MV CBCT-Based Synthetic CT Generation Using a Deep Learning Method for Rectal Cancer Adaptive Radiotherapy

Abstract

Due to image quality limitations, online Megavoltage cone beam CT (MV CBCT), which represents real online patient anatomy, cannot be used to perform adaptive radiotherapy (ART). In this study, we used a deep learning method, the cycle-consistent adversarial network (CycleGAN), to improve the MV CBCT image quality and Hounsfield-unit (HU) accuracy for rectal cancer patients to make the generated synthetic CT (sCT) eligible for ART. Forty rectal cancer patients treated with the intensity modulated radiotherapy (IMRT) were involved in this study. The CT and MV CBCT images of 30 patients were used for model training, and the images of the remaining 10 patients were used for evaluation. Image quality, autosegmentation capability and dose calculation capability using the autoplanning technique of the generated sCT were evaluated. The mean absolute error (MAE) was reduced from 135.84 ± 41.59 HU for the CT and CBCT comparison to 52.99 ± 12.09 HU for the CT and sCT comparison. The structural similarity (SSIM) index for the CT and sCT comparison was 0.81 ± 0.03, which is a great improvement over the 0.44 ± 0.07 for the CT and CBCT comparison. The autosegmentation model performance on sCT for femoral heads was accurate and required almost no manual modification. For the CTV and bladder, although modification was needed for autocontouring, the Dice similarity coefficient (DSC) indices were high, at 0.93 and 0.94 for the CTV and bladder, respectively. For dose evaluation, the sCT-based plan has a much smaller dose deviation from the CT-based plan than that of the CBCT-based plan. The proposed method solved a key problem for rectal cancer ART realization based on MV CBCT. The generated sCT enables ART based on the actual patient anatomy at the treatment position.

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

“…Zhao et al. ( 18 ) used the modified CycleGAN to generate sCT from MV CBCT; the auto-segmentation and dose calculation based on sCT showed promising results. Liang et al.…”

Multimodal image translation via deep learning inference model trained in video domain

“…Their values for CBCT versus CT and synthetic CT versus CT are improved from 46.68 ± 9.25, 28.05 ± 1.21, 0.97 ± 0.0084, 0.92 ± 0.014 to 23.27 ± 5.53, 32.67 ± 1.98, 0.99 ± 0.0059, 0.97 ± 0.028, respectively. The previously published Cycle-GAN model [ 17 ] testing results are also presented in the last column for the detailed comparison. Significantly better image quality evaluation values are presented in synthetic CT images compared with CBCT images.…”

“…For a proof-of-concept demonstration, we collected rectum cancer patients’ CT and CBCT images as our datasets to evaluate the proposed method. A new CT-linac uRT-linac 506c designed by United Imaging Healthcare Co. Ltd, which integrated a diagnostic-quality 16-slices helical CT and a C-arm linac together, was used for data acquisition [ 17 ]. The helical CT can be used for simulation, and the electronic portal imaging detector (EPID) system was used for 3D MV CBCT acquisition.…”

“…To speed up the training convergence, we scaled the CT and CBCT image pixel values to the range of (-1, 1) according to the formula , where indicates the original image pixel values locating in the range of (-1000, 2048). We collected CT and CBCT images from 100 patients for model training [ 17 ], and the images from the other 10 new patients were used for model evaluation.…”

“…It was first applied to translate MR to CT images on head and neck sites [ 24 ], then applied on pelvic sites with more training data and incorporating gradient consistency loss [ 25 ]. It was also trained to learn translation functions from a source domain CBCT to a target domain CT with high accuracy and efficiency in radiotherapy related studies [ 17 ].…”