Assessment of a commercially available automatic deformable registration system

Evaluation of DIR algorithm performance in real patients for radiotherapy treatments: A systematic review of operator-dependent strategies

PURPOSE

The performance of deformable medical image registration (DIR) algorithms has become a major concern.

METHODS

We aimed to obtain updated information on DIR algorithm performance quantification through a literature review of articles published between 2010 and 2022. We focused only on studies using operator-based methods to treat real patients. The PubMed, Google Scholar and Embase databases were searched following PRISMA guidelines.

RESULTS

One hundred and seven articles were identified. The mean number of patients and registrations per publication was 20 and 63, respectively. We found 23 different geometric metrics appearing at least twice, and the dosimetric impact of DIR was quantified in 32 articles. Forty-eight different at-risk organs were described, and target volumes were studied in 43 publications. Prostate, head-and-neck and thoracic locations represented more than ¾ of the studied locations. We summarized the type of DIR and the images used, and other key elements. Intra/interobserver variability, threshold values and the correlation between metrics were also discussed.

CONCLUSIONS

This literature review covers the past decade and should facilitate the implementation of DIR algorithms in clinical practice by providing practical and pertinent information to quantify their performance on real patients.

Free-to-use DIR solutions in radiotherapy: Benchmark against commercial platforms through a contour-propagation study

PURPOSE

A contour propagation study has been conducted to benchmark three algorithms for Deformable Image Registration (DIR) freely available online against well-established commercial solutions.

METHODS

ElastiX, BRAINS and Plastimach, available as modules in the open source platform 3DSlicer, were tested as the recent AAPM Task group 132 guidelines proposes. The overlap of the DIR-mapped ROIs in four computational anthropomorphic phantoms was measured. To avoid bias every algorithm was left to run without any human interaction nor particular registration strategy. The accuracy of the algorithms was measured using the Dice Similarity Coefficient (DSC) and Mean Distance to Conformity (MDC) metrics. The registration quality was compared to the recommended geometrical accuracy suggested by AAPM TG132 and to the results of a large population-based study performed with commercial DIR solutions.

RESULTS

The considered free-to-use DIR solutions generally meet acceptable accuracy and good overlap (DSC > 0.85). Mild failures (DSC < 0.75) were detected only for the smallest structures. In case of extremely severe deformations acceptable accuracy was not met (MDC > 3 mm). The morphing capability of the tested algorithms equals those of commercial systems when the user interaction is avoided. Underperformances were detected only in cases where a specific registration strategy is mandatory to obtain a satisfying match.

CONCLUSIONS

All of the considered algorithms show performances not inferior to previously published data and have the potential to be good candidates for use in the clinical routine. The results and conclusions only apply to the considered phantoms and should not be considered to be generally applicable and extendable to patient cases.

Technical Note: A method for quality assurance of landmark sets for use in evaluation of deformable image registration accuracy of lung parenchyma

PURPOSE

To develop a quality control method to improve the accuracy of corresponding landmark sets used for deformable image registration (DIR) evaluation in the lung parenchyma.

METHODS

An iterative workflow was developed as a method for quality assurance of landmark sets. Starting with the initial landmark set for a given image pair, a landmark-based deformation was applied to one of the images. A difference image and a color overlay were generated using the deformed image and the other image of the pair. Inspection of these generated images at locations of landmarks allowed for the identification of misplaced landmarks. The observer responsible for creating the initial landmark set was tasked with review and revision of points flagged by the quality assurance procedure. Using the updated landmark sets, the process was repeated until all points were acceptable to the reviewer.

RESULTS

Eighteen landmark sets, containing a mean (SD) of 170 (31) landmarks, were created using CT images from non-small cell lung cancer patients exhibiting large geometric changes and atelectasis resolution, making landmark specification challenging. Following the quality assurance procedure, the final landmark sets contained a mean (SD) of 165 (25) landmarks, as points too difficult to match were removed and points were added to regions deficient in landmarks. For landmark sets in which changes were made, maximum and mean differences in landmark positions before and after quality assurance ranged between 8.7-81.5 mm and 0.3-9.6 mm, respectively.

CONCLUSIONS

An effective method for improving the accuracy of landmark correspondence was presented. This quality assurance approach enables more accurate evaluation of DIR for lung parenchyma in clinical image pairs in the absence of a ground truth deformation and may be applicable to other feature-rich anatomical sites.

Characterization of a new physical phantom for testing rigid and deformable image registration

The purpose of this study was to describe a new user-friendly, low-cost phantom that was developed to test the accuracy of rigid and deformable image registration (DIR) systems and to demonstrate the functional efficacy of the new phantom. The phantom was constructed out of acrylic and includes a variety of inserts that simulate different tissue shapes and properties. It can simulate deformations and location changes in patient anatomy by changing the rotations of both the phantom and the inserts. CT scans of this phantom were obtained and used to test the rigid and deformable registration accuracy of the Velocity software. Eight rotation and translation scenarios were used to test the rigid registration accuracy, and 11 deformation scenarios were used to test the DIR accuracy. The mean rotation accuracies in the X-Y (axial) and X-Z (coronal) planes were 0.50° and 0.13°, respectively. The mean translation accuracy was 1 mm in both the X and Y direction and was tested in soft tissue and bone. The DIR accuracies for soft tissue and bone were 0.93 (mean Dice similarity coefficient), 8.3 and 4.5 mm (mean Hausdouff distance), 0.95 and 0.79 mm (mean distance), and 1.13 and 1.12 (mean volume ratio) for soft tissue content (DTE oil) and bone, respectively. The new phantom has a simple design and can be constructed at a low cost. This phantom will allow DIR systems to be effectively and efficiently verified to ensure system performance.

A method to detect landmark pairs accurately between intra-patient volumetric medical images

PURPOSES

An image processing procedure was developed in this study to detect large quantity of landmark pairs accurately in pairs of volumetric medical images. The detected landmark pairs can be used to evaluate of deformable image registration (DIR) methods quantitatively.

METHODS

Landmark detection and pair matching were implemented in a Gaussian pyramid multi-resolution scheme. A 3D scale-invariant feature transform (SIFT) feature detection method and a 3D Harris-Laplacian corner detection method were employed to detect feature points, i.e., landmarks. A novel feature matching algorithm, Multi-Resolution Inverse-Consistent Guided Matching or MRICGM, was developed to allow accurate feature pairs matching. MRICGM performs feature matching using guidance by the feature pairs detected at the lower resolution stage and the higher confidence feature pairs already detected at the same resolution stage, while enforces inverse consistency.

RESULTS

The proposed feature detection and feature pair matching algorithms were optimized to process 3D CT and MRI images. They were successfully applied between the inter-phase abdomen 4DCT images of three patients, between the original and the re-scanned radiation therapy simulation CT images of two head-neck patients, and between inter-fractional treatment MRIs of two patients. The proposed procedure was able to successfully detect and match over 6300 feature pairs on average. The automatically detected landmark pairs were manually verified and the mismatched pairs were rejected. The automatic feature matching accuracy before manual error rejection was 99.4%. Performance of MRICGM was also evaluated using seven digital phantom datasets with known ground truth of tissue deformation. On average, 11855 feature pairs were detected per digital phantom dataset with TRE = 0.77 ± 0.72 mm.

CONCLUSION

A procedure was developed in this study to detect large number of landmark pairs accurately between two volumetric medical images. It allows a semi-automatic way to generate the ground truth landmark datasets that allow quantitatively evaluation of DIR algorithms for radiation therapy applications.

CT- and MRI-based gross target volume comparison in vestibular schwannomas

AIM

This study represents an enumeration and comparison of gross target volumes (GTV) as delineated independently on contrast-enhanced computed tomography (CT) and T1 and T2 weighted magnetic resonance imaging (MRI) in vestibular schwannomas (VS).

BACKGROUND

Multiple imaging in radiotherapy improves target localization.

METHODS AND MATERIALS

42 patients of VS were considered for this prospective study with one patient showing bilateral tumor. The GTV was delineated separately on CT and MRI. Difference in volumes were estimated individually for all the 43 lesions and similarity was studied between CT and T1 and T2 weighted MRI.

RESULTS

The male to female ratio for VS was found to be 1:1.3. The tumor was right sided in 34.9% and left sided in 65.1%. Tumor volumes (TV) on CT image sets were ranging from 0.251 cc to 27.27 cc. The TV for CT, MRI T1 and T2 weighted were 5.15 ± 5.2 cc, 5.8 ± 6.23 cc, and 5.9 ± 6.13 cc, respectively. Compared to MRI, CT underestimated the volumes. The mean dice coefficient between CT versus T1 and CT versus T2 was estimated to be 68.85 ± 18.3 and 66.68 ± 20.3, respectively. The percentage of volume difference between CT and MRI (%VD: mean ± SD for T1; 28.84 ± 15.0, T2; 35.74 ± 16.3) and volume error (%VE: T1; 18.77 ± 10.1, T2; 23.17 ± 13.93) were found to be significant, taking the CT volumes as the baseline.

CONCLUSIONS

MRI with multiple sequences should be incorporated for tumor volume delineation and they provide a clear boundary between the tumor and normal tissue with critical structures nearby.

Anthropomorphic phantom to investigate the bladder dose in gynecological high-dose-rate brachytherapy

PURPOSE

This study presents a prototype of a phantom appropriate for experimental bladder dosimetry. This work presents details of the phantom construction and dosimetric results obtained using radiochromic film and optically stimulated luminescence dosimeters (OSLDs).

METHODS AND MATERIALS

The phantom was constructed of polymethyl methacrylate. Two artificial bladders were three-dimensional printed using previous computed tomography images. Radiochromic films and OSLDs were positioned on the artificial bladder walls, and the applicators were placed according to the original computed tomography image.

RESULTS

The prototype phantom simulated the behavior of the dose on the bladder surface, enabling bladder movement in all directions. The dosimetric study that was performed using radiochromic film and OSLDs exhibited concordance, in most cases, with the results obtained from the planning system.

CONCLUSIONS

The methodology presented offers conditions for researchers to investigate more accurately the behavior of the dose on the bladder surface during intracavitary brachytherapy procedures.

Validation of a dose warping algorithm using clinically realistic scenarios

Objective:

Dose warping following deformable image registration (DIR) has been proposed for interfractional dose accumulation. Robust evaluation workflows are vital to clinically implement such procedures. This study demonstrates such a workflow and quantifies the accuracy of a commercial DIR algorithm for this purpose under clinically realistic scenarios.

Methods:

12 head and neck (H&N) patient data sets were used for this retrospective study. For each case, four clinically relevant anatomical changes have been manually generated. Dose distributions were then calculated on each artificially deformed image and warped back to the original anatomy following DIR by a commercial algorithm. Spatial registration was evaluated by quantitative comparison of the original and warped structure sets, using conformity index and mean distance to conformity (MDC) metrics. Dosimetric evaluation was performed by quantitative comparison of the dose–volume histograms generated for the calculated and warped dose distributions, which should be identical for the ideal “perfect” registration of mass-conserving deformations.

Results:

Spatial registration of the artificially deformed image back to the planning CT was accurate (MDC range of 1–2 voxels or 1.2–2.4 mm). Dosimetric discrepancies introduced by the DIR were low (0.02 ± 0.03 Gy per fraction in clinically relevant dose metrics) with no statistically significant difference found (Wilcoxon test, 0.6 ≥ p ≥ 0.2).

Conclusion:

The reliability of CT-to-CT DIR-based dose warping and image registration was demonstrated for a commercial algorithm with H&N patient data.

Advances in knowledge:

This study demonstrates a workflow for validation of dose warping following DIR that could assist physicists and physicians in quantifying the uncertainties associated with dose accumulation in clinical scenarios.

Challenges in using ¹⁸F-fluorodeoxyglucose-PET-CT to define a biological radiotherapy boost volume in locally advanced pancreatic cancer

BACKGROUND

The best method of identifying regions within pancreatic tumours that might benefit from an increased radiotherapy dose is not known. We investigated the utility of pre-treatment FDG-PET in predicting the spatial distribution of residual metabolic activity following chemoradiotherapy (CRT) in locally advanced pancreatic cancer (LAPC).

METHODS

17 patients had FDG-PET/CT scans at baseline and six weeks post-CRT. Tumour segmentation was performed at 40% and 50% of SUVmax at baseline and 60%, 70%, 80% and 90% post-CRT. FDG-PET scans were non-rigidly registered to the radiotherapy planning CT using the CT component of the FDG-PET/CT. Percentage overlap of the post-CRT volumes with the pre-CRT volumes with one another and the gross tumour volume (GTV) was calculated.

RESULTS

SUVmax decreased during CRT (median pre- 8.0 and post- 3.6, p < 0.0001). For spatial correlation analysis, 9 pairs of scans were included (Four were excluded following complete metabolic response, one patient had a non-FDG avid tumour, one had no post-CRT imaging, one had diffuse FDG uptake that could not be separated from normal tissues and one had an elevated blood glucose). The Pre40% and 50% of SUVmax volumes covered a mean of 50.8% and 30.3% of the GTV respectively. The mean% overlap of the 90%, 80%, 70%, 60% of SUVmax post-CRT with the Pre40% and Pre50% volumes were 83.3%, 84.0%, 83.7%, 77.9% and 77.8%, 69.9%, 74.5%, 64.8% respectively.

CONCLUSIONS

Regions of residual metabolic activity following CRT can be predicted from the baseline FDG-PET and could aid definition of a biological target volume for non-uniform dose prescriptions.

Deformable versus rigid registration of PET/CT images for radiation treatment planning of head and neck and lung cancer patients: a retrospective dosimetric comparison

Background

The purpose of this study is to evaluate the clinical impact of using deformable registration in tumor volume definition between separately acquired PET/CT and planning CT images.

Methods

Ten lung and 10 head and neck cancer patients were retrospectively selected. PET/CT images were registered with planning CT scans using commercially available software. Radiation oncologists defined two sets of gross tumor volumes based on either rigidly or deformably registered PET/CT images, and properties of these volumes were then compared.

Results

The average displacement between rigid and deformable gross tumor volumes was 1.8 mm (0.7 mm) with a standard deviation of 1.0 mm (0.6 mm) for the head and neck (lung) cancer subjects. The Dice similarity coefficients ranged from 0.76-0.92 and 0.76-0.97 for the head and neck and lung subjects, respectively, indicating conformity. All gross tumor volumes received at least 95% of the prescribed dose to 99% of their volume. Differences in the mean radiation dose delivered to the gross tumor volumes were at most 2%. Differences in the fraction of the tumor volumes receiving 100% of the radiation dose were at most 5%.

Conclusions

The study revealed limitations in the commercial software used to perform deformable registration. Unless significant anatomical differences between PET/CT and planning CT images are present, deformable registration was shown to be of marginal value when delineating gross tumor volumes.

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.

Using patient-specific phantoms to evaluate deformable image registration algorithms for adaptive radiation therapy

The quality of adaptive treatment planning depends on the accuracy of its underlying deformable image registration (DIR). The purpose of this study is to evaluate the performance of two DIR algorithms, B‐spline‐based deformable multipass (DMP) and deformable demons (Demons), implemented in a commercial software package. Evaluations were conducted using both computational and physical deformable phantoms. Based on a finite element method (FEM), a total of 11 computational models were developed from a set of CT images acquired from four lung and one prostate cancer patients. FEM generated displacement vector fields (DVF) were used to construct the lung and prostate image phantoms. Based on a fast‐Fourier transform technique, image noise power spectrum was incorporated into the prostate image phantoms to create simulated CBCT images. The FEM‐DVF served as a gold standard for verification of the two registration algorithms performed on these phantoms. The registration algorithms were also evaluated at the homologous points quantified in the CT images of a physical lung phantom. The results indicated that the mean errors of the DMP algorithm were in the range of 1.0~3.1mm for the computational phantoms and 1.9 mm for the physical lung phantom. For the computational prostate phantoms, the corresponding mean error was 1.0–1.9 mm in the prostate, 1.9–2.4 mm in the rectum, and 1.8–2.1 mm over the entire patient body. Sinusoidal errors induced by B‐spline interpolations were observed in all the displacement profiles of the DMP registrations. Regions of large displacements were observed to have more registration errors. Patient‐specific FEM models have been developed to evaluate the DIR algorithms implemented in the commercial software package. It has been found that the accuracy of these algorithms is patient‐dependent and related to various factors including tissue deformation magnitudes and image intensity gradients across the regions of interest. This may suggest that DIR algorithms need to be verified for each registration instance when implementing adaptive radiation therapy.

PACS numbers: 87.10.Kn, 87.55.km, 87.55.Qr, 87.57.nj

A quantitative comparison of the performance of three deformable registration algorithms in radiotherapy

We present an evaluation of various non-rigid registration algorithms for the purpose of compensating interfractional motion of the target volume and organs at risk areas when acquiring CBCT image data prior to irradiation. Three different deformable registration (DR) methods were used: the Demons algorithm implemented in the iPlan Software (BrainLAB AG, Feldkirchen, Germany) and two custom-developed piecewise methods using either a Normalized Correlation or a Mutual Information metric (featurelet_NC and featurelet_MI). These methods were tested on data acquired using a novel purpose-built phantom for deformable registration and clinical CT/CBCT data of prostate and lung cancer patients. The Dice similarity coefficient (DSC) between manually drawn contours and the contours generated by a derived deformation field of the structures in question was compared to the result obtained with rigid registration (RR). For the phantom, the piecewise methods were slightly superior, the featurelet_NC for the intramodality and the featurelet_MI for the intermodality registrations. For the prostate cases in less than 50% of the images studied the DSC was improved over RR. Deformable registration methods improved the outcome over a rigid registration for lung cases and in the phantom study, but not in a significant way for the prostate study. A significantly superior deformation method could not be identified.

Commissioning of a new wide-bore MRI scanner for radiotherapy planning of head and neck cancer

OBJECTIVE

A combination of CT and MRI is recommended for radiotherapy planning of head and neck cancers, and optimal spatial co-registration is achieved by imaging in the treatment position using the necessary immobilisation devices on both occasions, something which requires wide-bore scanners. Quality assurance experiments were carried out to commission a newly installed 1.5-T wide-bore MRI scanner and a dedicated, flexible six-channel phased array head and neck coil.

METHODS

Signal-to-noise ratio (SNR) and spatial signal uniformity were quantified using a homogeneous aqueous phantom, and geometric distortion was quantified using a phantom with water-filled fiducials in a grid pattern. Volunteer scans were also used to determine the in vivo image quality. Clinically relevant T1 weighted and T2 weighted fat-suppressed sequences were assessed in multiple scan planes (both sequences fast spin echo based). The performance of two online signal uniformity correction schemes, one utilising low-resolution reference scans and the other not utilising low-resolution reference scans, was compared.

RESULTS

Geometric distortions, for a ±35-kHz bandwidth, were <1 mm for locations within 10 cm of the isocentre rising to 1.8 mm at 18 cm away. SNR was above 50, and uniformity in the axial plane was 71% and 95% before and after uniformity correction, respectively.

CONCLUSION

The combined performance of the wide-bore scanner and the dedicated coil was adjudged adequate, although superior-inferior spatial coverage was slightly limited in the lower neck.

ADVANCES IN KNOWLEDGE

These results will be of interest to the increasing number of oncology centres that are seeking to incorporate MRI into planning practice using dedicated equipment.

The role of regularization in deformable image registration for head and neck adaptive radiotherapy

Deformable image registration provides a robust mathematical framework to quantify morphological changes that occur along the course of external beam radiotherapy treatments. As clinical reliability of deformable image registration is not always guaranteed, algorithm regularization is commonly introduced to prevent sharp discontinuities in the quantified deformation and achieve anatomically consistent results. In this work we analyzed the influence of regularization on two different registration methods, i.e. B-Splines and Log Domain Diffeomorphic Demons, implemented in an open-source platform. We retrospectively analyzed the simulation computed tomography (CTsim) and the corresponding re-planning computed tomography (CTrepl) scans in 30 head and neck cancer patients. First, we investigated the influence of regularization levels on hounsfield units (HU) information in 10 test patients for each considered method. Then, we compared the registration results of the open-source implementation at selected best performing regularization levels with a clinical commercial software on the remaining 20 patients in terms of mean volume overlap, surface and center of mass distances between manual outlines and propagated structures. The regularized B-Splines method was not statistically different from the commercial software. The tuning of the regularization parameters allowed open-source algorithms to achieve better results in deformable image registration for head and neck patients, with the additional benefit of a framework where regularization can be tuned on a patient specific basis.

Clinical validation of a 4D-CT based method for lung ventilation measurement in phantoms and patients

BACKGROUND

Lung cancer patients referred to radiotherapy (RT) often present with regional lung function deficits, and it is therefore of interest to image their lung function prior to treatment. In this study a method was developed that uses a deformable image registration (DIR) between the peak-inhale and peak-exhale phases of a thoracic four-dimensional computed tomography (4D-CT) scan to extract ventilation information. The method calculates the displacement vector fields (DVFs) resulting from the DIR using the Jacobian map approach in order to extract information regarding regional lung volume change.

MATERIAL AND METHODS

The DVFs resulting from DIRs were analysed to compute the Jacobian determinant of vectors in the field, thus obtaining a map of the vector gradients of the entire registered CT image, i.e. voxel-wise local volume change. Geometric and quantitative validation was achieved using images of both phantoms and patients. In the phantom studies, translations and deformations of known size and direction were introduced to validate both the DIR algorithm and the method as a whole. Furthermore, five patients underwent 4D-CT for planning of stereotactic body RT (SBRT). The patients were immobilised in a stereotactic body frame (SBF) and for each patient, two thoracic 4D-CT scans were acquired, one scan with respiration restricted by an abdominal compression plate and the other under free breathing.

RESULTS

In the phantom studies deformation errors were found to be of the order of the expected precision of 3 mm, corresponding to the image slice distance, in lateral and vertical directions. For the longitudinal direction a more pronounced discrepancy was observed, with the algorithm predicting displacement lengths of less than half of the physically introduced deformation. Qualitatively the method performed as expected. In the patient study an inverse consistency test showed deviations of up to 5.8 mm, i.e. almost twice the image slice separation. Jacobian maps of the patient images indicated well-ventilated areas as anatomically expected.

CONCLUSION

The established method provides a means of using a (commercially available) DIR algorithm to obtain a quantitative measure of local lung volume change. With further phantom and patient validation studies, quantitative maps of specific ventilation should be possible to produce and use in a clinical setting.