Purpose: This study aims to develop a DIR error correction method capable of utilizing sparse ground‐truth motion information and recovering missing data with the Kolmogorov‐Zurbenko (KZ) filter. Methods: The error correction method employs a two‐step approach. First, sparse ground‐truth displacement vectors are integrated into a pre‐correction deformable vector field (DVF) to estimate a post‐correction DVF with coarse resolution. Second, the coarse post‐correction DVF is boosted to a full‐resolution DVF through convolution with the KZ filter. To validate the use of the KZ filter for missing data recovery, recovery errors were determined by comparing a DVF recovered from down‐sampling with the original full‐resolution DVF. The entire error correction method was tested on an in‐house developed digital lung motion phantom consisting of a primary volume, a DVF, and a secondary volume synthesized by applying the DVF on the primary volume. Five pre‐correction DVFs were obtained by performing DIR between the two volumes using Velocity, MIM, ILK and OHS algorithms in DIRART toolbox, and Elastix, and then corrected. Primary volumes were synthesized with pre‐ and post‐correction DVFs, respectively. The error correction method was evaluated with pre‐ and post‐correction registration errors, and intensity errors in synthesized primary volumes. Results: Our test results for sparsely down‐sampled (<0.4%) DVFs showed that the KZ filter outperformed the cubic polynomial interpolation method for whole lung DVF map recovery in terms of median error (0.60mm vs 0.73mm) and mean error (1.18mm vs 1.29mm). Pre‐ and post‐correction 3D registration errors per voxel for Velocity, MIM, ILK, OHS, and Elastix are reduced by 2.39 mm on average. Pre‐ and post‐correction intensity errors are reduced by 0.37 in unit of image intensity on average. Conclusion: We have implemented a two‐step method capable of utilizing sparse ground‐truth displacement vectors for DIR error reduction, allowing DIR accuracy improvement utilizing clinically available motion data. This study is supported by NIH grant 1R21CA165384.

Purpose:To develop a technique to estimate on‐board VC‐MRI using multi‐slice sparsely‐sampled cine images, patient prior 4D‐MRI, motion‐modeling and free‐form deformation for real‐time 3D target verification of lung radiotherapy.Methods:A previous method has been developed to generate on‐board VC‐MRI by deforming prior MRI images based on a motion model(MM) extracted from prior 4D‐MRI and a single‐slice on‐board 2D‐cine image. In this study, free‐form deformation(FD) was introduced to correct for errors in the MM when large anatomical changes exist. Multiple‐slice sparsely‐sampled on‐board 2D‐cine images located within the target are used to improve both the estimation accuracy and temporal resolution of VC‐MRI. The on‐board 2D‐cine MRIs are acquired at 20–30frames/s by sampling only 10% of the k‐space on Cartesian grid, with 85% of that taken at the central k‐space. The method was evaluated using XCAT(computerized patient model) simulation of lung cancer patients with various anatomical and respirational changes from prior 4D‐MRI to onboard volume. The accuracy was evaluated using Volume‐Percent‐Difference(VPD) and Center‐of‐Mass‐Shift(COMS) of the estimated tumor volume. Effects of region‐of‐interest(ROI) selection, 2D‐cine slice orientation, slice number and slice location on the estimation accuracy were evaluated.Results:VCMRI estimated using 10 sparsely‐sampled sagittal 2D‐cine MRIs achieved VPD/COMS of 9.07±3.54%/0.45±0.53mm among all scenarios based on estimation with ROI_MM‐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 VCMRI to VPD/COMS of 19.47±15.74%/1.57±2.54mm, 20.70±9.97%/2.34±0.92mm, and 16.02±13.79%/0.60±0.82mm, respectively. Reducing the number of cines to 8 enhanced temporal resolution of VC‐MRI by 25% while maintaining the estimation accuracy. Estimation using slices sampled uniformly through the tumor achieved better accuracy than slices sampled non‐uniformly.Conclusions:Preliminary studies showed that it is feasible to generate VC‐MRI from multi‐slice sparsely‐sampled 2D‐cine images for real‐time 3D‐target verification.This work was supported by the National Institutes of Health under Grant No. R01‐CA184173 and a research grant from Varian Medical Systems.

Purpose: To develop a technique generating ultrafast on‐board VC‐MRI using prior 4D‐MRI and on‐board phase‐skipped encoding k‐space acquisition for real‐time 3D target tracking of liver and lung radiotherapy. Methods: The end‐of‐expiration (EOE) volume in 4D‐MRI acquired during the simulation was selected as the prior volume. 3 major respiratory deformation patterns were extracted through the principal component analysis of the deformation field maps (DFMs) generated between EOE and all other phases. The on‐board VC‐MRI at each instant was considered as a deformation of the prior volume, and the deformation was modeled as a linear combination of the extracted 3 major deformation patterns. To solve the weighting coefficients of the 3 major patterns, a 2D slice was extracted from VC‐MRI volume to match with the 2D on‐board sampling data, which was generated by 8‐fold phase skipped‐encoding k‐space acquisition (i.e., sample 1 phase‐encoding line out of every 8 lines) to achieve an ultrafast 16–24 volumes/s frame rate. The method was evaluated using XCAT digital phantom to simulate lung cancer patients. The 3D volume of end‐ofinhalation (EOI) phase at the treatment day was used as ground‐truth onboard VC‐MRI with simulated changes in 1) breathing amplitude and 2) breathing amplitude/phase change from the simulation day. A liver cancer patient case was evaluated for in‐vivo feasibility demonstration. Results: The comparison between ground truth and estimated on‐board VC‐MRI shows good agreements. In XCAT study with changed breathing amplitude, the volume‐percent‐difference(VPD) between ground‐truth and estimated tumor volumes at EOI was 6.28% and the Center‐of‐Mass‐Shift(COMS) was 0.82mm; with changed breathing amplitude and phase, the VPD was 8.50% and the COMS was 0.54mm. The study of liver patient case also demonstrated a promising in vivo feasibility of the proposed method Conclusion: Preliminary results suggest the feasibility to estimate ultrafast VC‐MRI for on‐board target localization with phase skipped‐encoding k‐space acquisition. Research grant from NIH R01‐184173

Purpose: Most current DCE‐MRI texture analysis methods focus on the spatial information of chosen contrast‐enhanced MR volumes or pharmacokinetic (PK) parameter maps, and the temporal information is not well included. This work proposed a novel texture matrix called Gray Level Local Power Matrix (GLLPM) for the accurate and efficient spatiotemporal DCE‐MRI texture analysis in therapeutic response assessment. Methods: A retrospective study with two groups (n=8/group) of tumor implanted mice was conducted. The treatment/control groups received bevacizumab/saline treatment with pre‐ and post‐treatment DCE‐MRI exams. For each scan, the GLLPM was calculated and compared with classic 3D/4D Gray Level Co‐Occurrence Matrices (GLCOM) using the CA concentration maps in the first 10‐minute post‐injection time. The calculation time of each matrix was recorded for efficiency evaluation. Using each matrix, a set of 22 Haralick texture features’ dynamic curves were calculated. The Mann‐Whitney U‐test was used to assess the differences of the Area Under Curve (AUC) of all derived texture feature curves between treatment/control groups. The post‐treatment texture feature curves were fitted by cubic polynomial. Experiments using support vector machine in a leave‐one‐out approach were performed to validate the use of fitted polynomial coefficients of each texture feature curve in treatment/control group classification. Results: The computation efficiency of GLLPM had improved factors of 3 and 20 in comparison with 3D/4D GLCOM, respectively. 21 out of 22 GLLPM texture feature dynamic curves’ AUCs between treatment/control groups had significant differences in post‐treatment scan but not in pre‐treatment scan. N=19 dynamic curves from GLLPM can be fitted by cubic polynomial (R2>0.8), and N for 3D/4D GLCOM were 14 and 19, respectively. The averaged classification accuracies using the post‐treatment texture features curves based on GLLPM, 3D/4D GLCOM were (84.5±12.1)%, (65.6±10.5)% and (73.3±12.8)%, respectively. Conclusion: The proposed GLLPM and its features can be used for the efficient DCE‐MRI therapeutic response assessment.

Purpose: To compare the performance of shutter‐speed(SS) model with transcytolemmal water exchange analysis against the Tofts model in the study of the efficacy of an anti‐angiogenesis drug Methods: 16 mice with LS‐174T implanted were randomly assigned into treatment/control groups (n=8/group) and received bevacizumab/saline three times (Day1/Day4/Day8). All mice received one pre‐ (Day0) and two post‐treatment (Day2/Day9) DCE scans. For each scan, the CA extravasation rate constant KTtrans/KStrans from the Tofts/SS model were calculated. The intracellular water residence time τi which reflects limited transcytolemmal water exchange between cell and extravascular‐extracellular‐space were also analyzed using SS model. A biological subvolume(BV) within the tumor was automatically segmented based on the τi intensity distribution, and the SS model parameters within the BV (KS,BVtrans and τi, BV) were analyzed. Rank‐sum tests were conducted to assess the differences of each parameter's statistics (mean value/coefficient‐of‐variation (CV) /kurtosis/skewness/heterogeneity indices d1 and d2) between treatment/control groups. Experiment using support vector machine in a leave‐one‐out approach were performed to validate the use of the analyzed biomarkers for treatment/control classification. Results: The SS model was a better fit for all scans in terms of Bayesian information criterion. At Day9, the treatment group had significantly higher mean KTtrans(p=0.021), KStrans(p=0.021) and τi(p=0.045). In the identified BV, the treatment group had significantly higher mean KS,BVtrans at both Day2(p=0.038) and Day9(p=0.007). Additionally, at Day9, the treatment group had significantly higher mean τi, BV(p=0.045) and higher KS,BVtrans heterogeneity indices d1(p=0.010) and d2(p=0.021) values. When using KS,BVtrans statistics for treatment/control group classification, the highest accuracy was 68.8%/87.5% at Day2/Day9; this result was better than the result of 62.5%/87.5% using KStrans statistics and 50.0%/87.5% using KTtrans statistics. Conclusion: The SS model parameters may be more reliable than the Tofts model parameters for therapeutic assessment. The proposed biological subvolume in this work may be useful for early therapeutic effect monitoring.