A Godley

An on-line replanning method for head and neck adaptive radiotherapya)

Daily setup for head and neck (HN) radiotherapy (RT) can vary randomly due to neck rotation and anatomy change. These differences cannot be totally corrected by the current practice of image guided RT with translational repositioning. The authors present a novel rapid correction scheme that can be used on-line to correct both interfractional setup variation and anatomy change for HN RT. The scheme consists of two major steps: (1) Segment aperture morphing (SAM) and (2) segment weight optimization (SWO). SAM is accomplished by applying the spatial relationship between the apertures and the contours of the planning target and organs at risk (OARs) to the new target and OAR contours. The new target contours are transferred from planning target contours to the CT of the day by means of deformable registration (MIMVISTA). The dose distribution for each new aperture was generated using a planning system with a fast dose engine and hardware and was input into a newly developed SWO package using fast sequential quadratic programming. The entire scheme was tested based on the daily CT images acquired for representative HN IMRT cases treated with a linac and CT-on-Rails combo. It was found that the target coverage and/or OAR sparing was degraded based on the CT of the day with the current standard repositioning from rigid registration. This degradation can be corrected by the SAM/SWO scheme. The target coverage and OAR sparing for the SAM/SWO plans were found to be equivalent to the original plan. The SAM/SWO process took 5-8 min for the head and neck cases studied. The proposed aperture morphing with weight optimization is an effective on-line approach for correcting interfractional patient setup and anatomic changes for head and neck cancer radiotherapy.

Accumulating daily-varied dose distributions of prostate radiation therapy with soft-tissue–based kV CT guidance

Even with daily image guidance based on soft tissue registration, deviations of fractional doses can be quite large due to changes in patient anatomy. It is of interest to ascertain the cumulative effect of these deviations on the total delivered dose. Daily kV CT data acquired using an in‐room CT for five prostate cancer patients were analyzed. Each daily CT was deformably registered to the planning CT using an in‐house tool. The resulting deformation field was used to map the delivered daily dose onto the planning CT, then summed to obtain the cumulative (total delivered) dose to the patient. The delivered cumulative values of prostate D100 on average were only 2.9% less than their planned values, while the PTV D95 were 3.6% less. The delivered rectum and bladder V70s can be twice what was planned. The less than 3% difference between delivered and planned prostate coverage indicates that the PTV margin of 5 mm was sufficient with the soft‐tissue–based kV CT guidance for the cases studied. PACS number: 87.55.km

Interfractional Target Variations for Partial Breast Irradiation

PURPOSE In this work, we quantify the interfractional variations in the shape of the clinical target volume (CTV) by analyzing the daily CT data acquired during CT-guided partial breast irradiation (PBI) and compare the effectiveness of various repositioning alignment strategies considered to account for the variations. METHODS AND MATERIALS The daily CT data for 13 breast cancer patients treated with PBI in either prone (10 patients) or supine (3 patients) with daily kV CT guidance using CT on Rails (CTVision, Siemens, Malvern, PA) were analyzed. For approximately 25 points on the surface of the CTV, deformation vectors were calculated by means of deformable image registration and verified by visual inspection. These were used to calculate the distances along surface normals (DSN), which directly related to the required margin expansions for each point. The DSN values were determined for seven alignment methods based on volumetric imaging and also two-dimensional projections (portal imaging). RESULTS The margin expansion necessary to cover 99% of all points for all days was 2.7 mm when utilizing the alignment method based on deformation field data (the best alignment method). The center-of-mass based alignment yielded slightly worse results (a margin of 4.0 mm), and shifts obtained by operator placement (7.9 mm), two-dimensional-based methods (7.0-10.1 mm), and skin marks (13.9 mm) required even larger margin expansions. Target shrinkage was evident for most days by the negative values of DSN. Even with the best alignment, the range of DSN values could be as high as 7 mm, resulting in a large amount of normal tissue irradiation, unless adaptive replanning is employed. CONCLUSION The appropriate alignment method is important to minimize the margin requirement to cover the significant interfractional target deformations observed during PBI. The amount of normal tissue unnecessarily irradiated is still not insignificant, and can be minimized if adaptive radiotherapy is applied.

MO-EE-A3-05: Improving the Temporal Resolution of Dynamic MRI by Deformable Alignment of the Peripheral K-Space

Introduction Dynamic MRI may be used to image motion that is needed for accurate radiotherapytreatment planning. The tradeoff between temporal and spatial resolution has been a major issue affecting the use of dynamic MRI. Here we propose to improve temporal resolution by reconstructing high spatial resolution components with deformably registered motion data. Dynamic MRI with increased temporal resolution without scarifying spatial resolution for several sites are presenting. Methods and Materials High and low spatial resolution components of the image typically move similarly, i.e. share most of the motion vector information. The motion vectors can be obtained by deformable registration of the central k‐space, and applied to the high spatial resolution data. This allows one to acquire the peripheral k‐space at a lower temporal resolution than the central k‐space therefore increase the overall sampling rate. We tested the possible gain by this method for 2‐dimensional dynamic MRIs (SIEMENS 3T Verio) of swallowing, breathing and peristalsis. Results The mean error in the displacement vectors obtained using only the central 20% of k‐space vs. using the full k‐space was less than 0.5mm, which reduced to 0.2mm if 50% of the radius is used. Peristalsis motion with sudden changes shared least information between central and peripheral k‐space. The reconstructed images with as low as central 20% of the k‐space radius are very similar to imagesreconstructed with full k‐space. The resulting probability distance histograms and regions of interest show minimal variation. Conclusion Temporal resolution can be improved considerably without compromising the spatial resolution if the low‐resolution vector information is used to construct high spatial resolution data as demonstrated in this work. The deformable registration accuracy is highly critical for the overall performance of the method. Conflict of Interest: This work is supported partially by Siemens Healthcare.

Collapsed-cone based deformation field regularization for nonrigid image registration

Incorporating biomedical information into nonrigid image registration is an important approach to improve the registration quality and provide realistic results. However, previous tissue-dependent deformation field filtering incur a relatively high computation cost in order to obtain results of improved quality. In this paper, we propose a collapsed-cone based adaptive filtering method to reduce the computational overhead of regularization. The filter is designed to change its filtering parameters dynamically at each voxel according to the tissue characteristics and the deformation of the surrounding voxels. The proposed filter is integrated into the demons deformable registration method to evaluate its effectiveness and performance. The evaluation is performed on a set of 3D computed tomography (CT) images and the result quality is compared with the output of those without applying the tissue-dependent filter. The results show that our proposed method can preserve the global features better. Based on the measure of sum of squared differences (SSD), the proposed method is also found converging faster and leading to lower SSD.

SU‐GG‐J‐43: Accumulating Delivered Doses Based on Daily CT in Head and Neck Cancer Radiotherapy

Purpose: To accumulate actually delivered dose based on daily CTs to evaluate patient repositioning techniques and adaptive strategies for head and neck cancerradiotherapy.Methods: Daily CTs acquired for 28 Head and Neck cancer patients using a CT‐on‐Rails (CTVision, Siemens) were analyzed. These patients were treated with IMRT plans, designed with a 5 mm CTV to PTV margin with prescription in the range of 60–70Gy. Daily plans were reconstructed using a planning system (XiO, CMS) based on daily CTs with iso‐center shifted according to three clinically used alignment techniques: skin markers (BB), soft‐tissue and bone. The daily dose distributions were then deformed using an in‐house deformable registration tool and mapped on to the planning CTs. The cumulative delivered dose distributions with three alignment techniques were compared with planning doses.Results: Although patient repositioning shifts are different between three alignment methods, the accumulated doses for GTV and PTV are comparable. The differences in PTV mean doses between the delivered (accumulative) and the planned values are within 2% for all three alignment methods. However, the accumulated mean doses for parotid glands and salivary glands are different by as much as 20% from their planned values. The soft‐tissue alignment leads to smaller differences. Conclusion: The actual delivered doses for head and neck radiotherapy are calculated by deforming daily CTs to the planning CT for three clinically used patient alignment techniques. While the delivered accumulative doses in targets are similar to the planned doses, the delivered doses to critical structures are different and these differences depend on the alignment rection.

SU‐GG‐T‐20: GPU‐Accelerated Auto‐Segmentation for Online Adaptive Radiotherapy

Purpose: Accurate and fast delineation of targets and normal structures based on daily CTs is critical for online adaptive radiotherapy (ART). Manual contouring generally takes too long to be practical. Auto‐segmentation can run in a few minutes on modern computers, but it is desirable to reduce this time further. A simple, inexpensive option to accelerate auto‐segmentation is to use a graphics processor unit (GPU). This work aims to evaluate the performance of a GPU‐accelerated auto‐segmentation. Method and Materials: A GPU card (NVidia GTX 285) was added to a computer consisting of two 3GHz four core CPUs running auto‐segmentation software, ABAS (Atlas Based Auto Segmentation, Elekta CMS Software). To evaluate performance, this setup was tested using daily CTs acquired during IGRT using a CT‐on‐Rails (CTVision, Siemens) for 14 breast (4 prone, 10 supine) and 6 prostate cancer patients. Contours of targets and critical structures were populated from planning CTs to daily CTs using ABAS with and without the GPU. The accuracy of GPU‐accelerated contouring was measured using Dices Coefficient (DC). Results: For prostate cases, there is no noticeable difference between contours generated with and without GPU ( 95%). The average time to segment images was reduced by 33% with GPU (to 2.8 minutes). For supine breast cases, contours generated with and without GPU agreed equally with contours drawn by physicians. The GPU average segmentation time was 2.3 minutes, reduced by 47%. Conclusion: The use of GPU can substantially accelerate auto‐segmentation without degrading the segmentation accuracy. This time saving is significant for online ART.

TU‐D‐BRC‐07: An Online Replanning Technique with Deformation‐Based Aperture Morphing and Weight Optimization

Purpose: We have previously proposed a fast online replanning method using contour‐based segment aperture morphing (SAM) and segment weight optimization (SWO), that was shown to be effective in correcting interfractional variations. Here we propose to enhance the replanning method by developing new SAM algorithm using deformation field. Methods: A new software tool was developed to incorporate the full 3D deformation field between the planning and daily CTs to morph the planed apertures based on the daily CT. The previously reported deformable registration algorithm based on a fast symmetric Demons method with the use of masks was utilized to generate the deformation vector field. The vector field was used to generate daily contours and to morph aperture shapes. The SWO was applied to further improve the dosimetry for the morphed apertures. The resulting doses were compared with those obtained with the full‐scope reoptimization (the golden standard), rigid‐body‐registration based repositioning (the current IGRT practice), and the previous contour‐based SAM. The new deformation‐based SAM and SWO process was tested for selected prostate cases with large daily organ deformation. Results: The deformable image registration took 1 minute, and the entire deformation‐based SAM and SWO process was completed within 5 minutes. The plans generated by the new deformation‐based SAM were considerably better than those from the repositioning, and were comparable to the contour‐based SAM. SWO further improved the plan quality, resulting in the final plans that are comparable to the full‐blown reoptimization plan. Conclusion: An algorithm using deformation field for aperture morphing was successfully developed and tested. The replanning process including the new algorithm is effective and can be completed within a reasonable timeframe (5–10 minutes) for practical implementation. The new replanning process can be fully automated and can effectively correct for interfraction errors, even for complex cases with multiple targets and overlapping critical structures.

SU-FF-J-132: Calculation of Cumulative Dose for Daily CT-Guided Prostate Irradiation Using Deformable Image Registration

Purpose: To calculate the cumulative dose actually delivered to prostate cancer patients treated with daily image‐guided radiation therapy (IGRT) and to quantify differences between planned and delivereddoses to determine the need of better adaptive strategies. Method and Materials: Daily kV CT data, acquired for ten prostate cancer patients treated with daily IGRT repositioning based on soft‐tissue registration on a CT on Rails (CTVision, Siemens), were analyzed. The dose actually delivered at each fraction was reconstructed by applying the original plan to the daily CT considering the repositioning shifts performed. Each daily CT was deformably registered to the plan CT using a newly developed deformable image registration tool. The resulting deformation field was used to map the delivered daily dose onto the planning CT. The cumulative dose over all treatment fractions was calculated and compared with the planed dose.Results: Accumulated prostate dose is consistently lower than the planned dose. For example, prostate D100 and D95 were reduced by 0.1–6.6% and 0.3–1.2%, respectively, as compared to their planed values. The variation in cumulative dose is reduced as the treatment progresses. The rectum had the largest and most erratic volume disparity, leading to frequent overdosing (mean dose increasing up to 11%). The bladder delivereddose is more likely to be an under‐dose. Conclusion: The cumulative doses actually delivered to the prostate, rectum and bladder are different from their planned values, even with the soft‐tissue‐registration based daily repositioning, the most accurate IGRT, indicating a better adaptive strategy, such as re‐planning, is required to account for interfraction anatomy variation observed in certain treatment fractions/patients.

SU‐FF‐J‐167: MRI‐To‐CT Deformable Registration for Treatment Planning of Breast Irradiation

Purpose: Breast MRI provides enhanced tissue differentiation and physiological information that greatly improves the delineation of targets for breast cancerradiotherapy.MRI needs to be registered with CT as the CT provides the electron density information for dose calculation and alignment in IGRT. We are developing a technique to deformably register MRI to CT, allowing for the transfer of the MRI‐defined structures to the CT for dose calculation, and the dose distribution from CT to MRI for outcome analysis. Method and Materials: An MRI data series was acquired, using a breast coil on a large bore 3T scanner (Verio, Siemens), with prone patient position using multiple sequences including T1, T1 fat saturation (FS), T2, T2FS, and T1FS with contrast. A CT was taken shortly after MRI acquisition. The multi‐form MRIs were rigidly registered to the CT. Anisotropic (edge preserving) noise filtering is applied to the images. A deformable image registration program developed using ITK then registers the MRIs individually to the CT using B‐Spline with Mattes mutual information as a metric and LBFGSB as the optimizer. MRI‐defined structures were overlaid on the CT for treatment planning, and the dose distribution transferred from CT to MRIs for analysis of radiation response based on multi‐form MRIs, using the deformation field obtained. Results: Multi‐form MRIs for prone breast were successfully registered to CT as gauged by high mutual information. Registration amongst the MRI modalities showed negligible deformation between them. T1FS registered best with the CT. Consequently its deformation field was used to deform the other MRIs. This improved the agreement between these MRIs and the CT over their individual registration. The deformation field was used for dose overlay on multi‐form MRIs.Conclusion: The deformable image registration tool developed successfully registers multi‐form MRIs to CT for treatment planning of prone breast radiotherapy.