Compensation cycle consistent generative adversarial networks (Comp‐GAN) for synthetic CT generation from MR scans with truncated anatomy

Abstract

Background MR scans used in radiotherapy can be partially truncated due to the limited field of view (FOV), affecting dose calculation accuracy in MR‐based radiation treatment planning. Purpose We proposed a novel Compensation‐cycleGAN (Comp‐cycleGAN) by modifying the cycle‐consistent generative adversarial network (cycleGAN), to simultaneously create synthetic CT (sCT) images and compensate the missing anatomy from the truncated MR images. Methods Computed tomography (CT) and T1 MR images with complete anatomy of 79 head‐and‐neck patients were used for this study. The original MR images were manually cropped 10–25 mm off at the posterior head to simulate clinically truncated MR images. Fifteen patients were randomly chosen for testing and the rest of the patients were used for model training and validation. Both the truncated and original MR images were used in the Comp‐cycleGAN training stage, which enables the model to compensate for the missing anatomy by learning the relationship between the truncation and known structures. After the model was trained, sCT images with complete anatomy can be generated by feeding only the truncated MR images into the model. In addition, the external body contours acquired from the CT images with full anatomy could be an optional input for the proposed method to leverage the additional information of the actual body shape for each test patient. The mean absolute error (MAE) of Hounsfield units (HU), peak signal‐to‐noise ratio (PSNR), and structural similarity index (SSIM) were calculated between sCT and real CT images to quantify the overall sCT performance. To further evaluate the shape accuracy, we generated the external body contours for sCT and original MR images with full anatomy. The Dice similarity coefficient (DSC) and mean surface distance (MSD) were calculated between the body contours of sCT and original MR images for the truncation region to assess the anatomy compensation accuracy. Results The average MAE, PSNR, and SSIM calculated over test patients were 93.1 HU/91.3 HU, 26.5 dB/27.4 dB, and 0.94/0.94 for the proposed Comp‐cycleGAN models trained without/with body‐contour information, respectively. These results were comparable with those obtained from the cycleGAN model which is trained and tested on full‐anatomy MR images, indicating the high quality of the sCT generated from truncated MR images by the proposed method. Within the truncated region, the mean DSC and MSD were 0.85/0.89 and 1.3/0.7 mm for the proposed Comp‐cycleGAN models trained without/with body contour information, demonstrating good performance in compensating the truncated anatomy. Conclusions We developed a novel Comp‐cycleGAN model that can effectively create sCT with complete anatomy compensation from truncated MR images, which could potentially benefit the MRI‐based treatment planning.

Channel-wise attention enhanced and structural similarity constrained cycleGAN for effective synthetic CT generation from head and neck MRI images

“…The cycleGAN requires the fake image to keep all the information in the original image, as a result, its CT correction ability was reduced in some degree. Compared with previous studies, the results of Zhao et al [ 6 ] showed the average MAE, PSNR, and SSIM calculated over test patients were 91.3 HU, 27.4 dB, and 0.94 for the proposed Comp-cycleGAN models trained with body-contour information. However, our cycleSimulationGAN model scored better on the three quantitative metrics with MAE (60.88), PSNR (36.23), and SSIM (0.985).…”

“…Magnetic resonance imaging (MRI) has become an essential imaging modality in both staging and targets volume (TV) delineation for head and neck (H&N) cancer radiotherapy (RT) owing to its intrinsically superior soft-tissue contrast, functional information, high resolution and non-radiation [ 1 – 3 ]. Precise contouring is crucial when treating H&N cancer patients presenting with primary soft tissue invasion or subtle intra-cranial invasion for accurate dose delivery to TV, and consequently improve the treatment outcomes [ 4 – 6 ]. Several studies have proven that MRI was more accurate than Computer Tomography (CT) and Positron Emission Tomography (PET) and was even closer to the pathological specimen measurements considered the “gold standard”, which can also substantially decrease the inter-observer variability of TV and organs at risks (OARs) delineation [ 1 , 2 , 7 ].…”

vOARiability: Interobserver and intermodality variability analysis in OAR contouring from head and neck CT and MR images

“…SO) does not considerably affect the contouring performance. Although we have identified considerable contouring differences for specific OARs, we can conclude that almost all OARs can be contoured with a similar degree of variability in either the CT or MR modality, which provides favorable support for MR images from the perspective of MR-only [38][39][40] and MR-guided RT. 42,43…”

“…However, because of the often insufficient CT image contrast for soft tissues, several studies recommended the integration of complementary magnetic resonance (MR) modality. 4,33 This is important especially from the perspective of OAR contouring, 4,34,35 synthetic MR image generation for MR-aided RT, 36,37 synthetic CT image generation for MR-only RT [38][39][40][41] and MR-guided RT. 42,43 While some OARs can be accurately and reliably contoured in CT images (i.e., bone structures such as, e.g., the mandible), MR images are often used to better visualize soft tissues.…”

Fully-automated, CT-only GTV contouring for palliative head and neck radiotherapy

“…Synthetic Magnetic Resonance (sMR) images (Approach 5) were generated by a pre-trained, in-house developed Comp-GAN 22 . To improve the structural consistency between the sMR and input CT images, a structure-consistency loss was introduced in the cycleGAN model 22 , 23 .…”