Kazuma Kishi

Evaluation of various deformable image registration algorithms for thoracic images

We evaluated the accuracy of one commercially available and three publicly available deformable image registration (DIR) algorithms for thoracic four-dimensional (4D) computed tomography (CT) images. Five patients with esophagus cancer were studied. Datasets of the five patients were provided by DIR-lab (dir-lab.com) and consisted of thoracic 4D CT images and a coordinate list of anatomical landmarks that had been manually identified. Expert landmark correspondence was used for evaluating DIR spatial accuracy. First, the manually measured displacement vector field (mDVF) was obtained from the coordinate list of anatomical landmarks. Then the automatically calculated displacement vector field (aDVF) was calculated by using the following four DIR algorithms: B-spine implemented in Velocity AI (Velocity Medical, Atlanta, GA, USA), free-form deformation (FFD), Horn–Schunk optical flow (OF) and Demons in DIRART of MATLAB software. Registration error is defined as the difference between mDVF and aDVF. The mean 3D registration errors were 2.7 ± 0.8 mm for B-spline, 3.6 ± 1.0 mm for FFD, 2.4 ± 0.9 mm for OF and 2.4 ± 1.2 mm for Demons. The results showed that reasonable accuracy was achieved in B-spline, OF and Demons, and that these algorithms have the potential to be used for 4D dose calculation, automatic image segmentation and 4D CT ventilation imaging in patients with thoracic cancer. However, for all algorithms, the accuracy might be improved by using the optimized parameter setting. Furthermore, for B-spline in Velocity AI, the 3D registration error was small with displacements of less than ∼10 mm, indicating that this software may be useful in this range of displacements.

Evaluation of accuracy of B-spline transformation-based deformable image registration with different parameter settings for thoracic images

Deformable image registration (DIR) is fundamental technique for adaptive radiotherapy and image-guided radiotherapy. However, further improvement of DIR is still needed. We evaluated the accuracy of B-spline transformation-based DIR implemented in elastix. This registration package is largely based on the Insight Segmentation and Registration Toolkit (ITK), and several new functions were implemented to achieve high DIR accuracy. The purpose of this study was to clarify whether new functions implemented in elastix are useful for improving DIR accuracy. Thoracic 4D computed tomography images of ten patients with esophageal or lung cancer were studied. Datasets for these patients were provided by DIR-lab (dir-lab.com) and included a coordinate list of anatomical landmarks that had been manually identified. DIR between peak-inhale and peak-exhale images was performed with four types of parameter settings. The first one represents original ITK (Parameter 1). The second employs the new function of elastix (Parameter 2), and the third was created to verify whether new functions improve DIR accuracy while keeping computational time (Parameter 3). The last one partially employs a new function (Parameter 4). Registration errors for these parameter settings were calculated using the manually determined landmark pairs. 3D registration errors with standard deviation over all cases were 1.78 (1.57), 1.28 (1.10), 1.44 (1.09) and 1.36 (1.35) mm for Parameter 1, 2, 3 and 4, respectively, indicating that the new functions are useful for improving DIR accuracy, even while maintaining the computational time, and this B-spline-based DIR could be used clinically to achieve high-accuracy adaptive radiotherapy.

Evaluation of On-Board kV Cone Beam Computed Tomography–Based Dose Calculation With Deformable Image Registration Using Hounsfield Unit Modifications

PURPOSE The purpose of this study was to estimate the accuracy of the dose calculation of On-Board Imager (Varian, Palo Alto, CA) cone beam computed tomography (CBCT) with deformable image registration (DIR), using the multilevel-threshold (MLT) algorithm and histogram matching (HM) algorithm in pelvic radiation therapy. METHODS AND MATERIALS One pelvis phantom and 10 patients with prostate cancer treated with intensity modulated radiation therapy were studied. To minimize the effect of organ deformation and different Hounsfield unit values between planning CT (PCT) and CBCT, we modified CBCT (mCBCT) with DIR by using the MLT (mCBCT(MLT)) and HM (mCBCT(HM)) algorithms. To evaluate the accuracy of the dose calculation, we compared dose differences in dosimetric parameters (mean dose [D(mean)], minimum dose [D(min)], and maximum dose [D(max)]) for planning target volume, rectum, and bladder between PCT (reference) and CBCTs or mCBCTs. Furthermore, we investigated the effect of organ deformation compared with DIR and rigid registration (RR). We determined whether dose differences between PCT and mCBCTs were significantly lower than in CBCT by using Student t test. RESULTS For patients, the average dose differences in all dosimetric parameters of CBCT with DIR were smaller than those of CBCT with RR (eg, rectum; 0.54% for DIR vs 1.24% for RR). For the mCBCTs with DIR, the average dose differences in all dosimetric parameters were less than 1.0%. CONCLUSIONS We evaluated the accuracy of the dose calculation in CBCT, mCBCT(MLT), and mCBCT(HM) with DIR for 10 patients. The results showed that dose differences in D(mean), D(min), and D(max) in mCBCTs were within 1%, which were significantly better than those in CBCT, especially for the rectum (P<.05). Our results indicate that the mCBCT(MLT) and mCBCT(HM) can be useful for improving the dose calculation for adaptive radiation therapy.

Evaluation of patient DVH‐based QA metrics for prostate VMAT: correlation between accuracy of estimated 3D patient dose and magnitude of MLC misalignment

The purpose of this study was to evaluate the accuracy of commercially available software, using patient DVH‐based QA metrics, by investigating the correlation between estimated 3D patient dose and magnitude of MLC misalignments. We tested 3DVH software with an ArcCHECK. Two different calculating modes of ArcCHECK Planned Dose Perturbation (ACPDP) were used: “Normal Sensitivity” and “High Sensitivity”. Ten prostate cancer patients treated with hypofractionated VMAT (67.6 Gy/26 Fr) in our hospital were studied. For the baseline plan, we induced MLC errors (−0.75,−0.5,−0.25,0.25,0.5, and 0.75 mm for each single bank). We calculated the dose differences between the ACPDP dose with error and TPS dose with error using gamma passing rates and using DVH‐based QA metrics. The correlations between dose estimation error and MLC position error varied with each structure and metric. A comparison using 1%/1 mm gamma index showed that the larger was the MLC error‐induced, the worse were the gamma passing rates. Slopes of linear fit to dose estimation error versus MLC position error for mean dose and D95 to the PTV were 1.76 and 1.40% mm−1, respectively, for “Normal Sensitivity”, and −0.53 and 0.88% mm−1, respectively, for “High Sensitivity”, showing better accuracy for “High Sensitivity” than “Normal Sensitivity”. On the other hand, the slopes for mean dose to the rectum and bladder, V35 to the rectum and bladder and V55 to the rectum and bladder, were −1.00,−0.55,−2.56,−1.25,−3.53, and 1.85% mm−1, respectively, for “Normal Sensitivity”, and −2.89,−2.39,−4.54,−3.12,−6.24, and −4.11% mm−1, respectively, for “High Sensitivity”, showing significant better accuracy for “Normal Sensitivity” than “High Sensitivity”. Our results showed that 3DVH had some residual error for both sensitivities. Furthermore, we found that “Normal Sensitivity” might have better accuracy for the DVH metric for the PTV and that “High Sensitivity” might have better accuracy for DVH metrics for the rectum and bladder. We must be willing to tolerate this residual error in clinical care. PACS number: 87.55Qr

Evaluation of four-dimensional computed tomography (4D-CT)-based pulmonary ventilation: The high correlation between 4D-CT ventilation and 81mKr-planar images was found

PURPOSE To evaluate four-dimensional computed tomography (4D-CT)-derived pulmonary ventilation by comparing with (81m)Kr-gas ventilation (VRI). We also proposed two methods to improve the functional accuracy of 4D-CT ventilation images and evaluated these methods. METHODS AND MATERIALS Eleven lung cancer patients with 4D-CT and VRI were analyzed. Hounsfield unit-based (VHU) and a Jacobian-based (VJac) 4D-CT ventilation images were calculated. They were evaluated by voxel-by-voxel spearmans rank correlation coefficient (r) between 4D-CT ventilation and VRI images. After applying an averaging ventilation method and a slope calculating method, correlations were also calculated. RESULTS 4D-CT ventilation showed the high correlation to VRI (r=0.875 with VHU). An averaging method brought significantly higher (p=0.012) correlations to nuclear medicine images with VHU. The improvement was not significant (p=0.619) with VJac. Slope calculating method improved the correlation with VHU and slightly worsened the correlation with VJac. CONCLUSIONS The averaging method we proposed might be useful to improve 4D-CT ventilation images. We found good agreement between 4D-CT ventilation and nuclear medicine ventilation, indicating the high physiologic accuracy of 4D-CT ventilation.

OC-0271: Positional accuracy valuation of a three dimensional printed device for head and neck immobilisation

ESTRO 35 2016 _____________________________________________________________________________________________________ original GTV when contouring the GTV on the anatomy of the second CT scan.SIB created two plans. One is 1st CT / 1st Plan and the other is SIB sum (25 fractions (deformed CT) and 5 fractions ( 2nd CT )) . A deformed CT (dCT) with structures was created by deforming the 1st CT to the 2nd CT. We summed up dose used in 1st Plan and 2nd Plan using a commercially software ( MIM Maestro 6.3 ). The two types of plans were compared with respect to DVHs for other dosimetric parameters of the PTVboost, PTVel, brainstem, spinal cord and parotid gland.

The impact of audio-visual biofeedback with a patient-specific guiding waveform on respiratory motion management: Comparison of two different respiratory management systems

Irregular breathing can influence the outcome of four-dimensional computed tomography imaging for causing artifacts. Audio-visual biofeedback systems associated with patient-specific guiding waveform are known to reduce respiratory irregularities. In Japan, abdomen and chest motion selfcontrol devices (Abches), representing simpler visual coaching techniques without guiding waveform are used instead; however, no studies have compared these two systems to date. Here, we evaluate the effectiveness of respiratory coaching to reduce respiratory irregularities by comparing two respiratory management systems. We collected data from eleven healthy volunteers. Bar and wave models were used as audio-visual biofeedback systems. Abches consisted of a respiratory indicator indicating the end of each expiration and inspiration motion. Respiratory variations were quantified as root mean squared error (RMSE) of displacement and period of breathing cycles. All coaching techniques improved respiratory variation, compared to free-breathing. Displacement RMSEs were 1.43 ± 0.84, 1.22 ± 1.13, 1.21 ± 0.86, and 0.98 ± 0.47 mm for free-breathing, Abches, bar model, and wave model, respectively. Freebreathing and wave model differed significantly (p < 0.05). Period RMSEs were 0.48 ± 0.42, 0.33 ± 0.31, 0.23 ± 0.18, and 0.17 ± 0.05 s for free-breathing, Abches, bar model, and wave model, respectively. For variation in both displacement and period, wave model was superior to free-breathing, bar model, and Abches. The average reduction in displacement and period RMSE compared with wave model were 27% and 47%, respectively. The efficacy of audio-visual biofeedback to reduce respiratory irregularity compared with Abches. Our results showed that audio-visual biofeedback combined with a wave model can potentially provide clinical benefits in respiratory management, although all techniques could reduce respiratory irregularities.

Clinical Factors Relating to Cervical Body Volume Reduction during Curative External Beam Radiation Therapy for Head and Neck Cancer

Purpose/Objectives: Substantial cervical body volume reduction (CBVR) occurs during fractionated external beam radiation therapy (EBRT) for head-and-neck cancer (HNC) and could have potential dosimetric influences. This study aims to investigate measurable clinical factors before treatment initiation correlating with CBVR during curative EBRT in HNC patients, and to determine which patients receive the great benefit from routine adaptive radiation therapy (ART). Materials/Methods: Fifty-six patients with oropharyngeal squamous cell carcinoma (OSCC) and 67 patients with hypopharyngeal squamous cell carcinoma (HSCC) had received curative EBRT between 2006 and 2013 were enrolled. For EBRT planning, computed tomography (CT) images were acquired before EBRT initiation and between two to seven weeks after the start of EBRT for replanning in each patient. A MATLAB program was used to evaluate the CBVR rate (CBVRR) between the initial and replanning CT imaging. The following factors were assessed for correlation with CBVRR: the T and N stage, induction and concurrent chemotherapy, the initial gross tumor volume (GTV), the GTV reduction rate (GTVRR) between the initial and replanning CT imaging, the initial body weight (BW) and the BW loss rate (BWLR) during the EBRT course. Results: In the OSCC group, the CBVRR ranged from 1.8 to 17.1% (median, 6.8%). In the HSCC group, the CBVRR ranged from 1.2 to 23.7% (median, 6.5%). In non-parametric univariate analysis, the N3 stage demonstrated a greater trend with the CBVRR than the N2c≥ stage in the HSCC group (p=0.023), whereas marginal inclination (p=0.096) was found in the OSCC group. The CBVRR was substantially related to the GTVRR (p=0.001) in the HSCC group.

Assessment of a Commercially Available Automatic Deformable Image Registration

Ventilation imaging can be performed using thoracic four dimensional computed tomography (4D-CT) images (max inhale phase and max exhale phase) and deformable image registration (DIR). If this method was administered in multi institution, some institution would use commercially available automatic DIR software. But, there are not many reports about commercially available automatic DIR. In this study, we evaluated the accuracy of a commercially available automatic deformable image registration (DIR) algorithm using 4D-CT images. For evaluating the accuracy of DIR, registration error was calculated by difference between manual displacement and automatic calculated displacement (DIR outputs). A B-spline DIR algorithm implemented in a Velocity AI ver. 2.7.0 software (Velocity Medial, GA, USA) was evaluated. 4D-CT images including 300 landmarks /case, throughout the lung, provided by DIR-lab (www.dir-lab.com). In this study, five patients were studied. The goal of DIR was to find a point to point correspondence between inhale image and exhale image. First, manual displacement was calculated by land mark points between max inhale phase and max exhale phase. Next, DIR outputs were calculated by a Velocity AI. After that, registration error was calculated by difference between manual displacement and DIR outputs. The mean 3D registration error (standard deviation) for the five cases was 2.70 (2.24) mm. Fewer large errors were seen, but the frequent histogram had a peak at 1.5mm of 3D error, and the frequencies decline as one moves away from the peak. The average 3D registration errors for case1 were 0.94 mm for 1.5 mm motion distance magnitude, 1.96 mm for 6.0 mm and 3.70 mm for 9.0 mm, respectively. Our result clearly shows that the accuracy of DIR in Velocity AI was within 3.0 mm. Therefore commercially available automatic DIR may be useful for image-guided radiation therapy, adaptive radiation therapy and ventilation imaging.

Evaluation of dose calculation accuracy of modified CBCT using Multi -level-threshold algorithm

We evaluated dose calculation accuracy for modi- fied Cone-Beam computed tomography (CBCT) images using Multi -level-threshold algorithm, and compared it with un- modified CBCT images. Inhomogeneous pelvis phantom imag- es and three prostate cancer patient images were acquired from both planning CT (PCT) and CBCT. The Hounsfield Units (HUs) were measured for air, tissue, bone. Using in- house software, three different tissue types were differentiated in both imaging modalities. The HUs of CBCT images were replaced by the mean HUs of the same tissue type of the PCT. An intensity modulated radiation therapy (IMRT) plan was created on PCT and copied to the modified CBCT. To evaluate the dosimetric accuracy, dose distributions based on CBCT images were compared it with PCT for four datasets (one inhomogeneous phantom, three prostate cancer patients ). HU- ED calibration acquired with PCT was used in both images. In the inhomogeneous pelvis phantom, differences in dose param- eter were seen. Regarding PTV, Rectum, and Bladder, differ- ences in mean dose between PCT and CBCT were 0.4%, 0.4%, and 0.1%, respectively, whereas those between PCT and modi- fied CBCT were 0.2%, 0.1%, and 0%, respectively. In the patient study, the average dose difference of PTV, Rectum, and Bladder between PCT and CBCT were 1.0%, 0.2%, and 0.6%, respectively, compared to PCT and modified CBCT, 0.5%, 0.1%, and 0.2%, respectively. We found that the dose difference in the plans based on PCT and CBCT were de- creased after this modification. The uncertainty in patient positioning between PCT and CBCT may cause additional discrepancies in the calculated results compared to the results in the phantom study. An accurate dose calculation based on CBCT images is possible when density distributions are cor- rected. This method does not need to acquire HU-ED calibra- tion for CBCT, and may be able to remove scatter.