Search

Purpose: In radiation therapy, a key step for a successful cancer treatment is image-based treatment planning. One objective of the planning phase is the fast and accurate segmentation of organs at risk and target structures from medical images. However, manual delineation of organs, which is still the gold standard in many clinical environments, is time-consuming and prone to inter-observer variations. Consequently, many automated segmentation methods have been developed., Methods: In this work, we train two hierarchical 3D neural networks to segment multiple organs at risk in the head and neck area. First, we train a coarse network on size-reduced medical images to locate the organs of interest. Second, a subsequent fine network on full-resolution images is trained for a final accurate segmentation. The proposed method is purely deep learning based; accordingly, no pre-registration or post-processing is required., Results: The approach has been applied on a publicly available computed tomography dataset, created for the MICCAI 2015 Auto-Segmentation challenge. In an extensive evaluation process, the best configurations for the trained networks have been determined. Compared to the existing methods, the presented approach shows state-of-the-art performance for the segmentation of seven different structures in the head and neck area., Conclusion: We conclude that 3D neural networks outperform the most existing model- and atlas-based methods for the segmentation of organs at risk in the head and neck area. The ease of use, high accuracy and the test time efficiency of the method make it promising for image-based treatment planning in clinical practice.

In the original version of this Article the values in the rightmost column of Table 1 were inadvertently shifted relative to the other columns. This has now been corrected in the PDF and HTML versions of the Article.

Purpose: To investigate advanced multimodal methods for pseudo-computed tomography (CT) generation from standard magnetic resonance imaging sequences and to validate the results by intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) plans. We present 2 novel methods that employ key techniques to enhance pseudo-CTs and investigate the effect on image quality and applicability for IMRT and VMAT planning., Materials and Methods: The data set contains CT and magnetic resonance image scans from 15 patients who underwent cranial radiation therapy. For each patient, pseudo-CTs of the head were generated with a patch-based and a voxel-based algorithm. The accuracy of the pseudo-CTs in comparison to clinical CTs was evaluated by mean absolute error, bias, and the Dice coefficient (of bone). IMRT and VMAT plans were created for each patient. Dose distributions were calculated with both the pseudo-CT and the clinical CT scans and compared by gamma tests, dose-volume histograms, and isocenter doses., Results: The generated pseudo-CTs exhibited average mean absolute errors of 118.7 ± 10.4 HU for the voxel-based algorithm and 73.0 ± 6.4 HU for the patch-based algorithm. The dose calculations were in good agreement and showed gamma test (2 mm, 2%) pass rates for both beam setups (IMRT and VMAT) of over 99% for 14 patients and over 98% for 1 patient., Conclusions: We showed that the key techniques of our 2 novel algorithms advance the quality of pseudo-CT significantly and generate very competitive pseudo-CTs compared with previously published methods. This quality was confirmed by low dose error in comparison to the ground-truth CT. With the achieved level of accuracy, our patch-based algorithm especially is a candidate for clinical routine use in IMRT and VMAT planning., (Copyright © 2018 Elsevier Inc. All rights reserved.)

An accurate prediction of brain tumor progression is crucial for optimized treatment of the tumors. Gliomas are primarily treated by combining surgery, external beam radiotherapy, and chemotherapy. Among them, radiotherapy is a non-invasive and effective therapy, and an understanding of tumor growth will allow better therapy planning. In particular, estimating parameters associated with tumor growth, such as the diffusion coefficient and proliferation rate, is crucial to accurately characterize physiology of tumor growth and to develop predictive models of tumor infiltration and recurrence. Accurate parameter estimation, however, is a challenging task due to inaccurate tumor boundaries and the approximation of the tumor growth model. Here, we introduce a Bayesian framework for a subject-specific tumor growth model that estimates the tumor parameters effectively. This is achieved by using an improved elliptical slice sampling method based on an adaptive sample region. Experimental results on clinical data demonstrate that the proposed method provides a higher acceptance rate, while preserving the parameter estimation accuracy, compared with other state-of-the-art methods such as Metropolis-Hastings and elliptical slice sampling without any modification. Our approach has the potential to provide a method to individualize therapy, thereby offering an optimized treatment.

Purpose: This work aims at investigating intensity corrected cone-beam x-ray computed tomography (CBCT) images for accurate dose calculation in adaptive intensity modulated proton therapy (IMPT) for prostate and head and neck (H&N) cancer. A deformable image registration (DIR)-based method and a scatter correction approach using the image data obtained from DIR as prior are characterized and compared on the basis of the same clinical patient cohort for the first time., Methods: Planning CT (pCT) and daily CBCT data (reconstructed images and measured projections) of four H&N and four prostate cancer patients have been considered in this study. A previously validated Morphons algorithm was used for DIR of the planning CT to the current CBCT image, yielding a so-called virtual CT (vCT). For the first time, this approach was translated from H&N to prostate cancer cases in the scope of proton therapy. The warped pCT images were also used as prior for scatter correction of the CBCT projections for both tumor sites. Single field uniform dose and IMPT (only for H&N cases) treatment plans have been generated with a research version of a commercial planning system. Dose calculations on vCT and scatter corrected CBCT (CBCT cor ) were compared by means of the proton range and a gamma-index analysis. For the H&N cases, an additional diagnostic replanning CT (rpCT) acquired within three days of the CBCT served as additional reference. For the prostate patients, a comprehensive contour comparison of CBCT and vCT, using a trained physician's delineation, was performed., Results: A high agreement of vCT and CBCT cor was found in terms of the proton range and gamma-index analysis. For all patients and indications between 95% and 100% of the proton dose profiles in beam's eye view showed a range agreement of better than 3 mm. The pass rate in a (2%,2 mm) gamma-comparison was between 96% and 100%. For H&N patients, an equivalent agreement of vCT and CBCT cor to the reference rpCT was observed. However, for the prostate cases, an insufficient accuracy of the vCT contours retrieved from DIR was found, while the CBCT cor contours showed very high agreement to the contours delineated on the raw CBCT., Conclusions: For H&N patients, no considerable differences of vCT and CBCT cor were found. For prostate cases, despite the high dosimetric agreement, the DIR yields incorrect contours, probably due to the more pronounced anatomical changes in the abdomen and the reduced soft-tissue contrast in the CBCT. Using the vCT as prior, these inaccuracies can be overcome and images suitable for accurate delineation and dose calculation in CBCT-based adaptive IMPT can be retrieved from scatter correction of the CBCT projections.

Purpose: To quantify and compare the impact of interfractional setup and anatomic variations on proton therapy (PT) and intensity modulated radiation therapy (IMRT) for prostate cancer., Methods and Materials: Twenty patients with low-risk or intermediate-risk prostate cancer randomized to receive passive-scattering PT (n=10) and IMRT (n=10) were selected. For both modalities, clinical treatment plans included 50.4 Gy(RBE) to prostate and proximal seminal vesicles, and prostate-only boost to 79.2 Gy(RBE) in 1.8 Gy(RBE) per fraction. Implanted fiducials were used for prostate localization and endorectal balloons were used for immobilization. Patients in PT and IMRT arms received weekly computed tomography (CT) and cone beam CT (CBCT) scans, respectively. The planned dose was recalculated on each weekly image, scaled, and mapped onto the planning CT using deformable registration. The resulting accumulated dose distribution over the entire treatment course was compared with the planned dose using dose-volume histogram (DVH) and γ analysis., Results: The target conformity index remained acceptable after accumulation. The largest decrease in the average prostate D98 was 2.2 and 0.7 Gy for PT and IMRT, respectively. On average, the mean dose to bladder increased by 3.26 ± 7.51 Gy and 1.97 ± 6.84 Gy for PT and IMRT, respectively. These values were 0.74 ± 2.37 and 0.56 ± 1.90 for rectum. Differences between changes in DVH indices were not statistically significant between modalities. All volume indices remained within the protocol tolerances after accumulation. The average pass rate for the γ analysis, assuming tolerances of 3 mm and 3%, for clinical target volume, bladder, rectum, and whole patient for PT/IMRT were 100/100, 92.6/99, 99.2/100, and 97.2/99.4, respectively., Conclusion: The differences in target coverage and organs at risk dose deviations for PT and IMRT were not statistically significant under the guidelines of this protocol., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Purpose: Severe (Radiation Therapy Oncology Group [RTOG] grade 3 or greater) esophagitis generally occurs in 15% to 25% of non-small cell lung cancer (NSCLC) patients undergoing concurrent chemotherapy and radiation therapy (CCRT), which may result in treatment breaks that compromise local tumor control and pose a barrier to dose escalation. Here, we report a novel contralateral esophagus-sparing technique (CEST) that uses intensity modulated radiation therapy (IMRT) to reduce the incidence of severe esophagitis., Methods and Materials: We reviewed consecutive patients with thoracic malignancies undergoing curative CCRT in whom CEST was used. The esophageal wall contralateral (CE) to the tumor was contoured as an avoidance structure, and IMRT was used to guide a rapid dose falloff gradient beyond the target volume in close proximity to the esophagus. Esophagitis was recorded based on the RTOG acute toxicity grading system., Results: We identified 20 consecutive patients treated with CCRT of at least 63 Gy in whom there was gross tumor within 1 cm of the esophagus. The median radiation dose was 70.2 Gy (range, 63-72.15 Gy). In all patients, ≥99% of the planning and internal target volumes was covered by ≥90% and 100% of prescription dose, respectively. Strikingly, no patient experienced grade ≥3 esophagitis (95% confidence limits, 0%-16%) despite the high total doses delivered. The median maximum dose, V45, and V55 of the CE were 60.7 Gy, 2.1 cc, and 0.4 cc, respectively, indicating effective esophagus cross-section sparing by CEST., Conclusion: We report a simple yet effective method to avoid exposing the entire esophagus cross-section to high doses. By using proposed CE dose constraints of V45 <2.5 cc and V55 <0.5 cc, CEST may improve the esophagus toxicity profile in thoracic cancer patients receiving CCRT even at doses above the standard 60- to 63-Gy levels. Prospective testing of CEST is warranted., (Copyright © 2015 Elsevier Inc. All rights reserved.)

Purpose: To investigate the spatial correlation between high uptake regions of 2-deoxy-2-[(18)F]-fluoro-D-glucose positron emission tomography ((18)F-FDG PET) before and after therapy in recurrent lung cancer., Methods and Materials: We enrolled 106 patients with inoperable lung cancer into a prospective study whose primary objectives were to determine first, the earliest time point when the maximum decrease in FDG uptake representing the maximum metabolic response (MMR) is attainable and second, the optimum cutoff value of MMR based on its predicted tumor control probability, sensitivity, and specificity. Of those patients, 61 completed the required 4 serial (18)F-FDG PET examinations after therapy. Nineteen of 61 patients experienced local recurrence at the primary tumor and underwent analysis. The volumes of interest (VOI) on pretherapy FDG-PET were defined by use of an isocontour at ≥50% of maximum standard uptake value (SUVmax) (≥50% of SUVmax) with correction for heterogeneity. The VOI on posttherapy images were defined at ≥80% of SUVmax. The VOI of pretherapy and posttherapy (18)F-FDG PET images were correlated for the extent of overlap., Results: The size of VOI at pretherapy images was on average 25.7% (range, 8.8%-56.3%) of the pretherapy primary gross tumor volume (GTV), and their overlap fractions were 0.8 (95% confidence interval [CI]: 0.7-0.9), 0.63 (95% CI: 0.49-0.77), and 0.38 (95% CI: 0.19-0.57) of VOI of posttherapy FDG PET images at 10 days, 3 months, and 6 months, respectively. The residual uptake originated from the pretherapy VOI in 15 of 17 cases., Conclusions: VOI defined by the SUVmax-≥50% isocontour may be a biological target volume for escalated radiation dose., (Copyright © 2014 Elsevier Inc. All rights reserved.)

Previous methods to estimate the inherent accuracy of deformable image registration (DIR) have typically been performed relative to a known ground truth, such as tracking of anatomic landmarks or known deformations in a physical or virtual phantom. In this study, we propose a new approach to estimate the spatial geometric uncertainty of DIR using statistical sampling techniques that can be applied to the resulting deformation vector fields (DVFs) for a given registration. The proposed DIR performance metric, the distance discordance metric (DDM), is based on the variability in the distance between corresponding voxels from different images, which are co-registered to the same voxel at location (X) in an arbitrarily chosen 'reference' image. The DDM value, at location (X) in the reference image, represents the mean dispersion between voxels, when these images are registered to other images in the image set. The method requires at least four registered images to estimate the uncertainty of the DIRs, both for inter- and intra-patient DIR. To validate the proposed method, we generated an image set by deforming a software phantom with known DVFs. The registration error was computed at each voxel in the 'reference' phantom and then compared to DDM, inverse consistency error (ICE), and transitivity error (TE) over the entire phantom. The DDM showed a higher Pearson correlation (Rp) with the actual error (Rp ranged from 0.6 to 0.9) in comparison with ICE and TE (Rp ranged from 0.2 to 0.8). In the resulting spatial DDM map, regions with distinct intensity gradients had a lower discordance and therefore, less variability relative to regions with uniform intensity. Subsequently, we applied DDM for intra-patient DIR in an image set of ten longitudinal computed tomography (CT) scans of one prostate cancer patient and for inter-patient DIR in an image set of ten planning CT scans of different head and neck cancer patients. For both intra- and inter-patient DIR, the spatial DDM map showed large variation over the volume of interest (the pelvis for the prostate patient and the head for the head and neck patients). The highest discordance was observed in the soft tissues, such as the brain, bladder, and rectum, due to higher variability in the registration. The smallest DDM values were observed in the bony structures in the pelvis and the base of the skull. The proposed metric, DDM, provides a quantitative tool to evaluate the performance of DIR when a set of images is available. Therefore, DDM can be used to estimate and visualize the uncertainty of intra- and/or inter-patient DIR based on the variability of the registration rather than the absolute registration error.

We present a novel method for inferring tissue labels in atlas-based image segmentation using Gaussian process regression. Atlas-based segmentation results in probabilistic label maps that serve as input to our method. We introduce a contour-driven prior distribution over label maps to incorporate image features of the input scan into the label inference problem. The mean function of the Gaussian process posterior distribution yields the MAP estimate of the label map and is used in the subsequent voting. We demonstrate improved segmentation accuracy when our approach is combined with two different patch-based segmentation techniques. We focus on the segmentation of parotid glands in CT scans of patients with head and neck cancer, which is important for radiation therapy planning.

Purpose: Breath-hold (BH) treatments can be used to reduce cardiac dose for patients with left-sided breast cancer and unfavorable cardiac anatomy. A surface imaging technique was developed for accurate patient setup and reproducible real-time BH positioning., Methods and Materials: Three-dimensional surface images were obtained for 20 patients. Surface imaging was used to correct the daily setup for each patient. Initial setup data were recorded for 443 fractions and were analyzed to assess random and systematic errors. Real time monitoring was used to verify surface placement during BH. The radiation beam was not turned on if the BH position difference was greater than 5 mm. Real-time surface data were analyzed for 2398 BHs and 363 treatment fractions. The mean and maximum differences were calculated. The percentage of BHs greater than tolerance was calculated., Results: The mean shifts for initial patient setup were 2.0 mm, 1.2 mm, and 0.3 mm in the vertical, longitudinal, and lateral directions, respectively. The mean 3-dimensional vector shift was 7.8 mm. Random and systematic errors were less than 4 mm. Real-time surface monitoring data indicated that 22% of the BHs were outside the 5-mm tolerance (range, 7%-41%), and there was a correlation with breast volume. The mean difference between the treated and reference BH positions was 2 mm in each direction. For out-of-tolerance BHs, the average difference in the BH position was 6.3 mm, and the average maximum difference was 8.8 mm., Conclusions: Daily real-time surface imaging ensures accurate and reproducible positioning for BH treatment of left-sided breast cancer patients with unfavorable cardiac anatomy., (Copyright © 2012 Elsevier Inc. All rights reserved.)

Purpose: To characterize and quantify multiorgan respiration-induced motion in the abdomen in liver and pancreatic cancer patients., Methods and Materials: Four-dimensional computed tomography scans were acquired for 18 patients treated for abdominal tumors. Contours of multiple abdominal organs were drawn by the radiation oncologist at one respiratory phase; these contours were propagated to other respiratory phases by deformable registration. Three-dimensional organ models were generated from the resulting contours at each phase. Motions of the bounding box and center of mass were extracted and analyzed for the clinical target volume and organs at risk., Results: On average, the center of mass motion for liver clinical target volumes was 9.7 mm (SD 5 mm) in the superior-inferior direction, with a range of 3 to 18 mm; for pancreatic tumors, the average was 5 mm (SD 1 mm) m with a range of 3 to 7 mm. Abdominal organs move in unison, but with varying amplitudes. Gating near exhale (T40-T60) reduces the range of motion by a factor of ∼10., Conclusion: We have used deformable registration to calculate the trajectories of abdominal organs in four dimensions, based on center of mass and bounding box motion metrics. Our results are compared with previously reported studies. Possible reasons for differences are discussed., (Copyright © 2012 Elsevier Inc. All rights reserved.)

Image registration is inherently ill-posed, and lacks a unique solution. In the context of medical applications, it is desirable to avoid solutions that describe physically unsound deformations within the patient anatomy. Among the accepted methods of regularizing non-rigid image registration to provide solutions applicable to medical practice is the penalty of thin-plate bending energy. In this paper, we develop an exact, analytic method for computing the bending energy of a three-dimensional B-spline deformation field as a quadratic matrix operation on the spline coefficient values. Results presented on ten thoracic case studies indicate the analytic solution is between 61-1371x faster than a numerical central differencing solution.

Purpose: We compared four-dimensional (4D) layer-stacking and conventional carbon ion beam distribution in the treatment of lung cancer between ungated and gated respiratory strategies using 4DCT data sets., Methods and Materials: Twenty lung patients underwent 4DCT imaging under free-breathing conditions. Using planning target volumes (PTVs) at respective respiratory phases, two types of compensating bolus were designed, a full single respiratory cycle for the ungated strategy and an approximately 30% duty cycle for the exhalation-gated strategy. Beams were delivered to the PTVs for the ungated and gated strategies, PTV(ungated) and PTV(gated), respectively, which were calculated by combining the respective PTV(Tn)s by layer-stacking and conventional irradiation. Carbon ion beam dose distribution was calculated as a function of respiratory phase by applying a compensating bolus to 4DCT. Accumulated dose distributions were calculated by applying deformable registration., Results: With the ungated strategy, accumulated dose distributions were satisfactorily provided to the PTV, with D95 values for layer-stacking and conventional irradiation of 94.0% and 96.2%, respectively. V20 for the lung and Dmax for the spinal cord were lower with layer-stacking than with conventional irradiation, whereas Dmax for the skin (14.1 GyE) was significantly lower (21.9 GyE). In addition, dose conformation to the GTV/PTV with layer-stacking irradiation was better with the gated than with the ungated strategy., Conclusions: Gated layer-stacking irradiation allows the delivery of a carbon ion beam to a moving target without significant degradation of dose conformity or the development of hot spots., (Copyright © 2011 Elsevier Inc. All rights reserved.)

For radiotherapy planning, contouring of target volume and healthy structures at risk in CT volumes is essential. To automate this process, one of the available segmentation techniques can be used for many thoracic organs except the esophagus, which is very hard to segment due to low contrast. In this work we propose to initialize our previously introduced model based 3D level set esophagus segmentation method with a principal curve tracing (PCT) algorithm, which we adapted to solve the esophagus centerline detection problem. To address challenges due to low intensity contrast, we enhanced the PCT algorithm by learning spatial and intensity priors from a small set of annotated CT volumes. To locate the esophageal wall, the model based 3D level set algorithm including a shape model that represents the variance of esophagus wall around the estimated centerline is utilized. Our results show improvement in esophagus segmentation when initialized by PCT compared to our previous work, where an ad hoc centerline initialization was performed. Unlike previous approaches, this work does not need a very large set of annotated training images and has similar performance.

Purpose: In proton therapy, uncertainty in the location of the distal dose edge can lead to cautious treatment plans that reduce the dosimetric advantage of protons. After radiation exposure, vertebral bone marrow undergoes fatty replacement that is visible on magnetic resonance imaging (MRI). This presents an exciting opportunity to observe radiation dose distribution in vivo. We used quantitative spine MRI changes to precisely detect the distal dose edge in proton radiation patients., Methods and Materials: We registered follow-up T1-weighted MRI images to planning computed tomography scans from 10 patients who received proton spine irradiation. A radiation dose-MRI signal intensity curve was created using the lateral beam penumbra in the sacrum. This curve was then used to measure range errors in the lumbar spine., Results: In the lateral penumbra, there was an increase in signal intensity with higher dose throughout the full range of 0-37.5 Gy (RBE). In the distal fall-off region, the beam sometimes appeared to penetrate farther than planned. The mean overshoot in 10 patients was 1.9 mm (95% confidence interval, 0.8-3.1 mm), on the order of the uncertainties inherent to our range verification method., Conclusions: We have demonstrated in vivo proton range verification using posttreatment spine MRI changes. Our analysis suggests the presence of a systematic overshoot of a few millimeters in some proton spine treatments, but the range error does not exceed the uncertainty incorporated into the treatment planning margin. It may be possible to extend our technique to MRI sequences that show early bone marrow changes, enabling adaptive treatment modification., (Copyright (c) 2010 Elsevier Inc. All rights reserved.)

In this paper we propose a supervised 3D segmentation algorithm to locate the esophagus in thoracic CT scans using a variational framework. To address challenges due to low contrast, several priors are learned from a training set of segmented images. Our algorithm first estimates the centerline based on a spatial model learned at a few manually marked anatomical reference points. Then an implicit shape model is learned by subtracting the centerline and applying PCA to these shapes. To allow local variations in the shapes, we propose to use nonlinear smooth local deformations. Finally, the esophageal wall is located within a 3D level set framework by optimizing a cost function including terms for appearance, the shape model, smoothness constraints and an air/contrast model.

Purpose: We compared respiratory-gated and respiratory-ungated treatment strategies using four-dimensional (4D) scattered carbon ion beam distribution in pancreatic 4D computed tomography (CT) datasets., Methods and Materials: Seven inpatients with pancreatic tumors underwent 4DCT scanning under free-breathing conditions using a rapidly rotating cone-beam CT, which was integrated with a 256-slice detector, in cine mode. Two types of bolus for gated and ungated treatment were designed to cover the planning target volume (PTV) using 4DCT datasets in a 30% duty cycle around exhalation and a single respiratory cycle, respectively. Carbon ion beam distribution for each strategy was calculated as a function of respiratory phase by applying the compensating bolus to 4DCT at the respective phases. Smearing was not applied to the bolus, but consideration was given to drill diameter. The accumulated dose distributions were calculated by applying deformable registration and calculating the dose-volume histogram., Results: Doses to normal tissues in gated treatment were minimized mainly on the inferior aspect, which thereby minimized excessive doses to normal tissues. Over 95% of the dose, however, was delivered to the clinical target volume at all phases for both treatment strategies. Maximum doses to the duodenum and pancreas averaged across all patients were 43.1/43.1 GyE (ungated/gated) and 43.2/43.2 GyE (ungated/gated), respectively., Conclusions: Although gated treatment minimized excessive dosing to normal tissue, the difference between treatment strategies was small. Respiratory gating may not always be required in pancreatic treatment as long as dose distribution is assessed. Any application of our results to clinical use should be undertaken only after discussion with oncologists, particularly with regard to radiotherapy combined with chemotherapy.

Purpose: To quantify pancreas and pancreatic tumor movement due to respiratory motion using volumetric cine CT images., Materials and Methods: Six patients with pancreatic tumors were scanned in cine mode with a 256 multi-slice CT scanner under free breathing conditions. Gross tumor volume (GTV) and pancreas were manually contoured on the CT data set by a radiation oncologist. Intrafractional respiratory movement of the GTV and pancreas was calculated, and the results were compared between the respiratory ungated and gated phases, which is a 30% duty cycle around exhalation., Results: Respiratory-induced organ motion was observed mainly in the anterior abdominal side than the posterior side. Average GTV displacement (ungated/gated phases) was 0.7 mm/0.2mm in both the left and right directions, and 2.5mm/0.9 mm in the anterior, 0.1 mm/0mm in the posterior, and 8.9 mm/2.6mm in the inferior directions. Average pancreas center of mass displacement relative to that at peak exhalation was mainly in the inferior direction, at 9.6mm in the ungated phase and 2.3mm in the gated phase., Conclusions: By allowing accurate determination of the margin, quantitative analysis of tumor and pancreas displacement provides useful information in treatment planning in all radiation approaches for pancreatic tumors.