Segmentation of 71 Anatomical Structures Necessary for the Evaluation of Guideline-Conforming Clinical Target Volumes in Head and Neck Cancers.

The delineation of the clinical target volumes (CTVs) for radiation therapy is time-consuming, requires intensive training and shows high inter-observer variability. Supervised deep-learning methods depend heavily on consistent training data; thus, State-of-the-Art research focuses on making CTV labels more homogeneous and strictly bounding them to current standards. International consensus expert guidelines standardize CTV delineation by conditioning the extension of the clinical target volume on the surrounding anatomical structures. Training strategies that directly follow the construction rules given in the expert guidelines or the possibility of quantifying the conformance of manually drawn contours to the guidelines are still missing. Seventy-one anatomical structures that are relevant to CTV delineation in head- and neck-cancer patients, according to the expert guidelines, were segmented on 104 computed tomography scans, to assess the possibility of automating their segmentation by State-of-the-Art deep learning methods. All 71 anatomical structures were subdivided into three subsets of non-overlapping structures, and a 3D nnU-Net model with five-fold cross-validation was trained for each subset, to automatically segment the structures on planning computed tomography scans. We report the DICE, Hausdorff distance and surface DICE for 71 + 5 anatomical structures, for most of which no previous segmentation accuracies have been reported. For those structures for which prediction values have been reported, our segmentation accuracy matched or exceeded the reported values. The predictions from our models were always better than those predicted by the TotalSegmentator. The sDICE with 2 mm margin was larger than 80% for almost all the structures. Individual structures with decreased segmentation accuracy are analyzed and discussed with respect to their impact on the CTV delineation following the expert guidelines. No deviation is expected to affect the rule-based automation of the CTV delineation.

Essential parameters needed for a U-Net-based segmentation of individual bones on planning CT images in the head and neck region using limited datasets for radiotherapy application.

Objective.The field of radiotherapy is highly marked by the lack of datasets even with the availability of public datasets. Our study uses a very limited dataset to provide insights on essential parameters needed to automatically and accurately segment individual bones on planning CT images of head and neck cancer patients.Approach.The study was conducted using 30 planning CT images of real patients acquired from 5 different cohorts. 15 cases from 4 cohorts were randomly selected as training and validation datasets while the remaining were used as test datasets. Four experimental sets were formulated to explore parameters such as background patch reduction, class-dependent augmentation and incorporation of a weight map on the loss function.Main results.Our best experimental scenario resulted in a mean Dice score of 0.93 ± 0.06 for other bones (skull, mandible, scapulae, clavicles, humeri and hyoid), 0.93 ± 0.02 for ribs and 0.88 ± 0.03 for vertebrae on 7 test cases from the same cohorts as the training datasets. We compared our proposed solution approach to a retrained nnU-Net and obtained comparable results for vertebral bones while outperforming in the correct identification of the left and right instances of ribs, scapulae, humeri and clavicles. Furthermore, we evaluated the generalization capability of our proposed model on a new cohort and the mean Dice score yielded 0.96 ± 0.10 for other bones, 0.95 ± 0.07 for ribs and 0.81 ± 0.19 for vertebrae on 8 test cases.Significance.With these insights, we are challenging the utilization of an automatic and accurate bone segmentation tool into the clinical routine of radiotherapy despite the limited training datasets.

Biofidelic image registration for head and neck region utilizing an in-silico articulated skeleton as a transformation model.

Objective.We propose an integration scheme for a biomechanical motion model into a deformable image registration. We demonstrate its accuracy and reproducibility for adaptive radiation therapy in the head and neck region.Approach. The novel registration scheme for the bony structures in the head and neck regions is based on a previously developed articulated kinematic skeleton model. The realized iterative single-bone optimization process directly triggers posture changes of the articulated skeleton, exchanging the transformation model within the deformable image registration process. Accuracy in terms of target registration errors in the bones is evaluated for 18 vector fields of three patients between each planning CT and six fraction CT scans distributed along the treatment course.Main results. The median of target registration error distribution of the landmark pairs is 1.4 ± 0.3 mm. This is sufficient accuracy for adaptive radiation therapy. The registration performs equally well for all three patients and no degradation of the registration accuracy can be observed throughout the treatment.Significance. Deformable image registration, despite its known residual uncertainties, is until now the tool of choice towards online re-planning automation. By introducing a biofidelic motion model into the optimization, we provide a viable way towards an in-build quality assurance.

Analyzing human decisions in IGRT of head-and-neck cancer patients to teach image registration algorithms what experts know.

BACKGROUND: In IGRT of deformable head-and-neck anatomy, patient setup corrections are derived by rigid registration methods. In practice, experienced radiation therapists often correct the resulting vectors, thus indicating a different prioritization of alignment of local structures. Purpose of this study is to transfer the knowledge experts apply when correcting the automatically generated result (pre-match) to automated registration. METHODS: Datasets of 25 head-and-neck-cancer patients with daily CBCTs and corresponding approved setup correction vectors were analyzed. Local similarity measures were evaluated to identify the criteria for human corrections with regard to alignment quality, analogous to the radiomics approach. Clustering of similarity improvement patterns is applied to reveal priorities in the alignment quality. RESULTS: The radiation therapists prioritized to align the spinal cord closest to the high-dose area. Both target volumes followed with second and third highest priority. The bony pre-match influenced the human correction along the crania-caudal axis. Based on the extracted priorities, a new rigid registration procedure is constructed which is capable of reproducing the corrections of experts. CONCLUSIONS: The proposed approach extracts knowledge of experts performing IGRT corrections to enable new rigid registration methods that are capable of mimicking human decisions. In the future, the deduction of knowledge-based corrections for different cohorts can be established automating such supervised learning approaches.

Construction of a biomechanical head and neck motion model as a guide to evaluation of deformable image registration.

The use of deformable image registration methods in the context of adaptive radiotherapy leads to uncertainties in the simulation of the administered dose distributions during the treatment course. Evaluation of these methods is a prerequisite to decide if a plan adaptation will improve the individual treatment. Current approaches using manual references limit the validity of evaluation, especially for low-contrast regions. In particular, for the head and neck region, the highly flexible anatomy and low soft tissue contrast in control images pose a challenge to image registration and its evaluation. Biomechanical models promise to overcome this issue by providing anthropomorphic motion modelling of the patient. We introduce a novel biomechanical motion model for the generation and sampling of different postures of the head and neck anatomy. Motion propagation behaviour of the individual bones is defined by an underlying kinematic model. This model interconnects the bones by joints and thus is capable of providing a wide range of motion. Triggered by the motion of the individual bones, soft tissue deformation is described by an extended heterogeneous tissue model based on the chainmail approach. This extension, for the first time, allows the propagation of decaying rotations within soft tissue without the necessity for explicit tissue segmentation. Overall motion simulation and sampling of deformed CT scans including a basic noise model is achieved within 30 s. The proposed biomechanical motion model for the head and neck site generates displacement vector fields on a voxel basis, approximating arbitrary anthropomorphic postures of the patient. It was developed with the intention of providing input data for the evaluation of deformable image registration.

Comparison of Safety Margin Generation Concepts in Image Guided Radiotherapy to Account for Daily Head and Neck Pose Variations.

PURPOSE: Intensity modulated radiation therapy (IMRT) of head and neck tumors allows a precise conformation of the high-dose region to clinical target volumes (CTVs) while respecting dose limits to organs a risk (OARs). Accurate patient setup reduces translational and rotational deviations between therapy planning and therapy delivery days. However, uncertainties in the shape of the CTV and OARs due to e.g. small pose variations in the highly deformable anatomy of the head and neck region can still compromise the dose conformation. Routinely applied safety margins around the CTV cause higher dose deposition in adjacent healthy tissue and should be kept as small as possible. MATERIALS AND METHODS: In this work we evaluate and compare three approaches for margin generation 1) a clinically used approach with a constant isotropic 3 mm margin, 2) a previously proposed approach adopting a spatial model of the patient and 3) a newly developed approach adopting a biomechanical model of the patient. All approaches are retrospectively evaluated using a large patient cohort of over 500 fraction control CT images with heterogeneous pose changes. Automatic methods for finding landmark positions in the control CT images are combined with a patient specific biomechanical finite element model to evaluate the CTV deformation. RESULTS: The applied methods for deformation modeling show that the pose changes cause deformations in the target region with a mean motion magnitude of 1.80 mm. We found that the CTV size can be reduced by both variable margin approaches by 15.6% and 13.3% respectively, while maintaining the CTV coverage. With approach 3 an increase of target coverage was obtained. CONCLUSION: Variable margins increase target coverage, reduce risk to OARs and improve healthy tissue sparing at the same time.

The frequency of re-planning and its variability dependent on the modification of the re-planning criteria and IGRT correction strategy in head and neck IMRT.

BACKGROUND: To analyse the frequency of re-planning and its variability dependent on the IGRT correction strategy and on the modification of the dosimetric criteria for re-planning for the spinal cord in head and neck IG-IMRT. METHODS: Daily kV-control-CTs of six head and neck patients (=175 CTs) were analysed. All volumes of interest were re-contoured using deformable image registration. Three IGRT correction strategies were simulated and the resulting dose distributions were computed for all fractions. Different sets of criteria with varying dose thresholds for re-planning were investigated. All sets of criteria ensure equivalent target coverage of both CTVs, but vary in the tolerance threshold of the spinal cord. RESULTS: The variations of the D95 and D2 in respect to the planned values ranged from -7% to +3% for both CTVs, and -2% to +6% for the spinal cord. Despite different correction vectors of the three IGRT strategies, the dosimetric differences were small. The number of fractions not requiring re-planning varied between 0% and 11% dependent on the applied IGRT correction strategy. In contrast, this number ranged between 32% and 70% dependent on the dosimetric thresholds, even though these thresholds were only gently modified. CONCLUSIONS: The more precise the planned dose needs to be maintained over the treatment course, the more frequently re-planning is required. The influence of different IGRT correction strategies, even though geometrically notable, was found to be of only limited relevance for the re-planning frequency. In contrast, the definition and modification of thresholds for re-planning have a major impact on the re-planning frequency.

Accuracy quantification of a deformable image registration tool applied in a clinical setting.

The purpose of this study was to test the accuracy of a commercially available deformable image registration tool in a clinical situation. In addition, to demonstrate a method to evaluate the resulting transformation of such a tool to a reference defined by multiple experts. For 16 patients (seven head and neck, four thoracic, five abdominal), 30-50 anatomical landmarks were defined on recognizable spots of a planning CT and a corresponding fraction CT. A commercially available deformable image registration tool, Velocity AI, was used to align all fraction CTs with the respective planning CTs. The registration accuracy was quantified by means of the target registration error in respect to expert-defined landmarks, considering the interobserver variation of five observers. The interobserver uncertainty of the landmark definition in our data sets is found to be 1.2 ± 1.1 mm. In general the deformable image registration tool decreases the extent of observable misalignments from 4-8 mm to 1-4 mm for nearly 50% of the landmarks (to 77% in sum). Only small differences are observed in the alignment quality of scans with different tumor location. Smallest residual deviations were achieved in scans of the head and neck region (79%, ≤ 4 mm) and the thoracic cases (79%, ≤ 4 mm), followed by the abdominal cases (59%, ≤ 4 mm). No difference is observed in the alignment quality of different tissue types (bony vs. soft tissue). The investigated commercially available deformable image registration tool is capable of reducing a mean target registration error to a level that is clinically acceptable for the evaluation of retreatment plans and replanning in case of gross tumor change during treatment. Yet, since the alignment quality needs to be improved further, the individual result of the deformable image registration tool has still to be judged by the physician prior to application.

An optimised IGRT correction vector determined from a displacement vector field: a proof of principle of a decision-making aid for re-planning.

BACKGROUND: To present a new method that determines an optimised IGRT couch correction vector from a displacement vector field (DVF). The DVF is computed by a deformable image registration (DIR) method. The proposed method can improve the quality of volume-of-interest (VOI) alignment in image guided radiation therapy (IGRT), and can serve as a decision-making aid for re-planning. MATERIAL AND METHODS: The proposed method was demonstrated using the CT data sets of 11 head-and-neck cancer patients with daily kilovoltage control-CTs. A DVF was computed for each control-CT using a DIR method. The DVF was used for voxel tracking and re-contouring of the VOIs in the control-CTs. Then a rigid body transformation, which could be used as couch correction vector, was optimised. The aim of the optimisation process was to find a vector and rotations that map the deformed VOIs into a specified territory. This territory was defined by a margin extension of the VOIs at the time of the planning process. Within this extension, VOI motion and deformation was tolerated. The objective function in the optimisation process was the sum of all volume fractions outside the defined territories. RESULTS: The proposed method was able to find a correction vector, which resulted in a coverage of the target volumes of at least 98% in 52.3% of all fractions. In contrast, a standard IGRT correction using a rigid registration method only fulfilled this criterion in 22.6% of all fractions. The optimisation process took an average of 1.5 minutes per fraction. CONCLUSION: The knowledge of the deformation of the anatomy allows the determination of an optimised rigid correction vector using our method. The method ensures controlled mapping of the VOIs despite small deformations. If no optimised vector can be determined, re-planning should be considered. Thus, our method can also serve as a decision-making aid for re-planning.

A quantification strategy for missing bone mass in case of osteolytic bone lesions.

PURPOSE: Most of the patients who died of breast cancer have developed bone metastases. To understand the pathogenesis of bone metastases and to analyze treatment response of different bone remodeling therapies, preclinical animal models are examined. In breast cancer, bone metastases are often bone destructive. To assess treatment response of bone remodeling therapies, the volumes of these lesions have to be determined during the therapy process. The manual delineation of missing structures, especially if large parts are missing, is very time-consuming and not reproducible. Reproducibility is highly important to have comparable results during the therapy process. Therefore, a computerized approach is needed. Also for the preclinical research, a reproducible measurement of the lesions is essential. Here, the authors present an automated segmentation method for the measurement of missing bone mass in a preclinical rat model with bone metastases in the hind leg bones based on 3D CT scans. METHODS: The affected bone structure is compared to a healthy model. Since in this preclinical rat trial the metastasis only occurs on the right hind legs, which is assured by using vessel clips, the authors use the left body side as a healthy model. The left femur is segmented with a statistical shape model which is initialised using the automatically segmented medullary cavity. The left tibia and fibula are segmented using volume growing starting at the tibia medullary cavity and stopping at the femur boundary. Masked images of both segmentations are mirrored along the median plane and transferred manually to the position of the affected bone by rigid registration. Affected bone and healthy model are compared based on their gray values. If the gray value of a voxel indicates bone mass in the healthy model and no bone in the affected bone, this voxel is considered to be osteolytic. RESULTS: The lesion segmentations complete the missing bone structures in a reasonable way. The mean ratio vr∕vm of the reconstructed bone volume vr and the healthy model bone volume vm is 1.07, which indicates a good reconstruction of the modified bone. CONCLUSIONS: The qualitative and quantitative comparison of manual and semi-automated segmentation results have shown that comparing a modified bone structure with a healthy model can be used to identify and measure missing bone mass in a reproducible way.

Regional cumulative maximum dose to the spinal cord in head-and-neck cancer: considerations for re-irradiation.

PURPOSE: To present a new method that assesses the delivered maximum dose of different spinal cord sections in head-and-neck cancer treated with intensity-modulated radiation therapy (IMRT). This allows a more accurate estimation of the remaining cord dose tolerance in case of a later re-irradiation treatment planning. MATERIALS AND METHODS: The suggested workflow is demonstrated using daily acquired kilo-voltage control-CTs of four head-and-neck cancer patients (118 control-CTs). The local maximum dose inside different cord levels is determined and accumulated for the planning situation and over the treatment course for an IGRT and a non-IGRT approach. RESULTS: The approach is suitable to accurately detect and document the delivered maximum dose dependent on the cord levels. The delivered maximum dose differed up to 13% from the planned one in all sections due to setup uncertainties and the applied correction strategy. CONCLUSION: The presented approach facilitates later re-irradiation treatment planning due to detailed documentation of the delivered maximum dose to the spinal cord levels in the primary IMRT. The method also facilitates the interpretation of complex 3D dose information by reducing it to its essentials. This 2D illustration is an aid to orientation for the physician in the re-irradiation planning process.

IGRT versus non-IGRT for postoperative head-and-neck IMRT patients: dosimetric consequences arising from a PTV margin reduction.

BACKGROUND: To evaluate the impact of image-guided radiation therapy (IGRT) versus non-image-guided radiation therapy (non-IGRT) on the dose to the clinical target volume (CTV) and the cervical spinal cord during fractionated intensity-modulated radiation therapy (IMRT) for head-and-neck cancer (HNC) patients. MATERIAL AND METHODS: For detailed investigation, 4 exemplary patients with daily control-CT scans (total 118 CT scans) were analyzed. For the IGRT approach a target point correction (TPC) derived from a rigid registration focused to the high-dose region was used. In the non-IGRT setting, instead of a TPC, an additional cohort-based safety margin was applied. The dose distributions of the CTV and spinal cord were calculated on each control-CT and the resulting dose volume histograms (DVHs) were compared with the planned ones fraction by fraction. The D50 and D98 values for the CTV and the D5 values of the spinal cord were additionally reported. RESULTS: In general, the D50 and D98 histograms show no remarkable difference between both strategies. Yet, our detailed analysis also reveals differences in individual dose coverage worth inspection. Using IGRT, the D5 histograms show that the spinal cord less frequently receives a higher dose than planned compared to the non-IGRT setting. This effect is even more pronounced when looking at the curve progressions of the respective DVHs. CONCLUSIONS: Both approaches are equally effective in maintaining CTV coverage. However, IGRT is beneficial in spinal cord sparing. The use of an additional margin in the non-IGRT approach frequently results in a higher dose to the spinal cord than originally planned. This implies that a margin reduction combined with an IGRT correction helps to maintain spinal cord dose sparing best as possible. Yet, a detailed analysis of the dosimetric consequences dependent on the used strategy is required, to detect single fractions with unacceptable dosimetric deviations.

Local setup reproducibility of the spinal column when using intensity-modulated radiation therapy for craniospinal irradiation with patient in supine position.

PURPOSE: To evaluate local positioning errors of the lumbar spine during fractionated intensity-modulated radiotherapy of patients treated with craniospinal irradiation and to assess the impact of rotational error correction on these uncertainties for one patient setup correction strategy. METHODS AND MATERIALS: 8 patients (6 adults, 2 children) treated with helical tomotherapy for craniospinal irradiation were retrospectively chosen for this analysis. Patients were immobilized with a deep-drawn Aquaplast head mask. Additionally to daily megavoltage control computed tomography scans of the skull, once-a-week positioning of the lumbar spine was assessed. Therefore, patient setup was corrected by a target point correction, derived from a registration of the patient's skull. The residual positioning variations of the lumbar spine were evaluated applying a rigid-registration algorithm. The impact of different rotational error corrections was simulated. RESULTS: After target point correction, residual local positioning errors of the lumbar spine varied considerably. Craniocaudal axis rotational error correction did not improve or deteriorate these translational errors, whereas simulation of a rotational error correction of the right-left and anterior-posterior axis increased these errors by a factor of 2 to 3. CONCLUSION: The patient fixation used allows for deformations between the patient's skull and spine. Therefore, for the setup correction strategy evaluated in this study, generous margins for the lumbar spinal target volume are needed to prevent a local geographic miss. With any applied correction strategy, it needs to be evaluated whether or not a rotational error correction is beneficial.

Local setup errors in image-guided radiotherapy for head and neck cancer patients immobilized with a custom-made device.

PURPOSE: To evaluate the local positioning uncertainties during fractionated radiotherapy of head-and-neck cancer patients immobilized using a custom-made fixation device and discuss the effect of possible patient correction strategies for these uncertainties. METHODS AND MATERIALS: A total of 45 head-and-neck patients underwent regular control computed tomography scanning using an in-room computed tomography scanner. The local and global positioning variations of all patients were evaluated by applying a rigid registration algorithm. One bounding box around the complete target volume and nine local registration boxes containing relevant anatomic structures were introduced. The resulting uncertainties for a stereotactic setup and the deformations referenced to one anatomic local registration box were determined. Local deformations of the patients immobilized using our custom-made device were compared with previously published results. Several patient positioning correction strategies were simulated, and the residual local uncertainties were calculated. RESULTS: The patient anatomy in the stereotactic setup showed local systematic positioning deviations of 1-4 mm. The deformations referenced to a particular anatomic local registration box were similar to the reported deformations assessed from patients immobilized with commercially available Aquaplast masks. A global correction, including the rotational error compensation, decreased the remaining local translational errors. Depending on the chosen patient positioning strategy, the remaining local uncertainties varied considerably. CONCLUSIONS: Local deformations in head-and-neck patients occur even if an elaborate, custom-made patient fixation method is used. A rotational error correction decreased the required margins considerably. None of the considered correction strategies achieved perfect alignment. Therefore, weighting of anatomic subregions to obtain the optimal correction vector should be investigated in the future.

Quantitative assessment of image-guided radiotherapy for paraspinal tumors.

PURPOSE: To evaluate stereotactic positioning uncertainties of patients with paraspinal tumors treated with fractionated intensity-modulated radiotherapy; and to determine whether target-point correction via rigid registration is sufficient for daily patient positioning. PATIENTS AND METHODS: Forty-five patients with tumors at the cervical, thoracic, and lumbar spine received regular control computed-tomography (CT) scans using an in-room CT scanner. All patients were immobilized with the combination of Scotch cast torso and head masks. The positioning was evaluated regarding translational and rotational errors by applying a rigid registration algorithm based on mutual information. The registration box was fitted to the target volume for optimal registration in the high-dose area. To evaluate the suitability of the rigid registration result for correcting the target volume position we subsequently registered three small subsections of the upper, middle, and lower target volume. The resulting residual deviations reflect the extent of the elastic deformations, which cannot be covered by the rigid-body registration procedure. RESULTS: A total of 321 control CT scans were evaluated. The rotational errors were negligible. Translational errors were smallest for cervical tumors (-0.1 +/- 1.1, 0.3 +/- 0.8, and 0.1 +/- 0.9 mm along left-right, anterior-posterior, and superior-inferior axes), followed by thoracic (0.8 +/- 1.1, 0.3 +/- 0.8, and 1.1 +/- 1.3 mm) and lumbar tumors (-0.7 +/- 1.3, 0.0 +/- 0.9, and 0.5 +/- 1.6 mm). The residual deviations of the three subsections were <1 mm. CONCLUSIONS: The applied stereotactic patient setup resulted in small rotational errors. However, considerable translational positioning errors may occur; thus, on the basis of these data daily control CT scans are recommended. Rigid transformation is adequate for correcting the target volume position.