Deep learning–based 4D‐synthetic CTs from sparse‐view CBCTs for dose calculations in adaptive proton therapy

Abstract

Background: Time-resolved 4D cone beam-computed tomography (4D-CBCT) allows a daily assessment of patient anatomy and respiratory motion. However, 4D-CBCTs suffer from imaging artifacts that affect the CT number accuracy and prevent accurate proton dose calculations. Deep learning can be used to correct CT numbers and generate synthetic CTs (sCTs) that can enable CBCT-based proton dose calculations. Purpose: In this work, sparse view 4D-CBCTs were converted into 4D-sCT utilizing a deep convolutional neural network (DCNN). 4D-sCTs were evaluated in terms of image quality and dosimetric accuracy to determine if accurate proton dose calculations for adaptive proton therapy workflows of lung cancer patients are feasible. Methods: A dataset of 45 thoracic cancer patients was utilized to train and evaluate a DCNN to generate 4D-sCTs,based on sparse view 4D-CBCTs reconstructed from projections acquired with a 3D acquisition protocol.Mean absolute error (MAE) and mean error were used as metrics to evaluate the image quality of single phases and average 4D-sCTs against 4D-CTs acquired on the same day. The dosimetric accuracy was checked globally (gamma analysis) and locally for target volumes and organs-at-risk (OARs) (lung, heart, and esophagus). Furthermore, 4D-sCTs were also compared to 3D-sCTs. To evaluate CT number accuracy, proton radiography simulations in 4D-sCT and 4D-CTs were compared in terms of range errors. The clinical suitability of 4D-sCTs was demonstrated by performing a 4D dose reconstruction using patient specific treatment delivery log files and breathing signals.Results: 4D-sCTs resulted in average MAEs of 48.1 ± 6.5 HU (single phase) and 37.7 ± 6.2 HU (average). The global dosimetric evaluation showed gamma pass ratios of 92.3% ± 3.2% (single phase) and 94.4% ± 2.1% (average). The clinical target volume showed high agreement in D 98 between 4D-CT and 4D-sCT, with differences below 2.4% for all patients. Larger dose differences were observed in mean doses of OARs (up to 8.4%). The comparison with 3D-sCTs showed no substantial image quality and dosimetric differences for the 4D-sCT average. Individual 4D-sCT phases showed slightly lower dosimetric accuracy.

Improving CBCT image quality to the CT level using RegGAN in esophageal cancer adaptive radiotherapy

“…This difference can be attributed to the low number of projections available for a single phase of the 4D-CBCT. Thummerer A [ 25 ] verified the feasibility of deep-learning-based 4D sCTs from sparse-view CBCTs for dose calculations in adaptive proton therapy. In addition, MR-guided radiotherapy treatment planning utilizes the high soft tissue contrast of MRI to reduce uncertainty in delineation of the target and organs at risk.…”

ScatterNet for projection-based 4D cone-beam computed tomography intensity correction of lung cancer patients

“…Dose calculations on CBCTs, which are routinely acquired for patient positioning in many clinics, would avoid additional dose burden for the patient from repeat CT imaging. Most scatter correction studies focused on 3D images, yet there were a few exceptions also investigating 4D applications [5] , [6] , [12] , [13] , [14] . The use of 4DCBCTs and 4D dose calculations could address uncertainties introduced by respiratory motion [7] , [15] , [16] , [17] .…”

“…In their proton dose evaluation they showed accurate results for the network as well as for the DIR method, while the analytical method was worse. In a follow-up study, they trained separate networks with 4DCBCT and 4DCT pairs of axial, coronal, and sagittal slices of the 0% breathing phase [14] . Applying the result to all breathing phases, a complete scatter-corrected 4DCBCT could be generated.…”

Comparison and evaluation of different deep learning models of synthetic CT generation from CBCT for nasopharynx cancer adaptive proton therapy

“…Several evaluations have been conducted on the use of sCT in adaptive proton radiotherapy, with most of these studies utilizing deep convolutional neural network (DCNN) and cycleGAN to generate sCT 29–32 . In this study, we aim to assess the suitability of different deep learning models, including the newly proposed cGAN, Unet+cycleGAN, and the conventional Unet and cycleGAN models, for generating sCT in NPC.…”

“…28 Several evaluations have been conducted on the use of sCT in adaptive proton radiotherapy, with most of these studies utilizing deep convolutional neural network (DCNN) and cycleGAN to generate sCT. [29][30][31][32] In this study, we aim to assess the suitability of different deep learning models, including the newly proposed cGAN, Unet+cycleGAN, and the conventional Unet and cycleGAN models, for generating sCT in NPC. By analyzing their unique structures and characteristics,we can determine which model is more appropriate for sCT generation and holds potential for further clinical exploration in adaptive proton therapy.…”