An Improved Model For Renal Function Measurements And The Development Of An Extensible Medical Image Processing Platform

Renal function measurements based on 3D dynamic contrast-enhanced MRI data is now one of the hot topics, for reasons that renal function measurements plays an important role in diagnosis of renal diseases and dynamic contrast-enhanced MRI is very cheap, effective, non-damage compared with other traditional methods such as CT or nuclear medicine.Glomerualar filtration rate is an important parameter in renal function measurements. The measurement of GFR based on DCE-MRI data is now in the experimental phase, thus there are many defects and inconvenience in modelling, segmentation and software tools.In this paper, several medical image processing platforms are put under observation. The images acquired by DCE-MRI are 4D images, that is,3D for space plus 1D for time. Many softwares can not import such 4D images correctly and lack the support for 4D image processing. Due to these needs and considering extensibility, a new medical image processing software platform with extensible plugin framework is implemented. This software can import, segment, display and draw time-density curve for 4D images. Other developers can also extend this platform easily via C++ interfaces.This paper compares several mathematical models for renal function measurements and finds that these models lack robustness, thus room for improvements exists. A new segmentation of artery compartments of kidney based on time-density curve is proposed and using this result, tradition 3-compartment model can be improved.The new model is applied to the DCE-MRI data provided by Tianjin Medical University and the comparison of the results and golden standard of GFR proves that the new model is of higher precision.