Optimal input configuration of dynamic contrast enhanced MRI in convolutional neural networks for liver segmentation

摘要

Most MRI liver segmentation methods use a structural 3D scan as input, such as a T1 or T2 weighted scan. Segmentationperformance may be improved by utilizing both structural and functional information, as containedin dynamic contrast enhanced (DCE) MR series. Dynamic information can be incorporated in a segmentationmethod based on convolutional neural networks in a number of ways. In this study, the optimal inputconfiguration of DCE MR images for convolutional neural networks (CNNs) is studied.The performance of three different input configurations for CNNs is studied for a liver segmentation task.The three configurations are I) one phase image of the DCE-MR series as input image; II) the separate phasesof the DCE-MR as input images; and III) the separate phases of the DCE-MR as channels of one input image.The three input configurations are fed into a dilated fully convolutional network and into a small U-net. TheCNNs were trained using 19 annotated DCE-MR series and tested on another 19 annotated DCE-MR series.The performance of the three input configurations for both networks is evaluated against manual annotations.The results show that both neural networks perform better when the separate phases of the DCE-MR seriesare used as channels of an input image in comparison to one phase as input image or the separate phases asinput images. No significant difference between the performances of the two network architectures was found forthe separate phases as channels of an input image.