Gamma camera-based methods for estimating renal uptake of 99mTc-diethylenetriaminepentaacetic acid (DTPA) without blood or urine sampling have been widely used. Of these, the method introduced by Gates has been the most common method. Currently, most of gamma cameras are equipped with a commercial program for GFR determination, a semi-quantitative analysis by manually drawing region of interest (ROI) over each kidney. Then, the GFR value can be computed from the scintigraphic determination of 99mTc-DTPA uptake within the kidney automatically. Delineating the kidney area is difficult when applying a fixed threshold value. Moreover, hand-drawn ROIs are tedious, time consuming, and dependent highly on operator skill. Thus, we developed a fully automatic renal ROI estimation system based on the temporal changes in intensity counts, intensity-pair distribution image contrast enhancement method, adaptive thresholding, and morphological operations that can locate the kidney area and obtain the GFR value from a 99mTc-DTPA renogram. To evaluate the performance of the proposed approach, 30 clinical dynamic renograms were introduced. The fully automatic approach failed in one patient with very poor renal function. Four patients had a unilateral kidney, and the others had bilateral kidneys. The automatic contours from the remaining 54 kidneys were compared with the contours of manual drawing. The 54 kidneys were included for area error and boundary error analyses. There was high correlation between two physicians' manual contours and the contours obtained by our approach. For area error analysis, the mean true positive area overlap is 91%, the mean false negative is 13.4%, and the mean false positive is 9.3%. The boundary error is 1.6 pixels. The GFR calculated using this automatic computer-aided approach is reproducible and may be applied to help nuclear medicine physicians in clinical practice.

Original languageEnglish
Pages (from-to)1010-1023
Number of pages14
JournalJournal of Digital Imaging
Issue number6
StatePublished - 12 2011


  • Adaptive thresholding
  • Diethylenetriaminepentaacetic acid
  • Glomerular filtration rate
  • Image contrast enhancement
  • Morphology operation
  • Renogram


Dive into the research topics of 'Fully automatic region of interest selection in glomerular filtration rate estimation from 99mTc-DTPA renogram'. Together they form a unique fingerprint.

Cite this