Image Registration System Research Articles

A neural network-based 2D/3D image registration quality evaluator for pediatric patient setup in external beam radiotherapy.

Our purpose was to develop a neural network‐based registration quality evaluator (RQE) that can improve the 2D/3D image registration robustness for pediatric patient setup in external beam radiotherapy. Orthogonal daily setup X‐ray images of six pediatric patients with brain tumors receiving proton therapy treatments were retrospectively registered with their treatment planning computed tomography (CT) images. A neural network‐based pattern classifier was used to determine whether a registration solution was successful based on geometric features of the similarity measure values near the point‐of‐solution. Supervised training and test datasets were generated by rigidly registering a pair of orthogonal daily setup X‐ray images to the treatment planning CT. The best solution for each registration task was selected from 50 optimizing attempts that differed only by the randomly generated initial transformation parameters. The distance from each individual solution to the best solution in the normalized parametrical space was compared to a user‐defined error tolerance to determine whether that solution was acceptable. A supervised training was then used to train the RQE. Performance of the RQE was evaluated using test dataset consisting of registration results that were not used in training. The RQE was integrated with our in‐house 2D/3D registration system and its performance was evaluated using the same patient dataset. With an optimized sampling step size (i.e., 5 mm) in the feature space, the RQE has the sensitivity and the specificity in the ranges of 0.865–0.964 and 0.797–0.990, respectively, when used to detect registration error with mean voxel displacement (MVD) greater than 1 mm. The trial‐to‐acceptance ratio of the integrated 2D/3D registration system, for all patients, is equal to 1.48. The final acceptance ratio is 92.4%. The proposed RQE can potentially be used in a 2D/3D rigid image registration system to improve the overall robustness by rejecting unsuccessful registration solutions. The RQE is not patient‐specific, so a single RQE can be constructed and used for a particular application (e.g., the registration for images acquired on the same anatomical site). Implementation of the RQE in a 2D/3D registration system is clinically feasible.PACS numbers: 87.57.nj, 87.85.dq, 87.55.Qr

Are men who are biopsied without prior prostate magnetic resonance imaging getting substandard care?

Are men who are biopsied without prior prostate magnetic resonance imaging getting substandard care?

PO-0950: Validation of the deformable image registration system elastix in the head and neck region

PO-0950: Validation of the deformable image registration system elastix in the head and neck region

Quantitative Analysis Tools and Digital Phantoms for Deformable Image Registration Quality Assurance.

This article proposes quantitative analysis tools and digital phantoms to quantify intrinsic errors of deformable image registration (DIR) systems and establish quality assurance (QA) procedures for clinical use of DIR systems utilizing local and global error analysis methods with clinically realistic digital image phantoms. Landmark-based image registration verifications are suitable only for images with significant feature points. To address this shortfall, we adapted a deformation vector field (DVF) comparison approach with new analysis techniques to quantify the results. Digital image phantoms are derived from data sets of actual patient images (a reference image set, R, a test image set, T). Image sets from the same patient taken at different times are registered with deformable methods producing a reference DVFref. Applying DVFref to the original reference image deforms T into a new image R'. The data set, R', T, and DVFref, is from a realistic truth set and therefore can be used to analyze any DIR system and expose intrinsic errors by comparing DVFref and DVFtest. For quantitative error analysis, calculating and delineating differences between DVFs, 2 methods were used, (1) a local error analysis tool that displays deformation error magnitudes with color mapping on each image slice and (2) a global error analysis tool that calculates a deformation error histogram, which describes a cumulative probability function of errors for each anatomical structure. Three digital image phantoms were generated from three patients with a head and neck, a lung and a liver cancer. The DIR QA was evaluated using the case with head and neck.

Investigation of a tongue-internal coordinate system for two-dimensional ultrasound

In order to compare ultrasound recordings of tongue motion across utterances or across speakers, it is necessary to register the ultrasound images with respect to a common frame of reference. Methods for doing this typically rely either (1) on fixing the position of the ultrasound transducer relative to the skull by means of a helmet or a similar device, or (2) re-aligning the images by various means, such as optical tracking of head and transducer motion. These methods require sophisticated laboratory setups, and are less conducive to fieldwork or other studies in which such methods are impractical. In this study, we investigated the possibility of defining a rough coordinate system for image registration based on anatomical properties of the tongue itself. This coordinate system is anchored to the lower-jaw rather than the skull, but may potentially be transformed into an approximately skull-relative coordinate system by integrating video recordings of jaw motion.

TU‐A‐9A‐01: A Precise Deformable Image Registration System Using Feature‐Based Irregular Meshes

Purpose:To develop a deformable image registration system using irregular meshes generated based on image features. Finer meshes are generated at organ or tissue boundaries so that higher deformation resolution is achieved with limited number of control points.Methods:This deformable image registration system consists of two parts: a finite element modeling system and a GPU‐based registration system. The finite element modeling system is used to generate irregular meshes for the entire body or region of interest without segmentation. This system uses a Laplacian of Gaussion operator to extract features such as boundaries between organs and tissues. The placement of control points (vertices) of meshes is optimized based on feature intensities. The registration framework uses elastic energy as regularity and guarantees the final deformation vector field to be diffeomorphic. This system has been compared with registration algorithms based on regular meshes or voxels (Demons). The comparison is performed on XCAT digital phantom data and patient data.Results:Both XCAT male phantom and female phantom were used. Two respiratory phases were treated as floating and fix images for every phantom. Similarity measures were performed for image intensity and displacement vector field (DVF) for four sets of testing data with three types of deformation algorithms (irregular mesh, regular grid, and voxel). It clearly shows that the voxel‐based Demons algorithm leads to slightly better images. However, the DVF results show that mesh‐based algorithms behave much better. Compared with regular meshes, the irregular mesh we proposed leads to much faster convergence and converge to better results. The calculation is accelerated by GPU cards and a typical registration of patient data could be completed in about 1 minute.Conclusion:The feature‐based irregular meshing method provides a fast and accurate deformable image registration.This research is supported by CPRIT individual investigator award RP110329 and Varian research grant.

SU-E-T-542: Comparison of Stereotactic Radiosurgery (SRS) of Brain Lesions Using Gamma Knife, VMAT, IMRT, and Conformal Arcs

Purpose: To compare dose distributions in stereotactic radiation surgery of brain lesions using gamma Knife, VMAT, conformal arcs, and IMRT in order to provide an optimal treatment. Methods: Dose distributions from single shot of 4C model of Gamma Knife at the helmet collimation sizes of 4, 8, 14, and 18 mm in diameter were compared with full arcs with the square shapes of 4×4 (or 5×5), 8×8 (or 10×10), and spherical shapes of 16 or 20 mm in diameter using EDR3 films in the same gamma knife QA phantom. Plans for ten SRS cases with single and multiple lesions were created in gamma knife plans and Pinnacle plans. The external beam plans had enlarged field size by 2-mm and used single conformal full circle arc for solitary lesion and none coplanar arcs/beams for multiple lesions. Coverage, conformity index, dose to critical organs, and integral dose to the brain and nearby critical structures were compared on all plans. Structures and dose matrices were registered in a Velocity deformable image registration system. Results: Single full circle arc from Elekta beam-modulate MLC (4-mm leaf thickness) and agility MLC (5-mm leaf thickness) have larger penumbra and less flatness than that of Gamma Knife singlemore » shot. None-coplanar arcs or beams were required to achieve similar dose distribution. In general, Gamma Knife plans provided significant less integral dose than that of linac-based plans. Benefits of IMRT and VMAT versus gamma Knife and conformal arcs were not significant. Conclusion: Our dose measurement and treatment planning evaluation clearly demonstrated dose distribution differences amount current popular SRS modalities for small solitary and multiple brain lesions. The trend of using MLC shape beams or arcs to replace conventional cones should be revisited in order to keep lower integral dose if the late correlates with some radiation-induced side effects. Pilot grant from Elekta LLC.« less

A Study of the Anatomic Changes and Dosimetric Consequences in Adaptive CRT of Non-Small-Cell Lung Cancer Using Deformable CT and CBCT Image Registration

The aim of this study is to evaluate anatomic lung tumor changes and dosimetric consequences utilizing the deformable daily kilovolt (KV) cone-beam computer tomography (CBCT) image registration. Five patients diagnosed with NSCLC were treated with three-dimensional conformal radiotherapy (3D CRT) and 10 daily KV CBCT image sets were acquired for each patient. Each CBCT image and plan CT were imported into the deformable image registration (DIR) system. The plan CT image was deformed by the DIR system and a new contour on CBCT was obtained by using the auto-contouring function of the DIR. These contours were individually marked as CBCT f1, CBCT f2,..., and CBCT f10, and imported into a treatment planning system (TPS). The daily CBCT plan was individually generated with the same planning criteria based on new contours. These plans were individually marked as CBCTp1, CBCTp2,..., and CBCTp10, followed by generating a dose accumulation plan (DA plan) in original pCT image contour sets by adding all CBCT plans using Varian Eclipse TPS. The maximum, minimum and mean doses to the plan target volume (PTV) in the 5 DA plans were the same with the CT plans. However, the volume of radiation 5, 10, 20, 30, and 50 Gy of the total lungs in DA plans were less than those of the CT plans. The maximum dose of the spinal cord in the DA plans were average 27.96% less than the CT plans. The mean dose for the left, right, and total lungs in the DA plans were reduced by 13.80%, 23.65%, and 12.96%, respectively. The adaptive 3D CRT based on the deformable registration can reduce the dose to the lung and the spinal cord with the same PTV dose coverage. Moreover, it provides a method for further adaptive radiotherapy exploration.

Study of Remote Sensing Image Registration System

Remote sensing image registration system has been widely applied for agriculture, forestry, geology, ocean, meteorology, hydrology and military field, which studies image correction, enhancement and registration based on digital image processing technology. This paper presents system design and realization of registration technology for remote sensing image. Remote sensing image has geometric distortion and needs digital correction by adopting polynomial method at first. Then five kinds of technology are offered to enhance the contrast of remote sensing image including histogram enhancement, grey level transformation enhancement, image smoothing, image sharpening and pseudo color enhancement. Image geometric registration algorithm is carried out to fuse images of the same area acquired at different times on different wave band at last. Test result shows that the method presented has good accuracy and quick speed for realtime application.

Fully Automated Prostate Segmentation on MRI: Comparison With Manual Segmentation Methods and Specimen Volumes

The objective of our study was to compare calculated prostate volumes derived from tridimensional MR measurements (ellipsoid formula), manual segmentation, and a fully automated segmentation system as validated by actual prostatectomy specimens. Ninety-eight consecutive patients (median age, 60.6 years; median prostate-specific antigen [PSA] value, 6.85 ng/mL) underwent triplane T2-weighted MRI on a 3-T magnet with an endorectal coil while undergoing diagnostic workup for prostate cancer. Prostate volume estimates were determined using the formula for ellipsoid volume based on tridimensional measurements, manual segmentation of triplane MRI, and automated segmentation based on normalized gradient fields cross-correlation and graph-search refinement. Estimates of prostate volume based on ellipsoid volume, manual segmentation, and automated segmentation were compared with prostatectomy specimen volumes. Prostate volume estimates were compared using the Pearson correlation coefficient and linear regression analysis. The Dice similarity coefficient was used to quantify spatial agreement between manual segmentation and automated segmentation. The Pearson correlation coefficient revealed strong positive correlation between prostatectomy specimen volume and prostate volume estimates derived from manual segmentation (R = 0.89-0.91, p < 0.0001) and automated segmentation (R = 0.88-0.91, p < 0.0001). No difference was observed between manual segmentation and automated segmentation. Mean partial and full Dice similarity coefficients of 0.92 and 0.89, respectively, were achieved for axial automated segmentation. Prostate volume estimates obtained with a fully automated 3D segmentation tool based on normalized gradient fields cross-correlation and graph-search refinement can yield highly accurate prostate volume estimates in a clinically relevant time of 10 seconds. This tool will assist in developing a broad range of applications including routine prostate volume estimations, image registration, biopsy guidance, and decision support systems.

SU-E-T-468: Daily KV Cone-Beam CT and Deformable Image Registration as a Method for Studying Dosimetric Consequences of Anatomic Changes in Adaptive CRT of Lung Cancer

Purpose: To evaluate anatomic lung tumor changes assessment and actually delivered doses utilizing the deformable kV CBCT Image Registration and compare to 3D CRT for lung cancer. Methods: Five patients diagnosed with lung tumors were treated with 3D CRT and daily KV CBCT images were acquired. Each of the kV CBCT and plan CT images were imported into the Deformable Image Registration (DIR) system. The generated CT image was deformed by the DIR system and new contours were obtained by using the adaptive re-contouring function of the DIR. These contours of 30 sets were individually marked CBCT contours and imported into Varian Eclipce system. The plan for each pair of the new contours and the original contours of the CT image was individually designed with 3D CRT with the same filed angle. These CBCT plans were added by applying a dose accumulation plan (DA plan)and were compared to CT plans. Results: The maximum, minimum, and mean doses to the PTV in the DA plans were avarage separately 1.2%, 29.7%, and 17% less than the CT plans. V5, V10, V20, V30and V50 of the left, the right lung and total lung in DA plans were less than those of the CT plans. The maximum dose of the spinal cord in the DA plans were avarage 27. 9% less than the CT plans. The mean dose of the left, right lung and total lung in the DA plans were reduced by 13.8%, 23.7%, and 15.3% respectively. Conclusion: The adaptive 3D CRT of DA plans based on the deformable registration provides dose escalation on lung cancer, which is helpful in improving the local control and excluding the surrounding normal tissues. It provides a method for further adaptive radiotherapy exploration.

A stretching transform-based automatic nonrigid registration system for cerebrovascular digital subtraction angiography images

In the cerebrovascular digital subtraction angiography (DSA), patient motion is the primary cause of image quality degradation. In this article, we describe a nonrigid image registration system for motion artifact reduction in DSA which is fully automatic, effective, and computationally very efficient. In this system, the mask image is partitioned to generate the appropriate control points. The energy of histogram of differences method is adopted as similarity measurement, and the Powell algorithm is utilized for acceleration. A forward stretching transformation is used to complete the motion estimation, and an inverse stretching transformation is used to generate the target image by pixel mapping strategy. The method of this system is effective to maintain the topological relationships of the gray value before and after the image deformation. Preliminary experiments on the cerebrovascular DSA images illustrate the applicability of the system. After the deformation, the mask image remains clear and accurate contours, and the quality of the subtraction image after the registration is favorable. © 2013 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 23, 171–187, 2013

Junction Point Detection Algorithm for SAR Image

In this paper, we propose a novel junction point detector based on an azimuth consensus for remote sensing images. To eliminate the impact of noise and some noncorrelated edges of SAR image, an azimuth consensus constraint is developed. In addition to detecting the locations of junctions at the subpixel level, this operator recognizes their structures as well. A new formula that includes a minimization criterion for the total weighted distance is proposed to compute the locations of junction points accurately. Compared with other well-known detectors, including Forstner, JUDOCA, and CPDA, the experimental results indicate that our operator outperforms them both in location accuracy of junction points and in angle accuracy of branch edges. Moreover, our method possesses satisfying robustness to the impact of noise and changes of the SAR images. Our operator can be potentially used to solve a number of problems in computer vision, such as SAR image registration, wide-baseline matching, and UAV navigation system.

Comparative study of NOTES alone versus NOTES guided by a new image registration system for navigation in the mediastinum: a study in a porcine model

Natural orifice transluminal endoscopic surgery (NOTES) mediastinoscopy (MED) through the esophagus has proved to be feasible in the animal model. However, injury of the adjacent pleura and pneumothorax has been reported as a frequent adverse event when using a blind access. To assess the utility and safety of a CT-based image registration system (IRS) for navigation in the mediastinum. Prospective, randomized, controlled trial in 30 Yorkshire pigs. Thirty-minute MEDs were performed: 15 MEDs were performed with IRS guidance (MED-IRS), and 15 MEDs were performed with a blind access. Animal research laboratory. In both groups, the mediastinum was accessed through a 10-cm submucosal tunnel in the esophageal wall. Timed exploration was performed with identification of 8 mediastinal structures. Technical feasibility, adverse events, and the number of mediastinal structures identified. Thirty animals weighing 31.5 ± 3.5 kg were included in this study. MED was not possible in 2 animals in the "MED with blind access" group but was possible in all MEDs performed with IRS. The mean number of identified organs was slightly higher in "with IRS-MED" (6.13 ± 1.3) than with MED with blind access (4.7 ± 2.3; P = .066). Moreover, the right atrium and vena cava were identified in more cases with IRS-MED than in MED with blind access (13 vs 3 and 15 vs 11, P = .000 and P = .03, respectively). There were 3 (23%) adverse events with IRS-MED and 4 (27%) with "MED with blind access" (P = not significant), with pneumothorax being the most frequent (2 and 3, respectively). Nonsurvival animal study. This study demonstrates that the IRS system appears feasible in natural orifice transluminal endoscopic surgery MED and suggests that IRS guidance might be useful for selected procedures.

Comparative study of NOTES alone vs. NOTES guided by a new image registration system for navigation in the mediastinum: A study in a porcine model

Background and study aims: Natural Orifice Transluminal Endoscopic Surgery (NOTES) mediastinoscopy through the esophagus has proven to be feasible in the animal model. However, injury of the adjacent pleura and pneumothorax has been reported as a frequent complication when using a blind access. The aim of the current study was to assess the utility and safety of a CT-based Image Registered (IR) navigation system to identify mediastinal structures.

Automatic localization of vertebral levels in x-ray fluoroscopy using 3D-2D registration: a tool to reduce wrong-site surgery

Surgical targeting of the incorrect vertebral level (wrong-level surgery) is among the more common wrong-site surgical errors, attributed primarily to the lack of uniquely identifiable radiographic landmarks in the mid-thoracic spine. The conventional localization method involves manual counting of vertebral bodies under fluoroscopy, is prone to human error and carries additional time and dose. We propose an image registration and visualization system (referred to as LevelCheck), for decision support in spine surgery by automatically labeling vertebral levels in fluoroscopy using a GPU-accelerated, intensity-based 3D-2D (namely CT-to-fluoroscopy) registration. A gradient information (GI) similarity metric and a CMA-ES optimizer were chosen due to their robustness and inherent suitability for parallelization. Simulation studies involved ten patient CT datasets from which 50 000 simulated fluoroscopic images were generated from C-arm poses selected to approximate the C-arm operator and positioning variability. Physical experiments used an anthropomorphic chest phantom imaged under real fluoroscopy. The registration accuracy was evaluated as the mean projection distance (mPD) between the estimated and true center of vertebral levels. Trials were defined as successful if the estimated position was within the projection of the vertebral body (namely mPD <5 mm). Simulation studies showed a success rate of 99.998% (1 failure in 50 000 trials) and computation time of 4.7 s on a midrange GPU. Analysis of failure modes identified cases of false local optima in the search space arising from longitudinal periodicity in vertebral structures. Physical experiments demonstrated the robustness of the algorithm against quantum noise and x-ray scatter. The ability to automatically localize target anatomy in fluoroscopy in near-real-time could be valuable in reducing the occurrence of wrong-site surgery while helping to reduce radiation exposure. The method is applicable beyond the specific case of vertebral labeling, since any structure defined in pre-operative (or intra-operative) CT or cone-beam CT can be automatically registered to the fluoroscopic scene.

Real-time computed tomography-based augmented reality for natural orifice transluminal endoscopic surgery navigation

Natural orifice transluminal endoscopic surgery (NOTES) is technically challenging owing to endoscopic short-sighted visualization, excessive scope flexibility and lack of adequate instrumentation. Augmented reality may overcome these difficulties. This study tested whether an image registration system for NOTES procedures (IR-NOTES) can facilitate navigation. In three human cadavers 15 intra-abdominal organs were targeted endoscopically with and without IR-NOTES via both transgastric and transcolonic routes, by three endoscopists with different levels of expertise. Ease of navigation was evaluated objectively by kinematic analysis, and navigation complexity was determined by creating an organ access complexity score based on the same data. Without IR-NOTES, 21 (11·7 per cent) of 180 targets were not reached (expert endoscopist 3, advanced 7, intermediate 11), compared with one (1 per cent) of 90 with IR-NOTES (intermediate endoscopist) (P = 0·002). Endoscope movements were significantly less complex in eight of the 15 listed organs when using IR-NOTES. The most complex areas to access were the pelvis and left upper quadrant, independently of the access route. The most difficult organs to access were the spleen (5 failed attempts; 3 of 7 kinematic variables significantly improved) and rectum (4 failed attempts; 5 of 7 kinematic variables significantly improved). The time needed to access the rectum through a transgastric approach was 206·3 s without and 54·9 s with IR-NOTES (P = 0·027). The IR-NOTES system enhanced both navigation efficacy and ease of intra-abdominal NOTES exploration for operators of all levels. The system rendered some organs accessible to non-expert operators, thereby reducing one impediment to NOTES procedures.

TU‐A‐BRA‐02: Incorporating PET/CT Images into 3D Ultrasound‐Guided Biopsy of the Prostate

Purpose: Prostate cancer is the second leading cause of cancer mortality in American men. Systematic transrectal ultrasound (TRUS)‐guided prostate biopsy is considered as the standard method for prostate cancer detection. The current biopsy technique has a significant sampling error and can miss up to 30% of cancers. As a Result, a patient may be informed of a negative biopsy Result but may in fact be harboring an occult early‐stage cancer because the current ultrasound imaging technology has difficulty to differentiate carcinoma from benign prostate tissue and because the current TRUS‐guided biopsy is blind and random. We are developing methods to combine PET/CT with three‐dimensional (3D) ultrasound images for targeted biopsy of the prostate with the aim of improving cancer detection rate. Methods: The 3D ultrasound‐guided biopsy system consists of a 3D mechanical localization system and software workstation for image segmentation, registration, and biopsy planning. For PET imaging, we use a new molecular imaging tracer called anti‐1‐amino‐3‐18F‐fluorocyclobutane‐1‐carboxylic acid ([F18]FACBC) that has shown promising results for detecting and localizing prostate cancer in our human clinical trials. In order to plan biopsy, we developed a 3D, automatic segmentation method for the prostate ultrasound images. In order to incorporate PET/CT images into ultrasound‐guided biopsy, we developed image registration methods to fuse TRUS and PET/CT images. Results: The segmentation method was tested in ten patients with a DICE overlap ratio of 92.4% ±1.1%. The registration method has been tested in phantoms. The biopsy system was tested in prostate phantoms. Three‐dimensional ultrasound images were acquired from four human patients. We are integrating the system for PET/CT directed, 3D ultrasound‐guided, targeted biopsy in human patients. Conclusions: A PET/CT image‐directed, 3D ultrasound‐guided biopsy system has been developed for the prostate. This research is supported in part by NIH grant R01CA156775 (PI: Fei), Georgia Cancer Coalition Distinguished Clinicians and Scientists Award (PI: Fei), and the Emory Molecular and Translational Imaging Center (NIH P50CA128301).

SU-E-J-87: Building Deformation Error Histogram and Quality Assurance of Deformable Image Registration.

To quantify the systematic error of a Deformable Image Registration (DIR) system and establish Quality Assurance (QA) procedure. To address the shortfall of landmark approach which it is only available at the significant visible feature points, we adapted a Deformation Vector Map (DVM) comparison approach. We used two CT image sets (R and T image sets) taken for the same patient at different time and generated a DVM, which includes the DIR systematic error. The DVM was calculated using fine-tuned B-Spline DIR and L-BFGS optimizer. By utilizing this DVM we generated R' image set to eliminate the systematic error in DVM,. Thus, we have truth data set, R' and T image sets, and the truth DVM. To test a DIR system, we use R' and T image sets to a DIR system. We compare the test DVM to the truth DVM. If there is no systematic error, they should be identical. We built Deformation Error Histogram (DEH) for quantitative analysis. The test registration was performed with an in-house B-Spline DIR system using a stochastic gradient descent optimizer. Our example data set was generated with a head and neck patient case. We also tested CT to CBCT deformable registration. We found skin regions which interface with the air has relatively larger errors. Also mobile joints such as shoulders had larger errors. Average error for ROIs were as follows; CTV: 0.4mm, Brain stem: 1.4mm, Shoulders: 1.6mm, and Normal tissues: 0.7mm. We succeeded to build DEH approach to quantify the DVM uncertainty. Our data sets are available for testing other systems in our web page. Utilizing DEH, users can decide how much systematic error they would accept. DEH and our data can be a tool for an AAPM task group to compose a DIR system QA guideline. This project is partially supported by the Agency for Healthcare Research and Quality (AHRQ) grant 1R18HS017424-01A2.

WE-E-213CD-10: A Robust 2D/3D Image Registration System for Pediatric Patient Setup in External Beam Radiotherapy.

To develop a neural network based registration quality evaluator (RQE) that can improve the 2D/3D image registration robustness for pediatric patient setup in external beam radiotherapy. X-ray setup images of six pediatric patients with brain tumors received proton therapies were retrospectively registered with their treatment planning computed tomography (CT) images. A neural network based pattern classifier was used as an RQE to classify 2D/3D rigid registration solutions as successful or unsuccessful based on features of the similarity value surfaces near the point-of-solution. The supervised training and test datasets were generated by rigidly registering a pair of orthogonal daily setup x-ray images to the treatment planning CTs. The best solution for each registration task was selected from 50 optimizing attempts that differed only by the initial transformation parameters. The distance from each individual solution to the best solution in the normalized parametrical space was compared to a user- defined error threshold to determine whether that solution was successful. The supervised training was then used to train the RQE. The performance of the RQE was evaluated using the test dataset that consisted of registration results that were not used in training. With the optimized sampling step size, i.e., 5 mm, in the feature space, the RQE has the sensitivity and the specificity in the ranges of 0.865-0.964 and 0.797-0.990, respectively, when it was used to detect the registration error with the mean voxel displacement (MVD) greater than 1 mm. The proposed RQE can potentially be used in a 2D/3D rigid image registration system to improve its robustness by rejecting unsuccessful registration solutions. The RQE is non-patient specific and needs only to be trained once for each treatment site. The implementation the RQE in a 2D/3D registration system results only 10% to 20% increase of the overall computation time. The authors have no conflict of interest to disclose.