Style Transfer–assisted Deep Learning Method for Kidney Segmentation at Multiphase MRI
We developed and validated a semisupervised style transfer-assisted deep learning method for automated segmentation of the kidneys using multiphase contrast-enhanced (MCE) MRI acquisitions. A cycle-consistent generative adversarial network (CycleGAN) successfully generated anatomically coregistered synthetic multiphase contrast-enhanced (MCE) MRI datasets from T2-weighted MRI acquisitions simulating the precontrast, corticomedullary, early nephrographic, and nephrographic acquisitions. A deep learning segmentation model (mask region–based convolutional neural network [Mask R-CNN]) trained on a synthetic MCE MRI dataset achieved mean Dice scores between 0.91 and 0.93 for kidney segmentation on the original MCE MRI acquisitions.
Guo, et al. Radiol Artif Intell 2023
![Figure 1](/sites/default/files/styles/coh_large/public/2024-01/guo-radiolai-2023-fig1.jpg?itok=geSsHi4D)
Deep Learning Kidney Segmentation with Very Limited Training Data Using a Cascaded Convolution Neural Network
This study investigated the feasibility of kidney segmentation using deep learning convolution neural network (CNN) models trained with MR images from only a few subjects. Our few-shot kidney segmentation approach using 3D augmentation achieved a good performance even using a single Unet. Furthermore, a cascaded network significantly improved the performance of segmentation and was superior to a single Unet in certain cases.
Guo, et al. PLoS One 2022
![Figure 1](/sites/default/files/styles/coh_large/public/2023-08/guo-plosone-2022_1.png?itok=noCz04-N)
![Figure 5](/sites/default/files/styles/coh_large/public/2023-08/guo-plosone-2022_2.png?itok=1PClvpR2)
Deep Learning-based Deformable Registration of Dynamic Contrast-Enhanced MR Images of the Kidney
Deformable registration of three-dimensional (3D) dynamic contrast-enhanced (DCE) MRI data improves estimation of kidney kinetic parameters. In this study, we proposed a deep learning approach with two steps: a convolutional neural network (CNN) based affine registration network, followed by a U-Net trained for deformable registration between two MRI images. The proposed registration method was applied successively across consecutive dynamic phases of the 3D DCE-MRI dataset to reduce motion effects in the different kidney compartments (i.e., cortex, medulla).
Huang, et al. Proc SPIE Int Soc Opt Eng 2022
![Figure 2](/sites/default/files/styles/coh_large/public/2023-08/huang-proc-2022_1.png?itok=-usxKIig)
![Figure 6](/sites/default/files/styles/coh_large/public/2023-08/huang-proc-2022_2.png?itok=u4L7jgko)