Automated measurement method and tool of glomerular filtration rate using triphasic helical computed tomography images.

Hong Li,Edward Janosko,Jianping Gong,Qing Cai,Ping Geng,Yan Kang
DOI: https://doi.org/10.1016/j.urology.2010.09.029
IF: 2.633
2011-01-01
Urology
Abstract:OBJECTIVES To introduce an automated method to perform computed tomography-based glomerular filtration rate (CT GFR) measurement using the 2-point Patlak plot technique and triphasic helical CT images. We also evaluated the correlation between our automated method and the manual measurement results, as well as the results from cystatin C. METHODS The present study included 25 patients without an acute renal disorder. The CT scan protocol consisted of an unenhanced CT examination followed by 2 contrast-enhanced CT examinations in the arterial and parenchymal phase. Between the noncontrast scan and the arterial scan, 7 dynamic scans were obtained to provide more data on the arterial input function. The 2-point Patlak plot technique was used in the automated and manual measurement methods. For each patient, the cystatin C level, determined in blood samples, was used as a reference. RESULTS The correlation between our automated method and the cystatin C method in 25 patients was 0.8029 (P < .001). The correlation between the manually implemented CT GFR method and the cystatin C method was 0.8287 (P < .001). A strong correlation (r = .9518, P < .001) was seen between the automated and manual measurements using the same model; however, the automated process can be finished within 2 minutes. CONCLUSIONS The automated CT GFR measurement method could potentially be used because of its highly improved efficiency. Moreover, it would avoid the use of a differential renal nuclear study. This would be helpful for predicting the residual GFR after nephrectomy and could also be used to predict the residual GFR after partial nephrectomy for tumor or stone treatment. UROLOGY 77: 1259-1264, 2011. (C) 2011 Elsevier Inc.
What problem does this paper attempt to address?