GANeXt: A Fully ConvNeXt-Enhanced Generative Adversarial Network for MRI- and CBCT-to-CT Synthesis
Siyuan Mei, Yan Xia, Fuxin Fan
TL;DR
This work tackles synthetic CT generation from MRI and CBCT to support adaptive radiotherapy. It introduces GANeXt, a fully ConvNeXt-based GAN with a GeNeXt 3D generator, a conditional PatchGAN discriminator, and a multi-head SegPatchGAN, trained with a multi-term loss across MRI-to-CT and CBCT-to-CT tasks. The approach demonstrates state-of-the-art performance on the SynthRAD2025 dataset, with strong generalization to unseen data and clear advantages over both convolutional and transformer-based baselines. Limitations include underestimation of high-HU regions and reliance on accurate deformable registration, suggesting areas for future improvement and broader clinical deployment.
Abstract
The synthesis of computed tomography (CT) from magnetic resonance imaging (MRI) and cone-beam CT (CBCT) plays a critical role in clinical treatment planning by enabling accurate anatomical representation in adaptive radiotherapy. In this work, we propose GANeXt, a 3D patch-based, fully ConvNeXt-powered generative adversarial network for unified CT synthesis across different modalities and anatomical regions. Specifically, GANeXt employs an efficient U-shaped generator constructed from stacked 3D ConvNeXt blocks with compact convolution kernels, while the discriminator adopts a conditional PatchGAN. To improve synthesis quality, we incorporate a combination of loss functions, including mean absolute error (MAE), perceptual loss, segmentation-based masked MAE, and adversarial loss and a combination of Dice loss and cross-entropy for multi-head segmentation discriminator. For both tasks, training is performed with a batch size of 8 using two separate AdamW optimizers for the generator and discriminator, each equipped with a warmup and cosine decay scheduler, with learning rates of $5\times10^{-4}$ and $1\times10^{-3}$, respectively. Data preprocessing includes deformable registration, foreground cropping, percentile normalization for the input modality, and linear normalization of the CT to the range $[-1024, 1000]$. Data augmentation involves random zooming within $(0.8, 1.3)$ (for MRI-to-CT only), fixed-size cropping to $32\times160\times192$ for MRI-to-CT and $32\times128\times128$ for CBCT-to-CT, and random flipping. During inference, we apply a sliding-window approach with $0.8$ overlap and average folding to reconstruct the full-size sCT, followed by inversion of the CT normalization. After joint training on all regions without any fine-tuning, the final models are selected at the end of 3000 epochs for MRI-to-CT and 1000 epochs for CBCT-to-CT using the full training dataset.