Measurement of Kidney Volume from Contrast-Enhanced CT (computed tomography) Images

Abstract

An assessment of KV is essential for the determination of the quality of donated kidneys especially obtained from living donors. Unfortunately, KV-measurements based on conventional imaging techniques, such as USG had proved to be less than satisfactory. Moreover, such an estimation of KVs can be a tedious and time-consuming process. In this study, a cohort of 12 donors (seven males and five females) were included for a process of KV-estimation where additional information based on CT images (coronal and axial), MRI (e.g., renal volume), and USG (e.g., renal pole-to-pole length) were also provided. All the data were already existing and were accessed as a part of an ongoing project. Renal length measured by USG and on coronal CT images were compared, while the KVs estimated by MRI and CT images were also compared. The MorphoLibJ plugin inbuilt within the FIJI image analysis software was used as a semi-automated platform for the estimation of KVs with comparison to a reference method. There was a significant difference between renal length measured by USG and on coronal images (P value 0.0237 with a bias of -0.732). The renal length showed a strong correlation with the KV derived from CT (r = 0.632) and stronger correlation with KV provided from MRI images (r = 0.804). No statistical difference was seen between the KV measured on CT and MRI images (P value 0.0825 with a bias of 14.1). The DSC between the reference and the semi-automated method on axial and coronal plains to measure KV was 0.927 and 0.934, respectively. In conclusion, the renal length is not recommended to estimate KV in the presence of a 3D data set such as MRI and CT images. Finally, MorphoLibJ is a promising analytical tool for semi-automated KV-estimation with decent accuracy and speed. This thesis highlights its implications and challenges while realising the goal of achieving a fully automated platform for the measurement of KVs.