A Comparison Study Between CNN-Based Deformed Planning CT and CycleGAN-Based Synthetic CT Methods for Improving iCBCT Image Quality

Abstract

PurposeThe aim of this study is to compare two methods for improving the image quality of the Varian Halcyon cone-beam CT (iCBCT) system through the deformed planning CT (dpCT) based on the convolutional neural network (CNN) and the synthetic CT (sCT) generation based on the cycle-consistent generative adversarial network (CycleGAN).MethodsA total of 190 paired pelvic CT and iCBCT image datasets were included in the study, out of which 150 were used for model training and the remaining 40 were used for model testing. For the registration network, we proposed a 3D multi-stage registration network (MSnet) to deform planning CT images to agree with iCBCT images, and the contours from CT images were propagated to the corresponding iCBCT images through a deformation matrix. The overlap between the deformed contours (dpCT) and the fixed contours (iCBCT) was calculated for purposes of evaluating the registration accuracy. For the sCT generation, we trained the 2D CycleGAN using the deformation-registered CT-iCBCT slicers and generated the sCT with corresponding iCBCT image data. Then, on sCT images, physicians re-delineated the contours that were compared with contours of manually delineated iCBCT images. The organs for contour comparison included the bladder, spinal cord, femoral head left, femoral head right, and bone marrow. The dice similarity coefficient (DSC) was used to evaluate the accuracy of registration and the accuracy of sCT generation.ResultsThe DSC values of the registration and sCT generation were found to be 0.769 and 0.884 for the bladder (p < 0.05), 0.765 and 0.850 for the spinal cord (p < 0.05), 0.918 and 0.923 for the femoral head left (p > 0.05), 0.916 and 0.921 for the femoral head right (p > 0.05), and 0.878 and 0.916 for the bone marrow (p < 0.05), respectively. When the bladder volume difference in planning CT and iCBCT scans was more than double, the accuracy of sCT generation was significantly better than that of registration (DSC of bladder: 0.859 vs. 0.596, p < 0.05).ConclusionThe registration and sCT generation could both improve the iCBCT image quality effectively, and the sCT generation could achieve higher accuracy when the difference in planning CT and iCBCT was large.

Dosimetric comparison of deformable image registration and synthetic CT generation based on CBCT images for organs at risk in cervical cancer radiotherapy

“…The dpCT images were deformed from the pCT images to the CBCT images, and the sCT images were generated based on CBCT images, both of which had similar anatomical structures to the CBCT images, especially the high geometric similarity of the skin and bony structures. As we found in our previous work [27], the DIR method appeared to be ineffective for large deformable structures composed of soft tissue, and sCT might produce some arti cial structures inconsistent with the CBCT images, which caused the slight difference in the anatomical structures of the dpCT images and sCT images. In Fig.…”

“…The methods underlying the development of the DIR network and sCT generation network are described in detail in our previous work [27]. Brie y, the DIR network is based on a 3D multistage registration network (MSnet), which includes three stages of registration, each of which consists of two downsampling…”

“…In our previous study [27], the accuracy of the image quality and the structural consistency with CBCT images were compared between the DIR method and sCT generation method. The results demonstrated that both DIR and sCT generation could effectively improve the image quality of CBCT images.…”

Improving synthetic CT accuracy by combining the benefits of multiple normalized preprocesses

“…Various research groups have described linear normalization 11,[18][19][20][21][22][23] and nonlinear normalization preprocessing methods using the hyperbolic tangent function (Tanh), 24,25 where Tanh(x) = e x −e −x e x +e −x . With reference to the above literature, 11,[18][19][20][21][22][23][24][25] we defined two linear normalization and one nonlinear normalization preprocessing methods employed in this work as follows:…”

A novel approach for eliminating metal artifacts based on MVCBCT and CycleGAN

“…A trained GA-B could generate image A’, which has the structure of image A and the style of image B. CycleGAN models ( 24 ) include two generators and two discriminators and add cycle-consistency loss for training. CycleGAN has been widely used for interconversion between different types of medical images ( 25 – 30 ). The structure of CycleGAN used in this study is consistent with that reported in the literature ( 24 ), and the model structure is shown in Figure 1 .…”