Volumetric Cine Imaging for On-board Target Localization in Radiation Therapy

Accurate target localization is critical for liver and lung cancers due to uncertainties caused by respiratory motions. On-board four-dimensional (4D) or real-time verification of the tumor before and during Stereotactic Body Radiation Therapy (SBRT) is necessary because SBRT uses high fractional doses, tight Planning Target Volume (PTV) margins and a long treatment time. Current imaging of moving targets for on-board localization cannot image full volumetric information in real-time. The purposes of this dissertation research are to do the following. (1) Develop a real-time quasi-cine CBCT reconstruction method for on-board CBCT-guided target verification; (2) Develop a volumetric cine MRI (VC-MRI) technique for on-board MRI-guided target verification using an MRI-guided radiotherapy machine; (3) Develop an on-board 4D MRI technique for on-board MRI-guided target verification using kV projections from a conventional linear accelerator (LINAC); (4) Accelerate VC-MRI through undersampling acquisition while maintaining image quality; and (5) Improve geometric accuracy of VC-MRI through novel undersampling acquisition and deformation models.

A technique for 4D CBCT estimation was previously developed using a deformation field map (DFM)-based strategy. In the previous method, each phase of the 4D CBCT was generated by deforming a prior CT volume acquired during simulation. The DFM was solved by a global motion model (GMM) extracted by a global Principal Component Analysis (PCA) from prior 4D-CT and free-form deformation (FD) technique, using a data fidelity constraint. However, this technique has limitations in both reconstruction time (~5 minutes) and accuracy. In the new proposed study of this dissertation, a quasi-cine CBCT estimation technique was developed to address these issues for real-time application. Specifically, a new structural PCA method was developed to build a structural motion model (SMM) instead of GMM by accounting for potential relative motion pattern changes between different anatomical structures from simulation to treatment. The motion model extracted from planning 4D CT was divided into two structures: tumor and body excluding tumor, and the parameters of both structures were optimized together. Weighted free-form deformation (WFD) was employed afterwards to introduce flexibility in adjusting the weightings of different structures in the data fidelity constraint based on clinical interests. As such, the localization accuracy could be substantially improved with extremely limited angle kV projections. The technique was evaluated by simulating a 30 mm diameter lesion in a computerized patient model (XCAT) with various anatomical and respiratory changes from planning 4D CT to on-board volume. The estimation accuracy was evaluated by the volume percent difference (VPD)/center-of-mass-shift (COMS) between lesions in the estimated and "ground-truth" on-board quasi-cine CBCT. Different on-board projection acquisition scenarios and projection noise levels were simulated to investigate their effects on the estimation accuracy. The method was also evaluated against three lung patients.

The SMM-WFD method achieved substantially better accuracy than the GMM-FD method for CBCT estimation using extremely small scan angles or projections. Using orthogonal 15° scanning angles, the VPD/COMS were 3.47 ± 2.94% and 0.23 ± 0.22 mm for SMM-WFD and 25.23 ± 19.01% and 2.58 ± 2.54 mm for GMM-FD among all eight XCAT scenarios. Compared to GMM-FD, SMM-WFD was more robust against reduction of the scanning angles down to orthogonal 10° with VPD/COMS of 6.21 ± 5.61% and 0.39 ± 0.49 mm, and more robust against reduction of projection numbers down to only 8 projections in total for both orthogonal-view 30° and orthogonal-view 15° scan angles. SMM-WFD method was also more robust than the GMM-FD method against increasing levels of noise in the projection images. Additionally, the SMM-WFD technique provided better tumor estimation for all three lung patients compared to the GMM-FD technique.

The first technique developed in this dissertation showed that compared to the GMM-FD technique, the SMM-WFD technique can substantially improve the CBCT estimation accuracy using extremely small scan angles and low number of projections to provide fast low dose 4D target verification.

The next section of the dissertation describes ways in which volumetric cine MRI (VC-MRI) was developed for MRI-guided radiation therapy. Currently, there are no ways to image real-time MRI in three-dimensions (3D) for on-board MRI-guided radiotherapy. In the first subsection of the second section of this dissertation, a novel technique was developed to generate, for the first time, real-time 3D VC-MRI using patient prior images, motion modeling and on-board two-dimensional (2D) cine MRI.

One phase of a 4D MRI acquired during patient simulation is used as patient prior images. Three major respiratory deformation patterns of the patient are extracted from 4D MRI based on PCA. The on-board VC-MRI at any instant is considered as a deformation of the prior MRI. The deformation field is represented as a linear combination of the 3 major deformation patterns. The coefficients of the deformation patterns are solved by the data fidelity constraint using the acquired on-board single 2D cine MRI. The method was evaluated using both XCAT simulation of lung cancer patients and MRI data from 4 real liver cancer patients. The accuracy of the estimated VC-MRI was quantitatively evaluated using VPD, COMS, and target tracking errors. Effects of acquisition orientation, region-of-interest (ROI) selection, patient breathing pattern change, and noise on the estimation accuracy were also evaluated.

Image subtraction of ground-truth with estimated on-board VC-MRI showed fewer differences than image subtraction of ground-truth with prior image. Agreement between normalized profiles in the estimated and ground-truth VC-MRI was achieved with less than 6% error for both XCAT and patient data. Among all XCAT scenarios, the VPD between ground-truth and estimated lesion volumes was, on average, 8.43 ± 1.52% and the COMS was, on average, 0.93 ± 0.58 mm across all time steps for estimation based on the ROI region in the sagittal cine images. Matching to ROI in the sagittal view achieved better accuracy when there was substantial breathing pattern change. The technique was robust against noise levels up to Signal-to-Noise Ratio (SNR) = 20. For patient data, average tracking errors were less than 2 mm in all directions for all patients. The feasibility of generating real-time VC-MRI for on-board localization of moving targets was demonstrated.

Next, a technique was developed to explore the feasibility of using an on-board kV imaging system and patient prior MRI knowledge to generate on-board quasi-cine volumetric MRI for target localization. Very few clinics have MRI-guided radiotherapy units, but most clinics have kV imaging capabilities with a conventional LINAC. The technique developed in this section of the dissertation aims to utilize conventional LINAC imaging capabilities, along with prior patient 4D MRI to estimate on-board 4D MRI for MRI-guided radiotherapy. Prior 4D MRI volumes were separated into end-of-expiration (EOE) phase (MRIprior) and all other phases. MRIprior was used to generate a synthetic CT at EOE phase (sCTprior). On-board quasi-cine 3D or 4D MRI at each respiratory phase was considered a deformation of MRIprior. The Deformation Field Map (DFM) was estimated by matching Digitally Reconstructed Radiographs (DRRs) of the deformed sCTprior to on-board kV projections using a MM-FD deformation optimization algorithm. The on-board 4D MRI method was evaluated using both XCAT simulation and real patient data. The accuracy of the estimated MRI was quantitatively evaluated using VPD, Volume Dice Coefficient (VDC) and COMS. Effects of scan angle and number of projections were also evaluated.

In the XCAT study, VPD/VDC/COMS among all XCAT scenarios were 10.16±1.31%/0.95±0.01/0.88±0.15mm using orthogonal-view 30° scan angles with 102 projections. The on-board 4D MRI method was robust against various scan angles and projection numbers evaluated. In the patient study, estimated on-board 4D MRI was generated successfully when compared to the 'reference on-board 4D MRI' for the liver patient case. Preliminary results for this novel technique demonstrated the potential for MRI-based image guidance for liver SBRT using only a kV imaging system on a conventional LINAC.

The last section of the dissertation aims to accelerate the VC-MRI technique developed and improve the estimation accuracy by using novel undersampling and deformation models. VC-MRI was accelerated by using undersampled 2D cine MRI to provide real-time 3D guidance. Undersampled Cartesian and radial k-space acquisition strategies were investigated. The effects of k-space sampling percentage (SP) and distribution, tumor sizes and noise on the VC-MRI estimation were studied. The accelerated VC-MRI estimation was evaluated using XCAT simulation of lung cancer patients and data from liver cancer patients. VPD and COMS of the tumor volumes and tumor tracking errors were calculated.

For XCAT, VPD/COMS were 11.93  ±  2.37%/0.90  ±  0.27 mm and 11.53  ±  1.47%/0.85  ±  0.20 mm among all scenarios with Cartesian sampling (SP  =  10%) and radial sampling (21 spokes, SP  =  5.2%), respectively. When tumor size decreased, higher sampling rate achieved more accurate VC-MRI than lower sampling rate. VC-MRI was robust against noise levels up to SNR  =  20. For patient data, the tumor tracking errors in superior-inferior, anterior-posterior and lateral directions were 0.46  ±  0.20 mm, 0.56  ±  0.17 mm and 0.23  ±  0.16 mm, respectively, for Cartesian-based sampling with SP  =  20% and 0.60  ±  0.19 mm, 0.56  ±  0.22 mm and 0.42  ±  0.15 mm, respectively, for radial-based sampling with SP  =  8% (32 spokes).

Results from this method showed that VC-MRI could be accelerated from a single undersampled on-board 2D cine MRI. Phantom and patient studies showed that the temporal resolution of VC-MRI can potentially be improved by 5-10 times using a 2D cine image acquired with 10-20% k-space sampling.

Lastly, VC-MRI accuracy was improved by using multi-slice undersampled cine images, patient prior 4D MRI, motion modeling and free-form deformation. In this study, free-form deformation (FD) was introduced to correct for errors in the MM when large anatomical changes exist. Multiple-slice undersampled on-board 2D-cine images were used by reconstructing 10% of total k-space using a novel k-t SLR reconstruction method, which uses both the spatial and temporal resolution of the k-space data to reconstruct the 2D cine MRIs. The method was evaluated using XCAT simulation of lung cancer patients with various anatomical and respirational changes from prior 4D MRI to onboard volume. The accuracy was evaluated using VPD, VDC and COMS of the estimated tumor volume. Effects of ROI selection, 2D-cine slice orientation, slice number and slice location on the estimation accuracy were evaluated.

VC-MRI estimated using 10 undersampled sagittal 2D cine MRIs achieved VPD/VDC/COMS of 9.77±3.71%/0.95±0.02/0.75±0.26mm among all scenarios based on estimation with ROI MM and ROI FD. The FD optimization improved estimation significantly for scenarios with anatomical changes. Using ROI FD achieved better estimation than global FD. Changing the multi-slice orientation to axial, coronal, and axial/sagittal orthogonal reduced the accuracy of VC-MRI. Estimation using slices sampled uniformly through the tumor achieved better accuracy than using slices sampled non-uniformly.

In conclusion, the work presented in his dissertation builds upon previous research and develops novel solutions for generating real-time volumetric cine images for both CBCT and MRI. The completed research dissertation presents the following: (1) develops a quasi-real-time cine CBCT reconstruction method using structural PCA and weighted free-form deformation, (2) develops a VC-MRI technique using motion modeling and single slice 2D cine acquisition, (3) develops an on-board 4D-MRI technique using limited on-board kV projections from a conventional LINAC and deformation models, (4) accelerates VC-MRI through undersampling acquisition while maintaining image quality, and (5) improve geometric accuracy of VC-MRI through novel undersampling acquisition and deformation models.