Accurate Image Registration Research Articles

Identification and separation of rigid image registration error sources, demonstrated for MRI-only image guided radiotherapy

Purpose: Rigid image registration (RIR) accuracy is crucial for image guided radiotherapy (IGRT). However, existing clinical image registration assessment methods cannot separate and quantify RIR error sources. Herein, we develop an extension of the ‘full circle method’ for RIR consistency. Paired registration circuits are used to isolate sources of RIR error caused by reference dataset substitution, from those inherent to the underlying RIR. This approach was demonstrated in the context of MRI-only IGRT, assessing substitution of MRI-derived synthetic-CT (sCT) for conventional CT, in a cohort of rectal cancer patients. Materials and methods: Planning CT, MRI-derived sCT, and two CBCTs from seven rectal cancer patients were retrospectively registered with global and soft tissue clipbox based RIR. Paired registration circuits were constructed using two moving (cone beam CT) images and two reference images (CT and sCT), per patient. Differences between inconsistencies in registration circuits containing CT and sCT were used to determine changes in registration accuracy due to substitution of sCT for CT. Results: sCT was found to be equivalent to CT under global RIR, with median differences of 0.05 mm and 0.01°. Soft tissue clipbox based RIR with sCT exhibited gross misregistration (>5 mm or 3°) for 3 patients. Registration consistency was degraded compared to CT across the cohort, with median differences of 0.54 mm and 0.15°. Conclusion: A paired registration circuit methodology for assessing RIR accuracy without ground truth information was developed and demonstrated for MRI-only IGRT in rectal cancer. This highlighted a reduction in clipbox based RIR consistency when sCT was substituted for conventional CT. The developed method enabled separation of degraded registration accuracy, from other error sources within the overall registration inconsistency. This novel methodology is applicable to any RIR scenario and enables analysis of the change in RIR performance on modification of image data or process.

An Image Registration Method for Multisource High-Resolution Remote Sensing Images for Earthquake Disaster Assessment.

For earthquake disaster assessment using remote sensing (RS), multisource image registration is an important step. However, severe earthquakes will increase the deformation between the remote sensing images acquired before and after the earthquakes on different platforms. Traditional image registration methods can hardly meet the requirements of accuracy and efficiency of image registration of post-earthquake RS images used for disaster assessment. Therefore, an improved image registration method was proposed for the registration of multisource high-resolution remote sensing images. The proposed method used the combination of the Shi_Tomasi corner detection algorithm and scale-invariant feature transform (SIFT) to detect tie points from image patches obtained by an image partition strategy considering geographic information constraints. Then, the random sample consensus (RANSAC) and greedy algorithms were employed to remove outliers and redundant matched tie points. Additionally, a pre-earthquake RS image database was constructed using pre-earthquake high-resolution RS images and used as the references for image registration. The performance of the proposed method was evaluated using three image pairs covering regions affected by severe earthquakes. It was shown that the proposed method provided higher accuracy, less running time, and more tie points with a more even distribution than the classic SIFT method and the SIFT method using the same image partitioning strategy.

The Effects of Distribution of Image Matched Fiducial Markers on Accuracy of Computer‐Guided Implant Surgery

Image registration of the optical intraoral scan to computed tomography image is essential for computer-guided implant surgery. The remaining teeth, which are considered to be congruent structures observed in the scan and radiographic images, are used to perform the image registration. The purpose of this study was to evaluate the effects of the distribution of matching fiducial points on the accuracy of the image registration. A partially edentulous model that has three anterior remaining teeth was prepared. Two mini dental implants were inserted in the posterior edentulous areas on both sides, and computed tomography and surface scan data were obtained. Three groups were set according to the distribution of the image matching points used: localized distribution, unilateral distribution, and bilateral distribution. Fifteen graduate students performed the registration process in each group using the same image matching method. The accuracy of image registration was evaluated by measuring the geometric discrepancies between the radiographic and registered scan images in the anterior, middle, and posterior regions. One-way and two-way analysis of variance with the Tukey HSD post hoc test were used for statistical analysis (α = 0.05) RESULTS: In general, the registration discrepancy was lowest in the bilateral distribution group, followed by the unilateral distribution and localized distribution groups (p<0.001). In the regional analysis, the registration error tended to increase as the measurement region moved farther from the matching points. The distribution of the matching points and measurement regions had a statistical interaction in the accuracy of image registration. The accuracy of image registration of the surface scan to the computed tomography is affected by the matching point distribution that can be improved by placing artificial markers in the edentulous areas. This article is protected by copyright. All rights reserved.

Interest of positron-emission tomography and magnetic resonance imaging for radiotherapy planning and control

Computed tomography (CT) in the treatment position is currently indispensable for planning radiation therapy. Other imaging modalities, such as magnetic resonance imaging (MRI) and positron emission-tomography (PET), can be used to improve the definition of the tumour and/or healthy tissue but also to provide functional data of the target volume. Accurate image registration is essential for treatment planning, so MRI and PET scans should be registered at the planning CT scan. Hybrid PET/MRI scans with a hard plane can be used but pose the problem of the absence of CT scans. Finally, techniques for moving the patient on a rigid air-cushioned table allow PET/CT/MRI scans to be performed in the treatment position while limiting the patient's movements exist. At the same time, the advent of MRI–linear accelerator systems allows to redefine image-guided radiotherapy and to propose treatments with daily recalculation of the dose. The place of PET during treatment remains more confidential and currently only in research and prototype status. The same development of imaging during radiotherapy is underway in proton therapy.

A Semiautomatic Registration Method for Chang’E-1 IIM Imagery Based on Globally Geo-Reference LROC-WAC Mosaic Imagery

Global geo-reference Lunar Reconnaissance Orbiter Camera-wide angle camera (LROC-WAC) mosaic imagery provides the precise geographic information for the mapping of mineral elements based on Chang’E-1 interferometric imaging spectrometer (IIM) imagery. However, the traditional image registration methods fail to achieve the accurate registration in between due to heterogeneous characteristics. Therefore, this letter proposes a semiautomatic registration method for Chang’E-1 IIM imagery based on global geo-reference LROC-WAC mosaic imagery. Due to the lack of ground control points, the method implemented a random sample consensus (RANSAC)-guided affine transformation (AT) model to help predict the potential coarse correspondence. Afterward, a multiwindow image matching approach is performed for the fine correspondence. To verify the performance of the proposed method, experiments were performed using Chang’E-1 IIM imagery and geo-reference LROC-WAC mosaic imagery. Experimental results indicate that the proposed method can obtain a massive number of homologous points while minimizing manual intervention, which is comparable to manual results and has high image registration accuracy.

A graphical user interface for automated 2- or 3-dimensional image registration in dental treatment recovery planning: the DentIR application.

Cone-beam computed tomography (CBCT) scans enable quantification of interproximal bone loss after implant procedures in dental patients. In order for this quantification to be accurate, software is typically used to manipulate image sets captured before and after implantation to obtain their exact registration (i.e., alignment). However, no affordable CBCT image registration software is currently available for dental applications. Thus, the aim of the present study was to develop a freely available graphical user interface, called DentIR, that automates 2-dimensional (2-D) or 3-D image registration for use in planning dental treatment. The DentIR app was designed using the MATLAB environment, downloaded to a desktop personal computer (PC and Mac), and tested for its ease of use and alignment accuracy in the absence of the MATLAB environment. The DentIR app enabled previewing of the CBCT images in 3-D to allow for filtering of each frame to reduce noise and blurring before registration. The 2-D or 3-D registration was tested with four transformation methods. The accuracy of each method was assessed by comparing the mean squared error and the peak signal-to-noise ratio values that were provided by the DentIR app. The registered images could be saved as Portable Network Graphics (PNG) images. The free, user-friendly DentIR app was easily downloadable to Mac or PC platforms. It provided accurate image registration to aid in the planning of dental treatment. Future updates of the DentIR app include adding the ability to register more than two images at once, enhancing image editing options and enabling registration of a cropped portion of the image for more in-depth analyses.

Feasibility of Robot-Assisted Ultrasound Imaging with Force Feedback for Assessment of Thyroid Diseases.

Medical ultrasound is extensively used to define tissue textures and to characterize lesions, and it is the modality of choice for detection and follow-up assessment of thyroid diseases. Classical medical ultrasound procedures are performed manually by an occupational operator with a hand-held ultrasound probe. These procedures require high physical and cognitive burden and yield clinical results that are highly operator-dependent, therefore frequently diminishing trust in ultrasound imaging data accuracy in repetitive assessment. A robotic ultrasound procedure, on the other hand, is an emerging paradigm integrating a robotic arm with an ultrasound probe. It achieves an automated or semi-automated ultrasound scanning by controlling the scanning trajectory, region of interest, and the contact force. Therefore, the scanning becomes more informative and comparable in subsequent examinations over a long-time span. In this work, we present a technique for allowing operators to reproduce reliably comparable ultrasound images with the combination of predefined trajectory execution and real-time force feedback control. The platform utilized features a 7-axis robotic arm capable of 6-DoF force-torque sensing and a linear-array ultrasound probe. The measured forces and torques affecting the probe are used to adaptively modify the predefined trajectory during autonomously performed examinations and probe-phantom interaction force accuracy is evaluated. In parallel, by processing and combining ultrasound B-Mode images with probe spatial information, structural features can be extracted from the scanning volume through a 3D scan. The validation was performed on a tissue-mimicking phantom containing thyroid features, and we successfully demonstrated high image registration accuracy between multiple trials.

A novel rotation invariant and Manhattan metric-based pose refinement: Augmented reality-based oral and maxillofacial surgery.

Augmented reality (AR) is gaining attention in medicine because of the convenience and innovation that it brings to operating rooms. Furthermore, oral and maxillofacial surgery (OMS), which is one of sensitive and narrow spatial surgery, requires high accuracy in image registration and low processing time of the system. However, the current systems are suffering from image registration problems while matching two different posture images. We thus aimed to increase that overlay accuracy and decrease the processing time. The proposed system consists of an Iterative Closest Point (ICP) algorithm, which is the combination of a rotation invariant and Manhattan error metric, to provide the best initial parameters and to decrease the computational cost by sorting high and low processing pixel images, respectively. The study on maxillary and mandibular jaw bone demonstrates that the proposed work overlay accuracy ranges from 0.22 to 0.30 mm, and processing time ranges from 10 to 14 frames per second as opposed to the 0.23- to 0.35-mm overlay accuracy and the current 8 to 12 frames per second processing time. This research aimed to improve the visualization and fast AR system for the OMS. Thus, the proposed system achieved an improvement in overlay accuracy and processing time by implementing the Rotation Invariant and Manhattan error metric ICP algorithm.

Validation of an MR-guided online adaptive radiotherapy (MRgoART) program: Deformation accuracy in a heterogeneous, deformable, anthropomorphic phantom

Background and purposeTo investigate deformable image registration (DIR) and multi-fractional dose accumulation accuracy of a clinical MR-guided online adaptive radiotherapy (MRgoART) program, utilizing clinically-based magnitudes of abdominal deformation vector fields (DVFs). Materials and methodsA heterogeneous anthropomorphic multi-modality abdominal deformable phantom was comprised of MR and CT anatomically-relevant materials. Thermoluminescent dosimeters (TLDs) were affixed within regions of interest (ROIs). CT and MR simulation scans were acquired. CT was deformed to MR for dose calculations. MRgoART was executed on a MR-linac (MRIdian) for 5 Gy/5 fractions. Before each fraction, a deformation was applied. Ground truth was known for ROI volume, TLD position, and TLD dose measured by an accredited dosimetry calibration laboratory. To validate the range of applied deformations, phantom DVFs were compared to DVFs of clinical abdominal MRgoART fractions. MR-MR deformation accuracy was quantified through dice similarity coefficient (DSC), Hausdorff distance (HD), mean distance-to-agreement (MDA), and as mean-absolute-error (MAE) for CT-MR-MR deformation. Arithmetic-summation of calculated dose at respective TLD positions and deform-accumulated dose (MIM) was compared to TLD measured dose, respectively. MR-MR deformation statistics were quantified for MRIdian and MIM. ResultsMean phantom DVFs were 5.0 ± 2.9 mm compared to mean DVF of clinical abdominal patients at 5.2 ± 3.0 mm. Respective mean DSC, HD, MDA was 0.93 ± 0.03, 0.74 ± 0.80 cm, 0.08 ± 0.03 cm for MRIdian and 0.93 ± 0.03, 0.54 ± 0.27 cm, 0.08 ± 0.03 cm for MIM (N = 80 ROIs). Mean MAE was 20.5 HU. Respective mean and median dose differences were 0.3%, −0.3% for arithmetic-summation and 4.1%, 0.6% for deformed-accumulation. Maximum differences were 0.21 Gy (arithmetic-summation), 0.31 Gy (deformed-accumulation). ConclusionsMRgoART deformation and dosimetric accuracy has been benchmarked for mean fractional DVFs of 5 mm in a multiple-rigid-body deformable phantom. Deformation accuracy was within TG132 criteria and clinically acceptable end-to-end MRgoART dosimetric agreement was observed for this phantom. Further efforts are needed in validation of deform-accumulated dose.

Reducing Shadow Effects on the Co-Registration of Aerial Image Pairs

Image registration is an important preprocessing step prior to detecting changes using multi-temporal image data, which is increasingly accomplished using automated methods. In high spatial resolution imagery, shadows represent a major source of illumination variation, which can reduce the performance of automated registration routines. This study evaluates the statistical relationship between shadow presence and image registration accuracy, and whether masking and normalizing shadows leads to improved automatic registration results. Eighty-eight bitemporal aerial image pairs were co-registered using software called Scale Invariant Features Transform (&lt;small&gt;SIFT&lt;/small&gt;) and Random Sample Consensus (&lt;small&gt;RANSAC&lt;/small&gt;) Alignment (&lt;small&gt;SARA&lt;/small&gt;). Co-registration accuracy was assessed at different levels of shadow coverage and shadow movement within the images. The primary outcomes of this study are (1) the amount of shadow in a multi-temporal image pair is correlated with the accuracy/success of automatic co-registration; (2) masking out shadows prior to match point select does not improve the success of image-to-image co-registration; and (3) normalizing or brightening shadows can help match point routines find more match points and therefore improve performance of automatic co-registration. Normalizing shadows via a standard linear correction provided the most reliable co-registration results in image pairs containing substantial amounts of relative shadow movement, but had minimal effect for pairs with stationary shadows.

LungRegNet: An unsupervised deformable image registration method for 4D-CT lung.

To develop an accurate and fast deformable image registration (DIR) method for four-dimensional computed tomography (4D-CT) lung images. Deep learning-based methods have the potential to quickly predict the deformation vector field (DVF) in a few forward predictions. We have developed an unsupervised deep learning method for 4D-CT lung DIR with excellent performances in terms of registration accuracies, robustness, and computational speed. A fast and accurate 4D-CT lung DIR method, namely LungRegNet, was proposed using deep learning. LungRegNet consists of two subnetworks which are CoarseNet and FineNet. As the name suggests, CoarseNet predicts large lung motion on a coarse scale image while FineNet predicts local lung motion on a fine scale image. Both the CoarseNet and FineNet include a generator and a discriminator. The generator was trained to directly predict the DVF to deform the moving image. The discriminator was trained to distinguish the deformed images from the original images. CoarseNet was first trained to deform the moving images. The deformed images were then used by the FineNet for FineNet training. To increase the registration accuracy of the LungRegNet, we generated vessel-enhanced images by generating pulmonary vasculature probability maps prior to the network prediction. We performed fivefold cross validation on ten 4D-CT datasets from our department. To compare with other methods, we also tested our method using separate 10 DIRLAB datasets that provide 300 manual landmark pairs per case for target registration error (TRE) calculation. Our results suggested that LungRegNet has achieved better registration accuracy in terms of TRE than other deep learning-based methods available in the literature on DIRLAB datasets. Compared to conventional DIR methods, LungRegNet could generate comparable registration accuracy with TRE smaller than 2mm. The integration of both the discriminator and pulmonary vessel enhancements into the network was crucial to obtain high registration accuracy for 4D-CT lung DIR. The mean and standard deviation of TRE were 1.00±0.53mm and 1.59±1.58mm on our datasets and DIRLAB datasets respectively. An unsupervised deep learning-based method has been developed to rapidly and accurately register 4D-CT lung images. LungRegNet has outperformed its deep-learning-based peers and achieved excellent registration accuracy in terms of TRE.

Moving vehicle detection in aerial infrared image sequences via fast image registration and improved YOLOv3 network

ABSTRACTOwing to its powerful adaptability and robustness to the weak light, the infrared camera equipped on the unmanned aerial vehicle is more and more applied for aerial photography. Nowadays, how to make full use of the aerial infrared image sequences for the moving vehicle detection has attracted widespread attention. However, due to the low-resolution, low contrast, and few texture features of the infrared image, it is extremely difficult to detect the moving vehicles. In our work, an effective and efficient moving vehicle detection approach in the aerial infrared image sequences is proposed via fast image registration and You Only Look Once Version 3 (YOLOv3) network. First, to compensate for the motion of the aerial infrared camera, a fast infrared image registration method is put forward. To improve the accuracy and efficiency of image registration, we construct a multi-screening based mechanism for screening out the incorrect and redundant feature points. For feature description, the low-level and high-level descriptors are combined to further improve the registration accuracy. Then, the 2-frame difference and image masking are introduced to acquire the frame mask images, where only the region-of-interest is reserved, and the remaining regions are masked. Next, we construct a new structure of an improved YOLOv3 network with only 23 layers. Due to the insufficiency of the infrared vehicle samples, transfer learning is introduced to train the improved YOLOv3 network. Finally, the proposed approach is evaluated on the Defence Advanced Research Projects Agency (DARPA) Video verification of Identify (VIVID) and Northwest Polytechnical University (NPU) data sets. Experiments and comprehensive analyses show that the proposed approach can achieve satisfactory and competitive moving vehicle detection results.

Rotation-Invariant Siamese Network for Low-Altitude Remote-Sensing Image Registration

Multiple-view change caused by small unmanned aerial vehicles (UAVs) monitoring the ground, resulting in image distortion, multiview transformation, and low overlap. Thus, such change has a strong effect on the accuracy of image registration. In this study, we utilize a Siamese network to deal with the complexity registration of low-altitude remote-sensing images. A robust neighbor-guided patch representation is designed to describe feature points based on neighborhood relation reconstruction, and patch selection. The network is trained based on rotation-invariant layer to solve the inevitable rotation, and nonrigid deformation caused by multiview images in low-altitude remote-sensing images. With only three training images involving 4500 putative matches, the experiment results demonstrated that the learned network can process the scenarios of yaw rotation, pitch rotation, mixture, and extreme (e.g., mixture, scaling, and distortion occur simultaneously) of UAV better than other six state-of-the-art methods.

Multi-Focus Image Region Fusion and Registration Algorithm with Multi-Scale Wavelet

Aiming at the problems of poor brightness control effect and low registration accuracy in traditional multi focus image registration, a wavelet multi-scale multi focus image region fusion registration method is proposed. The multi-scale Retinex algorithm is used to enhance the image, the wavelet decomposition similarity analysis is used for image interpolation, and the EMD method is used to decompose the multi focus image. Finally, the image reconstruction is completed and the multi focus image registration is realized. In order to verify the multi focus image fusion registration effect of different methods, a comparative experiment was designed. Experimental results show that the proposed method can control the image brightness in a reasonable range, the root mean square error of image region fusion registration algorithm is less than 5, and the image registration accuracy is high. This method can achieve multi focus image region fusion registration.

GF-2 Panchromatic and Multispectral Remote Sensing Image Registration Algorithm

Image registration is an important step in remote sensing image preprocessing. The accuracy, efficiency, and automatic degree of image registration will directly affect the results of subsequent image processing and analysis. Due to the slow registration speed and low registration accuracy of SIFT algorithm for GF-2 panchromatic and multispectral remote sensing images, SIFT algorithm is improved in this article to improve the efficiency of image registration algorithm. In this article, the strategy of information entropy meshes is introduced, and feature extraction is carried out only for the regions with large information entropy, so that the running time of feature extraction can be reduced. By introducing Canny operator to eliminate the unstable edge response points, the number of feature points can be further reduced, thus the computation amount of the algorithm can be also reduced. Due to the poor registration effect of SIFT algorithm for remote sensing images, a new gradient calculation method and feature description method are used in this article to enhance the robustness of feature descriptors. In the feature matching stage, this article proposes a two-layer matching strategy, which uses mutual Euclidean distance for initial matching, then uses FSC algorithm for fine matching, and finally realizes the registration of GF-2 panchromatic and multispectral remote sensing images. Experimental results show that the proposed algorithm can obtain more correct matching point pairs when the number of extracted feature points is small. Moreover, the registration accuracy and speed are both higher than the comparison algorithm. This article uses the remote sensing images of urban scenes dominated by plains and mountainous scenes with small terrain fluctuation respectively for experiments. The algorithm in this article can achieve registration accuracy better than 0.5 pixels, and in terms of time consumption, it is only about 70% of SIFT algorithm, and the computational efficiency is better than SIFT algorithm, which can meet the actual GF-2 Pan and MSI registration tasks.

Objective Assessment of the Quality and Accuracy of Deformable Image Registration.

Background:The increased use of deformable registration algorithms in clinical practice has also increased the need for their validation.Aims and Objectives:The purpose of the study was to investigate the quality, accuracy, and plausibility of three commercial image registration algorithms for 4-dimensional computed tomography (4DCT) datasets using various similarity measures.Materials and Methods:4DCT datasets were acquired for 10 lung cancer patients. 23 similarity measures were used to evaluate image registration quality. To ensure selected method's invertibility and assess resultant mechanical stress, the determinant of the Jacobian for the displacement field and 3-D Eulerian strain tensor were calculated. All the measures and calculations were applied on to extended deformable multi pass (EXDMP) and deformable multi pass (DMP) methods.Results:The results indicate the same trend for several of the studied measures. The Jacobian determinant values were always positive for the DMP method. The Eulerian strain tensor had smaller values for the DMP method than EXDMP in all of the studied cases. The negative values of the Jacobian determinant point to non-physical behavior of the EXDMP method. The Eulerian strain tensor values indicate less tissue strain for the DMP method. Large differences were also observed in the results between complete and cropped datasets (coefficient of determination: 0.55 vs. 0.93).Conclusion:A number of error and distance measures showed the best performance among the tested measures. The evaluated measures might detect CT dataset differences with higher precision if the analysis is restricted to a smaller volume.

Unsupervised 3D PET-CT Image Registration Method Using a Metabolic Constraint Function and a Multi-Domain Similarity Measure

High-resolution CT images can clearly display anatomical structures but does not display functional information, while blurred PET images can display molecular and functional information of lesions but cannot clearly display morphological structures. Therefore, accurate PET-CT image registration, which is used for anatomical structure and functional information fusion, is a prerequisite for early stage cancer diagnosis. However, some hypermetabolic anatomical structures, such as brain and bladder, have low registration accuracy. To solve this problem, a 3D unsupervised network based on a metabolic constraint function and a multi-domain similarity measure (3D MC-MDS Net) is proposed for 3D PET-CT image registration. Specifically, a metabolic constraint model is established based on the standard uptake value (SUV) distribution of hypermetabolic regions such as brain, bladder, liver and heart, which reduces the excessive distortion on displacement vector field (DVF) caused by hypermetabolic anatomical structures in PET images. A DVF estimator is built based on 3D unsupervised convolutional neural networks and a spatial transformer is used for warping 3D PET images to 3D CT images. The generated registration results (PET image patches) and the original 3D CT image patches are used for calculating the spatial domain similarity (SD similarity) and frequency domain similarity (FD similarity). Finally, the loss function of the entire registration network is constructed by a weighted sum of SD similarity, FD similarity and a smoothness of DVF. A dataset consisted of 170 whole-body PET-CT images is used for registration accuracy evaluation. The proposed unsupervised registration network, 3D MC-MDS Net, can accurately learn the 3D registration model by using the training dataset with the metabolic constraint model, which significantly improves the registration accuracy.

A multi-scale framework with unsupervised joint training of convolutional neural networks for pulmonary deformable image registration

To achieve accurate and fast deformable image registration (DIR) for pulmonary CT, we proposed a Multi-scale DIR framework with unsupervised Joint training of Convolutional Neural Network (MJ-CNN). MJ-CNN contains three models at multi-scale levels for a coarse-to-fine DIR to avoid being trapped in a local minimum. It is trained based on image similarity and deformation vector field (DVF) smoothness, requiring no supervision of ground-truth DVF. The three models are first trained sequentially and separately for their own registration tasks, and then are trained jointly for an end-to-end optimization under the multi-scale framework. In this study, MJ-CNN was trained using public SPARE 4D-CT data. The trained MJ-CNN was then evaluated on public DIR-LAB 4D-CT dataset as well as clinical CT-to-CBCT and CBCT-to-CBCT registration. For 4D-CT inter-phase registration, MJ-CNN achieved comparable accuracy to conventional iteration optimization-based methods, and showed the smallest registration errors compared to recently published deep learning-based DIR methods, demonstrating the efficacy of the proposed multi-scale joint training scheme. Besides, MJ-CNN trained using one dataset (SPARE) could generalize to a different dataset (DIR-LAB) acquired by different scanners and imaging protocols. Furthermore, MJ-CNN trained on 4D-CTs also performed well on CT-to-CBCT and CBCT-to-CBCT registration without any re-training or fine-tuning, demonstrating MJ-CNN’s robustness against applications and imaging techniques. MJ-CNN took about 1.4 s for DVF estimation and required no manual-tuning of parameters during the evaluation. MJ-CNN is able to perform accurate DIR for pulmonary CT with nearly real-time speed, making it very applicable for clinical tasks.

Effect of Gantry Speed on Image Quality of Four-dimensional Cone-beam Computed Tomography Using a Dynamic Phantom of Lung

The purpose of this paper was to determine the optimal imaging conditions for four-dimensional cone-beam computed tomography (4D-CBCT) using an X-ray tube and a flat-panel detector mounted on a radiotherapy device. The optimal imaging conditions were examined by changing the gantry speed (GS) parameter that affected the exposure time. Exposed dose during imaging and image quality of moving phantom were compared between examined conditions. The weighted computed tomography dose index (CTDIW) decreased linearly with increasing GS. However, when GS was 180°/min or faster, the image quality degraded, and errors of 1 mm or more were observed regarding the size of mock tumor in the moving phantom. The accuracy of automatic image matching was within 0.1 mm when GS of 120°/min or slower was chosen. From the results of this study, we concluded that GS of 120°/min is the optimum imaging condition. Under this imaging condition, the exposure time and CTDIW can be reduced by about 50% without compromising the accuracy of image registration, compared to the conventional GS of 70°/min. In addition, it has been clarified that there is an event that image reconstruction is not performed correctly due to the influence of phantom artifacts without depending on GS.

3D Orientation-Preserving Variational Models for Accurate Image Registration

The Beltrami coefficient from complex analysis has recently been found to provide a robust constraint for obtaining orientation-preserving and diffeomorphic transformations for registration of plan...