Accuracy of finite element model-based multi-organ deformable image registration

Adapting liver motion models using a navigator channel technique

Deformable registration can improve the accuracy of tumor targeting; however for online applications, efficiency as well as accuracy is important. A navigator channel technique has been developed to combine a biomechanical model-based deformable registration algorithm with a population motion model and patient specific motion information to perform fast deformable registration for application in image-guided radiation therapy. A respiratory population-based liver motion model was generated from breath-hold CT data sets of ten patients using a finite element model as a framework. The population model provides a biomechanical reference template of the average liver motions, which were found to be (absolute mean +/-SD) 0.12 +/- 0.10, 0.84 +/- 0.13, and 1.24 +/- 0.18 cm in the left-right (LR), anterior-posterior (AP), and superior-inferior (SI) directions, respectively. The population motion model was then adapted to the specific liver motion of 13 patients based on their exhale and inhale CT images. The patient motion was calculated using a navigator channel (a narrow region of interest window) on liver boundaries in the images. The absolute average accuracy of the navigator channel to predict the 1D SI and AP motions of the liver was less than 0.11, which is less than the out-of-plane image voxel size, 0.25 cm. This 1D information was then used to adapt the 4D population motion model in the SI and AP directions to predict the patient specific liver motion. The absolute average residual error of the navigator channel technique to adapt the population motion to the patients' specific motion was verified using three verification methods: (1) vessel bifurcation, (2) tumor center of mass, and (3) MORFEUS deformable algorithm. All three verification methods showed statistically similar results where the technique's accuracy was approximately on the order of the voxel image sizes. This method has potential applications in online assessment of motion at the time of treatment to improve image-guided radiotherapy and monitoring of intrafraction motion.

Dynamic MRI of grid-tagged hyperpolarized helium-3 for the assessment of lung motion during breathing

PURPOSE

To develop a dynamic magnetic resonance imaging (MRI) tagging technique using hyperpolarized helium-3 (HP He-3) to track lung motion.

METHODS AND MATERIALS

An accelerated non-Cartesian k-space trajectory was used to gain acquisition speed, at the cost of introducing image artifacts, providing a viable strategy for obtaining whole-lung coverage with adequate temporal resolution. Multiple-slice two-dimensional dynamic images of the lung were obtained in three healthy subjects after inhaling He-3 gas polarized to 35%-40%. Displacement, strain, and ventilation maps were computed from the observed motion of the grid peaks.

RESULTS

Both temporal and spatial variations of pulmonary mechanics were observed in normal subjects, including shear motion between different lobes of the same lung.

CONCLUSION

These initial results suggest that dynamic imaging of grid-tagged hyperpolarized magnetization may potentially be a powerful tool for observing and quantifying pulmonary biomechanics on a regional basis and for assessing, validating, and improving lung deformable image registration algorithms.

Adapting population liver motion models for individualized online image-guided therapy

Respiratory motion varies on a daily basis in abdominal cancer patients, affecting the ability to successfully deliver local therapy and requiring increased treatment margins to account for this variation. Deformable registration techniques can accurately describe respiratory motion, however, online application can be limited by long computational times and user intervention. A technique has been developed to quickly quantify patient breathing motion from respiratory-sorted volumetric images by calculating 1D shifts in image intensities between spatially corresponding regions of interest (navigator channels) on patient's images. The 1D motion at the superior, inferior, anterior, and posterior liver edges was detected and applied to adapt a population liver respiratory motion model. For validation, deformable registration was performed for each patient using a validated technique, MORFEUS, for relative validation, and vessel bifurcations, identified on patient's inhale and exhale images, for absolute validation. The accuracy of the adapted-population model to describe the patient respiratory motion was (absolute mean +/- SD) 0.26 +/- 0.11 cm and 0.30 +/- 0.21 cm in the superior-inferior (SI) and anterior-posterior (AP) directions, respectively. The accuracy of predicting the tumor COM motion was 0.30 +/- 0.22 cm, and 0.34 +/- 0.31, while the absolute validation, based on bifurcations was 0.26 +/- 0.16 cm and 0.13 +/- 0.04 cm in the SI and AP directions, respectively. This technique was developed to complement and quickly adapt a full 3D biomechanical based deformable registration technique, MORFEUS, to be applied in the online setting.

A framework for evaluation of deformable image registration spatial accuracy using large landmark point sets

Expert landmark correspondences are widely reported for evaluating deformable image registration (DIR) spatial accuracy. In this report, we present a framework for objective evaluation of DIR spatial accuracy using large sets of expert-determined landmark point pairs. Large samples (>1100) of pulmonary landmark point pairs were manually generated for five cases. Estimates of inter- and intra-observer variation were determined from repeated registration. Comparative evaluation of DIR spatial accuracy was performed for two algorithms, a gradient-based optical flow algorithm and a landmark-based moving least-squares algorithm. The uncertainty of spatial error estimates was found to be inversely proportional to the square root of the number of landmark point pairs and directly proportional to the standard deviation of the spatial errors. Using the statistical properties of this data, we performed sample size calculations to estimate the average spatial accuracy of each algorithm with 95% confidence intervals within a 0.5 mm range. For the optical flow and moving least-squares algorithms, the required sample sizes were 1050 and 36, respectively. Comparative evaluation based on fewer than the required validation landmarks results in misrepresentation of the relative spatial accuracy. This study demonstrates that landmark pairs can be used to assess DIR spatial accuracy within a narrow uncertainty range.

Objective assessment of deformable image registration in radiotherapy: a multi-institution study

The looming potential of deformable alignment tools to play an integral role in adaptive radiotherapy suggests a need for objective assessment of these complex algorithms. Previous studies in this area are based on the ability of alignment to reproduce analytically generated deformations applied to sample image data, or use of contours or bifurcations as ground truth for evaluation of alignment accuracy. In this study, a deformable phantom was embedded with 48 small plastic markers, placed in regions varying from high contrast to roughly uniform regional intensity, and small to large regional discontinuities in movement. CT volumes of this phantom were acquired at different deformation states. After manual localization of marker coordinates, images were edited to remove the markers. The resulting image volumes were sent to five collaborating institutions, each of which has developed previously published deformable alignment tools routinely in use. Alignments were done, and applied to the list of reference coordinates at the inhale state. The transformed coordinates were compared to the actual marker locations at exhale. A total of eight alignment techniques were tested from the six institutions. All algorithms performed generally well, as compared to previous publications. Average errors in predicted location ranged from 1.5 to 3.9 mm, depending on technique. No algorithm was uniformly accurate across all regions of the phantom, with maximum errors ranging from 5.1 to 15.4 mm. Larger errors were seen in regions near significant shape changes, as well as areas with uniform contrast but large local motion discontinuity. Although reasonable accuracy was achieved overall, the variation of error in different regions suggests caution in globally accepting the results from deformable alignment.

Inter-observer variability of prostate delineation on cone beam computerised tomography images

AIM

To determine the inter-observer variability of defining the prostate gland on cone beam computerised tomography images for the purposes of image-guided radiotherapy.

MATERIALS AND METHODS

Five genitourinary oncologists contoured the prostate gland on five cone beam computerised tomography datasets. The variations in prostate boundary delineation and consequent isocentre placement between observers were measured. Variations in volume and centre of mass were calculated. The variation in boundary definition was determined with finite element modelling.

RESULTS

The average standard deviation for centre of mass displacements was small, measuring 0.7, 1.8 and 2.8mm in the left-right, anterior-posterior and superior-inferior directions, respectively. The standard deviation for volume determination was 8.93 cm(3) with large variability (3.98-19.00 cm(3)). The mean difference between the computerised tomography-derived volume and the mean cone beam-derived volume was 16% (range 0-23.7%). The mean standard deviations for left-right, anterior-posterior and superior-inferior boundary displacements were, respectively, 1.8, 2.1 and 3.6 mm. The maximum deviation seen was 9.7 mm in the superior direction.

CONCLUSION

Expert observers had difficulty agreeing upon the location of the prostate peri-prostatic interface on the images provided. The effect on the centre of mass determination was small, and inter-observer variability for prostate detection on cone beam computerised tomography images is not prohibitive to the use of soft tissue guidance protocols. Potential exists for significant systematic matching errors, and points to the need for rigorous therapist image recognition training and development of guidance protocols before clinical implementation of soft tissue cone beam image guidance.

Performance evaluation of automatic anatomy segmentation algorithm on repeat or four-dimensional computed tomography images using deformable image registration method

PURPOSE

Auto-propagation of anatomic regions of interest from the planning computed tomography (CT) scan to the daily CT is an essential step in image-guided adaptive radiotherapy. The goal of this study was to quantitatively evaluate the performance of the algorithm in typical clinical applications.

METHODS AND MATERIALS

We had previously adopted an image intensity-based deformable registration algorithm to find the correspondence between two images. In the present study, the regions of interest delineated on the planning CT image were mapped onto daily CT or four-dimensional CT images using the same transformation. Postprocessing methods, such as boundary smoothing and modification, were used to enhance the robustness of the algorithm. Auto-propagated contours for 8 head-and-neck cancer patients with a total of 100 repeat CT scans, 1 prostate patient with 24 repeat CT scans, and 9 lung cancer patients with a total of 90 four-dimensional CT images were evaluated against physician-drawn contours and physician-modified deformed contours using the volume overlap index and mean absolute surface-to-surface distance.

RESULTS

The deformed contours were reasonably well matched with the daily anatomy on the repeat CT images. The volume overlap index and mean absolute surface-to-surface distance was 83% and 1.3 mm, respectively, compared with the independently drawn contours. Better agreement (>97% and <0.4 mm) was achieved if the physician was only asked to correct the deformed contours. The algorithm was also robust in the presence of random noise in the image.

CONCLUSION

The deformable algorithm might be an effective method to propagate the planning regions of interest to subsequent CT images of changed anatomy, although a final review by physicians is highly recommended.

Development of a novel post-processing treatment planning platform for 4D radiotherapy

The aim of this study is to develop an Automatic Post-processing Tool for four-dimensional (4D) treatment planning (APT4D) that enables the user to perform some necessary procedures related to 4D treatment planning, such as automated image registration, automatic propagation of regions of interest, and dose distribution transformation. Demons-based deformable registrations were performed to map the moving phase images (such as the end-inhalation phase or 0% phase) to the reference phase (typically the end-exhalation fixed phase or 50% phase). Contours were automatically propagated into the moving phase using the image registration results. The dose distribution of each moving phase was transformed to the fixed phase and subsequently was summed as an average with equal weighting factor. To validate the application of APT4D utility, the 4D computed tomography (CT) images of a lung cancer patient and an abdominal cancer patient were acquired and resorted into ten respiratory phases. 4D plans based on the 4D CT images were developed. The correlation coefficient ranged from 0.992 to 0.999 for the re-sampled deformed moving phase image against the fixed phase image for the lung patient plan and from 0.977 to 0.999 for the abdominal patient plan. For all the organs, the match indices between the manual contours and automatic contour propagation results were around 0.92 to 0.95. The 4D composite dose-volume histogram showed dosimetric reductions for liver and kidneys in the high dose region. The APT4D adds automation, efficiency, and functionality, while integrating the whole process of 4D treatment planning.

Combination of intensity-based image registration with 3D simulation in radiation therapy

Modern techniques of radiotherapy like intensity modulated radiation therapy (IMRT) make it possible to deliver high dose to tumors of different irregular shapes at the same time sparing surrounding healthy tissue. However, internal tumor motion makes precise calculation of the delivered dose distribution challenging. This makes analysis of tumor motion necessary. One way to describe target motion is using image registration. Many registration methods have already been developed previously. However, most of them belong either to geometric approaches or to intensity approaches. Methods which take account of anatomical information and results of intensity matching can greatly improve the results of image registration. Based on this idea, a combined method of image registration followed by 3D modeling and simulation was introduced in this project. Experiments were carried out for five patients 4DCT lung datasets. In the 3D simulation, models obtained from images of end-exhalation were deformed to the state of end-inhalation. Diaphragm motions were around -25 mm in the cranial-caudal (CC) direction. To verify the quality of our new method, displacements of landmarks were calculated and compared with measurements in the CT images. Improvement of accuracy after simulations has been shown compared to the results obtained only by intensity-based image registration. The average improvement was 0.97 mm. The average Euclidean error of the combined method was around 3.77 mm. Unrealistic motions such as curl-shaped deformations in the results of image registration were corrected. The combined method required less than 30 min. Our method provides information about the deformation of the target volume, which we need for dose optimization and target definition in our planning system.

Quantifying the accuracy of automated structure segmentation in 4D CT images using a deformable image registration algorithm

Four-dimensional (4D) radiotherapy is the explicit inclusion of the temporal changes in anatomy during the imaging, planning, and delivery of radiotherapy. One key component of 4D radiotherapy planning is the ability to automatically ("auto") create contours on all of the respiratory phase computed tomography (CT) datasets comprising a 4D CT scan, based on contours manually drawn on one CT image set from one phase. A tool that can be used to automatically propagate manually drawn contours to CT scans of other respiratory phases is deformable image registration. The purpose of the current study was to geometrically quantify the difference between automatically generated contours with manually drawn contours. Four-DCT data sets of 13 patients consisting of ten three-dimensional CT image sets acquired at different respiratory phases were used for this study. Tumor and normal tissue structures [gross tumor volume (GTV), esophagus, right lung, left lung, heart and cord] were manually drawn on each respiratory phase of each patient. Large deformable diffeomorphic image registration was performed to map each CT set from the peak-inhale respiration phase to the CT image sets corresponding with subsequent respiration phases. The calculated displacement vector fields were used to deform contours automatically drawn on the inhale phase to the other respiratory phase CT image sets. The code was interfaced to a treatment planning system to view the resulting images and to obtain the volumetric, displacement, and surface congruence information; 692 automatically generated structures were compared with 692 manually drawn structures. The auto- and manual methods showed similar trends, with a smaller difference observed between the GTVs than other structures. The auto-contoured structures agree with the manually drawn structures, especially in the case of the GTV, to within published interobserver variations. For the GTV, fractional volumes agree to within 0.2+/-0.1, center of mass displacements agree to within 0.5+/-1.5 mm, and agreement of surface congruence is 0.0+/-1.1 mm. The surface congruence between automatic and manual contours for the GTV, heart, left lung, right lung and esophagus was less than 5 mm in 99%, 94%, 94%, 91% and 89%, respectively. Careful assessment of the performance of automatic algorithms is needed in the presence of 4D CT artifacts.

Mesh-morphing algorithms for specimen-specific finite element modeling

Despite recent advances in software for meshing specimen-specific geometries, considerable effort is still often required to produce and analyze specimen-specific models suitable for biomechanical analysis through finite element modeling. We hypothesize that it is possible to obtain accurate models by adapting a pre-existing geometry to represent a target specimen using morphing techniques. Here we present two algorithms for morphing, automated wrapping (AW) and manual landmarks (ML), and demonstrate their use to prepare specimen-specific models of caudal rat vertebrae. We evaluate the algorithms by measuring the distance between target and morphed geometries and by comparing response to axial loading simulated with finite element (FE) methods. First a traditional reconstruction process based on microCT was used to obtain two natural specimen-specific FE models. Next, the two morphing algorithms were used to compute mappings from the surface of one model, the source, to the other, the target, and to use this mapping to morph the source mesh to produce a target mesh. The microCT images were then used to assign element-specific material properties. In AW the mappings were obtained by wrapping the source and target surfaces with an auxiliary triangulated surface. In ML, landmarks were manually placed on corresponding locations on the surfaces of both source and target. Both morphing algorithms were successful in reproducing the shape of the target vertebra with a median distance between natural and morphed models of 18.8 and 32.2 microm, respectively, for AW and ML. Whereas AW-morphing produced a surface more closely resembling that of the target, ML guaranteed correspondence of the landmark locations between source and target. Morphing preserved the quality of the mesh producing models suitable for FE simulation. Moreover, there were only minor differences between natural and morphed models in predictions of deformation, strain and stress. We therefore conclude that it is possible to use mesh-morphing techniques to produce accurate specimen-specific FE models of caudal rat vertebrae. Mesh morphing techniques provide advantages over conventional specimen-specific finite element modeling by reducing the effort required to generate multiple target specimen models, facilitating intermodel comparisons through correspondence of nodes and maintenance of connectivity, and lends itself to parametric evaluation of "artificial" geometries with a focus on optimizing reconstruction.

Assessment of dose reconstruction errors in image-guided radiation therapy

Dose reconstruction can be used to improve the accuracy of dose evaluation throughout a treatment course. Its working mechanism is based on deformable image registration (DIR). The purpose of this paper is to develop a method to estimate the dose reconstruction error associated with the inaccuracy of DIR algorithms. To reach this goal, we quantified dominant errors in DIR in terms of unbalanced energy (UE), which were compared with the standard displacement error (SDE). Their high similarity, characterized by Pearson correlation coefficient, was verified through nine 'demons' registration instances performed within simulated reference frames. Based on the similarity, the dose-warping discrepancy at each voxel was defined as a line integral of the dose gradient within the voxel's neighborhood whose boundary was determined by the voxel's UE value. From this definition, the dose reconstruction error was then calculated at each voxel on nine prostate computed tomography images, obtained from a patient treatment course. The average of the Pearson correlation coefficients between UE and SDE over the simulated registration instances was above 70%. The mean value of the dose reconstruction errors in a target volume was calculated for each of nine treatment fractions. The averaged percentage of these mean values with respect to the prescribed dose on the target volume was 1.68%. These results are consistent with contour-based mean dose error evaluations. This paper has established a relation between a registration error and its induced dose reconstruction discrepancy. It allows an automatic validation method to be developed to estimate the dose accumulation error at each voxel in clinical settings.

Adaptive boundary conditions for physically based follow-up breast MR image registration

This paper presents an algorithm for non-rigid registration of breast MRI follow-up images that compensates for differences in patient positioning while maintaining real anatomical and pathological changes. The proposed method uses a biomechanical model to constrain the deformation of the internal breast tissue according to elastic continuum mechanics, which is driven by suitable boundary conditions that align the breast surfaces in the images to be registered. Typically, such boundary conditions impose one-to-one surface point correspondences that are established a priori. We investigate alternative, more flexible boundary conditions that do not depend on fixed point correspondences and do not assume completely accurate breast surface segmentation in both images. More specifically, we allow for sliding motion of one surface over the other during deformation as well as for restricted motion perpendicular to the initially segmented boundary surface, based on the internal elastic forces and local intensity information. We evaluate the impact of different boundary conditions on registration quality from the subtraction images obtained for repeated scans of healthy volunteers with intermediate repositioning, using rigid body and free form whole volume intensity based registration for comparison, and also present initial results for actual patient data. Our results demonstrate a drastic reduction in subtraction artifacts using our model, without compromising the biomechanical validity of the deformation field such as unrealistically large local volume changes as with traditional voxel intensity based registration.

A comparison framework for breathing motion estimation methods from 4-D imaging

Motion estimation is an important issue in radiation therapy of moving organs. In particular, motion estimates from 4-D imaging can be used to compute the distribution of an absorbed dose during the therapeutic irradiation. We propose a strategy and criteria incorporating spatiotemporal information to evaluate the accuracy of model-based methods capturing breathing motion from 4-D CT images. This evaluation relies on the identification and tracking of landmarks on the 4-D CT images by medical experts. Three different experts selected more than 500 landmarks within 4-D CT images of lungs for three patients. Landmark tracking was performed at four instants of the expiration phase. Two metrics are proposed to evaluate the tracking performance of motion-estimation models. The first metric cumulates over the four instants the errors on landmark location. The second metric integrates the error over a time interval according to an a priori breathing model for the landmark spatiotemporal trajectory. This latter metric better takes into account the dynamics of the motion. A second aim of this paper is to estimate the impact of considering several phases of the respiratory cycle as compared to using only the extreme phases (end-inspiration and end-expiration). The accuracy of three motion estimation models (two image registration-based methods and a biomechanical method) is compared through the proposed metrics and statistical tools. This paper points out the interest of taking into account more frames for reliably tracking the respiratory motion.

Automated finite-element analysis for deformable registration of prostate images

Two major factors preventing the routine clinical use of finite-element analysis for image registration are: 1) the substantial labor required to construct a finite-element model for an individual patient's anatomy and 2) the difficulty of determining an appropriate set of finite-element boundary conditions. This paper addresses these issues by presenting algorithms that automatically generate a high quality hexahedral finite-element mesh and automatically calculate boundary conditions for an imaged patient. Medial shape models called m-reps are used to facilitate these tasks and reduce the effort required to apply finite-element analysis to image registration. Encouraging results are presented for the registration of CT image pairs which exhibit deformation caused by pressure from an endorectal imaging probe and deformation due to swelling.

Evaluation of respiratory-induced target motion for esophageal tumors at the gastroesophageal junction

PURPOSE

To quantify the internal motion margin requirements for radiotherapy of tumors near the gastroesophageal junction (GEJ).

METHODS AND MATERIALS

Four-dimensional computed tomography (4DCT) scans were obtained for 25 patients with primary tumors located near the GEJ. The gross tumor volume (GTV) was manually contoured on the exhale-phase image from the 4DCT image set. A deformable image registration method was used to automatically propagate the contours to other phases of the 4DCT images. To quantify target motion, we measured the displacement of the GTV centroid and the variations in the target boundary and volume. Internal margins were calculated in the lateral (RL), anterior-posterior (AP), and superior-inferior (SI) directions.

RESULTS

The mean+/-standard deviation peak-to-peak GTV centroid motion was 0.39+/-0.27cm (range, 0.04-1.09cm) in the RL, 0.38+/-0.23cm (range, 0.10-0.94cm) in the AP, and 0.87+/-0.47cm (range, 0.43-2.63cm) in the SI directions, respectively. On average, the internal target volume was 72% (range, 9-172%) larger than the GTV defined on a single-phase CT image. Variations in tumor boundaries due to tissue motion and deformation suggested asymmetric margins: 1.0cm left [toward the stomach], 0.8cm right, 1.1cm anterior, 0.6cm posterior, 1.0cm superior (toward the distal esophagus), and 1.6cm inferior (toward the stomach).

CONCLUSION

Because tumors near the GEJ are subject to a marked but asymmetric amount of respiratory-induced intrafractional tumor motion, the use of asymmetric internal margins may be beneficial.

Four-Dimensional Treatment Planning for Stereotactic Body Radiotherapy

PURPOSE

To investigate the influence of tumor motion on the calculation of four-dimensional (4D) dose distributions of the gross tumor volume (GTV) in pulmonary stereotactic body radiotherapy.

METHODS AND MATERIALS

For 7 patients with eight pulmonary tumors, a respiratory-correlated 4D-computed tomography study was acquired. The internal target volume was the sum of all tumor positions in the planning 4D-computed tomography study, and a 5-mm margin was used for generation of the planning target volume. Three-dimensional (3D) treatment plans were generated with a dose prescription of 3 x 12.5 Gy to the planning target volume enclosing the 65% and 80% isodose. After model-based nonrigid image registration, the 4D dose distributions were calculated.

RESULTS

No significant difference was found in the dose to the GTV with the tumor in the end-exhalation, end-inhalation, or mid-ventilation phase of the breathing cycle. The high-dose region was confined to the solid tumor, and lower doses were delivered to the surrounding pulmonary tissue of lower density. This nonstatic, variant dose distribution increased the 4D dose to the GTV by 6.2%, on average, compared with calculations using on a static dose distribution during the breathing cycle. The 4D accumulation resulted in a biologic effective dose (BED) of 143 +/- 8 Gy and 106 +/- 4 Gy to the GTV in the plan-65% and plan-80%, respectively. The dose to the ipsilateral lung was not different between the 3D and 4D dose calculations or between plan-65% and plan-80%.

CONCLUSIONS

In this study, the dose to the GTV was not decreased or blurred in the 4D plan compared with the 3D plan. The 3D doses to the GTV, internal target volume, and dose at the isocenter were good approximations of the 4D dose calculations. The 3D dose at the planning target volume margin underestimated the 4D dose significantly.

Tracking lung tissue motion and expansion/compression with inverse consistent image registration and spirometry

Breathing motion is one of the major limiting factors for reducing dose and irradiation of normal tissue for conventional conformal radiotherapy. This paper describes a relationship between tracking lung motion using spirometry data and image registration of consecutive CT image volumes collected from a multislice CT scanner over multiple breathing periods. Temporal CT sequences from 5 individuals were analyzed in this study. The couch was moved from 11 to 14 different positions to image the entire lung. At each couch position, 15 image volumes were collected over approximately 3 breathing periods. It is assumed that the expansion and contraction of lung tissue can be modeled as an elastic material. Furthermore, it is assumed that the deformation of the lung is small over one-fifth of a breathing period and therefore the motion of the lung can be adequately modeled using a small deformation linear elastic model. The small deformation inverse consistent linear elastic image registration algorithm is therefore well suited for this problem and was used to register consecutive image scans. The pointwise expansion and compression of lung tissue was measured by computing the Jacobian of the transformations used to register the images. The logarithm of the Jacobian was computed so that expansion and compression of the lung were scaled equally. The log-Jacobian was computed at each voxel in the volume to produce a map of the local expansion and compression of the lung during the breathing period. These log-Jacobian images demonstrate that the lung does not expand uniformly during the breathing period, but rather expands and contracts locally at different rates during inhalation and exhalation. The log-Jacobian numbers were averaged over a cross section of the lung to produce an estimate of the average expansion or compression from one time point to the next and compared to the air flow rate measured by spirometry. In four out of five individuals, the average log-Jacobian value and the air flow rate correlated well (R2 = 0.858 on average for the entire lung). The correlation for the fifth individual was not as good (R2 = 0.377 on average for the entire lung) and can be explained by the small variation in tidal volume for this individual. The correlation of the average log-Jacobian value and the air flow rate for images near the diaphragm correlated well in all five individuals (R2 = 0.943 on average). These preliminary results indicate a strong correlation between the expansion/compression of the lung measured by image registration and the air flow rate measured by spirometry. Predicting the location, motion, and compression/expansion of the tumor and normal tissue using image registration and spirometry could have many important benefits for radiotherapy treatment. These benefits include reducing radiation dose to normal tissue, maximizing dose to the tumor, improving patient care, reducing treatment cost, and increasing patient throughput.

Development of Multiorgan Finite Element-Based Prostate Deformation Model Enabling Registration of Endorectal Coil Magnetic Resonance Imaging for Radiotherapy Planning

PURPOSE

Endorectal coil (ERC) magnetic resonance imaging (MRI) provides superior visualization of the prostate compared with computed tomography at the expense of deformation. This study aimed to develop a multiorgan finite element deformable method, Morfeus, to accurately co-register these images for radiotherapy planning.

METHODS

Patients with prostate cancer underwent fiducial marker implantation and computed tomography simulation for radiotherapy planning. A series of axial MRI scans were acquired with and without an ERC. The prostate, bladder, rectum, and pubic bones were manually segmented and assigned linear elastic material properties. Morfeus mapped the surface of the bladder and rectum between two imaged states, calculating the deformation of the prostate through biomechanical properties. The accuracy of deformation was measured as fiducial marker error and residual surface deformation between the inferred and actual prostate. The deformation map was inverted to deform from 100 cm(3) to no coil.

RESULTS

The data from 19 patients were analyzed. Significant prostate deformation occurred with the ERC (mean intrapatient range, 0.88 +/- 0.25 cm). The mean vector error in fiducial marker position (n = 57) was 0.22 +/- 0.09 cm, and the mean vector residual surface deformation (n = 19) was 0.15 +/- 0.06 cm for deformation from no coil to 100-cm(3) ERC, with an image vector resolution of 0.22 cm. Accurately deformed MRI scans improved soft-tissue resolution of the anatomy for radiotherapy planning.

CONCLUSIONS

This method of multiorgan deformable registration enabled accurate co-registration of ERC-MRI scans with computed tomography treatment planning images. Superior structural detail was visible on ERC-MRI, which has potential for improving target delineation.

FEM-based evaluation of deformable image registration for radiation therapy

This paper presents a new concept to automatically detect the neighborhood in an image where deformable registration is mis-performing. Specifically, the displacement vector field (DVF) from a deformable image registration is substituted into a finite-element-based elastic framework to calculate unbalanced energy in each element. The value of the derived energy indicates the quality of the DVF in its neighborhood. The new voxel-based evaluation approach is compared with three other validation criteria: landmark measurement, a finite element approach and visual comparison, for deformable registrations performed with the optical-flow-based 'demons' algorithm as well as thin-plate spline interpolation. This analysis was performed on three pairs of prostate CT images. The results of the analysis show that the four criteria give mutually comparable quantitative assessments on the six registration instances. As an objective concept, the unbalanced energy presents no requirement on boundary constraints in its calculation, different from traditional mechanical modeling. This method is automatic, and at voxel level suitable to evaluate deformable registration in a clinical setting.

Direct measurement of lung motion using hyperpolarized helium-3 MR tagging

PURPOSE

To measure lung motion between end-inhalation and end-exhalation using a hyperpolarized helium-3 (HP (3)He) magnetic resonance (MR) tagging technique.

METHODS AND MATERIALS

Three healthy volunteers underwent MR tagging studies after inhalation of 1 L HP (3)He gas diluted with nitrogen. Multiple-slice two-dimensional and volumetric three-dimensional MR tagged images of the lungs were obtained at end-inhalation and end-exhalation, and displacement vector maps were computed.

RESULTS

The grids of tag lines in the HP (3)He MR images were well defined at end-inhalation and remained evident at end-exhalation. Displacement vector maps clearly demonstrated the regional lung motion and deformation that occurred during exhalation. Discontinuity and differences in motion pattern between two adjacent lung lobes were readily resolved.

CONCLUSIONS

Hyperpolarized helium-3 MR tagging technique can be used for direct in vivo measurement of respiratory lung motion on a regional basis. This technique may lend new insights into the regional pulmonary biomechanics and thus provide valuable information for the deformable registration of lung.

Technical note: A physical phantom for assessment of accuracy of deformable alignment algorithms

The purpose of this study was to investigate the feasibility of a simple deformable phantom as a QA tool for testing and validation of deformable image registration algorithms. A diagnostic thoracic imaging phantom with a deformable foam insert was used in this study. Small plastic markers were distributed through the foam to create a lattice with a measurable deformation as the ground truth data for all comparisons. The foam was compressed in the superior-inferior direction using a one-dimensional drive stage pushing a flat "diaphragm" to create deformations similar to those from inhale and exhale states. Images were acquired at different compressions of the foam and the location of every marker was manually identified on each image volume to establish a known deformation field with a known accuracy. The markers were removed digitally from corresponding images prior to registration. Different image registration algorithms were tested using this method. Repeat measurement of marker positions showed an accuracy of better than 1 mm in identification of the reference marks. Testing the method on several image registration algorithms showed that the system is capable of evaluating errors quantitatively. This phantom is able to quantitatively assess the accuracy of deformable image registration, using a measure of accuracy that is independent of the signals that drive the deformation parameters.

Assessment of a Model-Based Deformable Image Registration Approach for Radiation Therapy Planning

PURPOSE

The aim of this study is to develop a surface-based deformable image registration strategy and to assess the accuracy of the system for the integration of multimodality imaging, image-guided radiation therapy, and assessment of geometrical change during and after therapy.

METHODS AND MATERIALS

A surface-model-based deformable image registration system has been developed that enables quantitative description of geometrical change in multimodal images with high computational efficiency. Based on the deformation of organ surfaces, a volumetric deformation field is derived using different volumetric elasticity models as alternatives to finite-element modeling.

RESULTS

The accuracy of the system was assessed both visually and quantitatively by tracking naturally occurring landmarks (bronchial bifurcations in the lung, vessel bifurcations in the liver, implanted gold markers in the prostate). The maximum displacements for lung, liver and prostate were 5.3 cm, 3.2 cm, and 0.6 cm respectively. The largest registration error (direction, mean +/- SD) for lung, liver and prostate were (inferior-superior, -0.21 +/- 0.38 cm), (anterior-posterior, -0.09 +/- 0.34 cm), and (left-right, 0.04 +/- 0.38 cm) respectively, which was within the image resolution regardless of the deformation model. The computation time (2.7 GHz Intel Xeon) was on the order of seconds (e.g., 10 s for 2 prostate datasets), and deformed axial images could be viewed at interactive speed (less than 1 s for 512 x 512 voxels).

CONCLUSIONS

Surface-based deformable image registration enables the quantification of geometrical change in normal tissue and tumor with acceptable accuracy and speed.

A magnetic resonance imaging study of prostate deformation relative to implanted gold fiducial markers

PURPOSE

To describe prostate deformation during radiotherapy and determine the margins required to account for prostate deformation after setup to intraprostatic fiducial markers (FM).

METHODS AND MATERIALS

Twenty-five patients with T1c-T2c prostate cancer had three gold FMs implanted. The patients presented with a full bladder and empty rectum for two axial magnetic resonance imaging (MRI) scans using a gradient recalled echo (GRE) sequence capable of imaging the FMs. The MRIs were done at the time of radiotherapy (RT) planning and a randomly assigned fraction. A single observer contoured the prostate surfaces. They were entered into a finite element model and aligned using the centroid of the three FMs.

RESULTS

During RT, the prostate volume decreased by 0.5%/fraction (p = 0.03) and the FMs in-migrated by 0.05 mm/fraction (p < 0.05). Prostate deformation was unrelated to differential bladder and bowel filling, but was related to a transurethral resection of the prostate (TURP) (p = 0.003). The standard deviation for systematic uncertainty of prostate surface contouring was 0.8 mm and for FM centroid localization was 0.4 mm. The standard deviation of random interfraction prostate deformation was 1.5 mm and for FM centroid variability was 1.1 mm. These uncertainties from prostate deformation can be incorporated into a margin recipe to determine the total margins required for RT.

CONCLUSIONS

During RT, the prostate exhibited: volume decrease, deformation, and in-migration of FMs. Patients with TURPs were prone to prostate deformation.

A deformable image registration method to handle distended rectums in prostate cancer radiotherapy

In image-guided adaptive radiotherapy, it is important to have the capability to automatically and accurately delineate the rectal wall, which is a major dose-limiting organ in prostate cancer radiotherapy. As image registration is a process to find the spatial correspondence between two images, a major challenge in intensity-based deformable image registration is to deal with the situation where no correspondence exists for some objects between the two images to be registered. One example is the variation of rectal contents due to the presence and absence of bowel gas. The intensity-based deformable image registration methods alone cannot create the correct spatial transformation if there is no correspondence between the source and target images. In this study we implemented an automatic image intensity modification procedure to create artificial gas pockets in the planning computed tomography (CT) images. A diffusion-based deformable image registration algorithm was developed to use an adaptive smoothing algorithm to better handle large organ deformations. The process was tested in 15 prostate cancer cases and 30 daily CT images containing the largest distended rectums. The manually delineated rectums agreed well with the autodelineated rectums when using the image-intensity modification procedure.

Image registration and data fusion in radiation therapy

This paper provides an overview of image registration and data fusion techniques used in radiation therapy, and examples of their use. They are used at all stages of the patient management process; for initial diagnosis and staging, during treatment planning and delivery, and after therapy to help monitor the patients' response to treatment. Most treatment planning systems now support some form of interactive or automated image registration and provide tools for mapping information, such as tissue outlines and computed dose from one imaging study to another. To complement this, modern treatment delivery systems offer means for acquiring and registering 2D and 3D image data at the treatment unit to aid patient setup. Techniques for adapting and customizing treatments during the course of therapy using 3D and 4D anatomic and functional imaging data are currently being introduced into the clinic. These techniques require sophisticated image registration and data fusion technology to accumulate properly the delivered dose and to analyse possible physiological and anatomical changes during treatment. Finally, the correlation of radiological changes after therapy with delivered dose also requires the use of image registration and fusion techniques.

Prospective comparison of computed tomography and magnetic resonance imaging for liver cancer delineation using deformable image registration

PURPOSE

The aim of this study was to compare magnetic resonance imaging (MRI) with computed tomography (CT) for liver cancer tumor definition for high-precision radiotherapy planning.

METHODS AND MATERIALS

Diagnostic quality MRI scans and triphasic CT scans, with the liver immobilized in exhale, were obtained at the time of radiation planning for 26 patients with unresectable liver metastases (n = 8), hepatocellular carcinoma (n = 10), and cholangiocarcinoma (n = 8). On the CT and MRI series best demonstrating the tumor, the liver and gross tumor volumes (GTVs) were contoured, and intrahepatic anatomic reference points were identified. Deformable registration was used to register the liver from the CT with that from the MRI.

RESULTS

A difference in the number of tumor foci was seen on CT vs. MRI in 5 patients with hepatocellular carcinoma: MRI showed more foci in 3 patients, CT in 2. After deformable registration of the livers, the population median of the average distance between the CT tumor surface and MRI tumor surface was 3.7 mm (2.2-21.3 mm). The median percentage of tumor surface area that differed by > or = 5 mm was 26% (1-86%). Median percentage concordance volumes were 81% (77-86%) in metastases, 77% (60-88%) in hepatocellular carcinoma and 64% (25-85%) in cholangiocarcinoma.

CONCLUSION

Differences between MRI-defined liver cancer GTVs and CT-defined GTVs can be substantial and are more common in primary liver cancer.

Deformable registration of 4D computed tomography data

Four-dimensional radiotherapy requires deformable registration to track delivered dose across varying anatomical states. Deformable registration based on B-splines was implemented to register 4D computed tomography data to a reference respiratory phase. To assess registration performance, anatomical landmarks were selected across ten respiratory phases in five patients. These point landmarks were transformed according to global registration parameters between different respiratory phases. Registration uncertainties were computed by subtraction of transformed and reference landmark positions. The selection of appropriate registration masks to separate independently moving anatomical subunits is crucial to registration performance. The average registration error for five landmarks for each of five patients was 2.1 mm. This level of accuracy is acceptable for most radiotherapy applications.

Assessment of residual error in liver position using kV cone-beam computed tomography for liver cancer high-precision radiation therapy

PURPOSE

To evaluate the residual error in liver position using breath-hold kilovoltage (kV) cone-beam computed tomography (CT) following on-line orthogonal megavoltage (MV) image-guided breath-hold liver cancer conformal radiotherapy.

METHODS AND MATERIALS

Thirteen patients with liver cancer treated with 6-fraction breath-hold conformal radiotherapy were investigated. Before each fraction, orthogonal MV images were obtained during exhale breath-hold, with repositioning for offsets>3 mm, using the diaphragm for cranio-caudal (CC) alignment and vertebral bodies for medial-lateral (ML) and anterior posterior (AP) alignment. After repositioning, repeat orthogonal MV images, orthogonal kV fluoroscopic movies, and kV cone-beam CTs were obtained in exhale breath-hold. The cone-beam CT livers were registered to the planning CT liver to obtain the residual setup error in liver position.

RESULTS

After repositioning, 78 orthogonal MV image pairs, 61 orthogonal kV image pairs, and 72 kV cone-beam CT scans were obtained. Population random setup errors (sigma) in liver position were 2.7 mm (CC), 2.3 mm (ML), and 3.0 mm (AP), and systematic errors (Sigma) were 1.1 mm, 1.9 mm, and 1.3 mm in the superior, medial, and posterior directions. Liver offsets>5 mm were observed in 33% of cases; offsets>10 mm and liver deformation>5 mm were observed in a minority of patients.

CONCLUSIONS

Liver position after radiation therapy guided with MV orthogonal imaging was within 5 mm of planned position in the majority of patients. kV cone-beam CT image guidance should improve accuracy with reduced dose compared with orthogonal MV image guidance for liver cancer radiation therapy.

Three-dimensional velocity mapping of lung motion using vessel bifurcation pattern matching

We present a new quantification technique for three-dimensional (3D) lung motion by means of tracking the anatomical features inside the lung using a set of sequential 3D-CT images (a 4D-CT image). The method is based on the conservation of topology, such as connections and junctions of vessels, during the motion. Lung CT images are used to do lung volume modeling, lung vessel extracting and thinning, and coordinates of vessel bifurcations are derived as feature points. Such feature points are tracked in a series of 3D-CT images, i.e., the points are individually tracked between two successive 3D-CT images, in which the lung is deformed. Consequently, 3D displacement vectors are obtained. The feature point tracking is carried out using point pattern matching with a probabilistic relaxation method. We examined this technique using a lung 3D-CT image and artificially deformed one, and separately scanned CT images for a rigid bifurcation phantom. The studies estimated that the error of the vectors is within approximately 1 voxel, i.e., 1 mm or less. Therefore, the accuracy is expected to be high enough for radiation therapy. This technique enables us to quantify realistic 3D organ motion without any fiducial markers. It can be applied to the quantification of tumor (target volume) deformation by gridding interpolation into all voxels. We expect it to be useful for dose estimation in mobile organs and for 4D treatment planning in radiation therapy.

Deformable registration of the planning image (kVCT) and the daily images (MVCT) for adaptive radiation therapy

The incorporation of daily images into the radiotherapy process leads to adaptive radiation therapy (ART), in which the treatment is evaluated periodically and the plan is adaptively modified for the remaining course of radiotherapy. Deformable registration between the planning image and the daily images is a key component of ART. In this paper, we report our researches on deformable registration between the planning kVCT and the daily MVCT image sets. The method is based on a fast intensity-based free-form deformable registration technique. Considering the noise and contrast resolution differences between the kVCT and the MVCT, an 'edge-preserving smoothing' is applied to the MVCT image prior to the deformable registration process. We retrospectively studied daily MVCT images from commercial TomoTherapy machines from different clinical centers. The data set includes five head-neck cases, one pelvis case, two lung cases and one prostate case. Each case has one kVCT image and 20-40 MVCT images. We registered the MVCT images with their corresponding kVCT image. The similarity measures and visual inspections of contour matches by physicians validated this technique. The applications of deformable registration in ART, including 'deformable dose accumulation', 'automatic re-contouring' and 'tumour growth/regression evaluation' throughout the course of radiotherapy are also studied.

Prostate contouring uncertainty in megavoltage computed tomography images acquired with a helical tomotherapy unit during image-guided radiation therapy

PURPOSE

To evaluate the image-guidance capabilities of megavoltage computed tomography (MVCT), this article compares the interobserver and intraobserver contouring uncertainty in kilovoltage computed tomography (KVCT) used for radiotherapy planning with MVCT acquired with helical tomotherapy.

METHODS AND MATERIALS

Five prostate-cancer patients were evaluated. Each patient underwent a KVCT and an MVCT study, a total of 10 CT studies. For interobserver variability analysis, four radiation oncologists, one physicist, and two radiation therapists (seven observers in total) contoured the prostate and seminal vesicles (SV) in the 10 studies. The intraobserver variability was assessed by asking all observers to repeat the contouring of 1 patient's KVCT and MVCT studies. Quantitative analysis of contour variations was performed by use of volumes and radial distances.

RESULTS

The interobserver and intraobserver contouring uncertainty was larger in MVCT compared with KVCT. Observers consistently segmented larger volumes on MVCT where the ratio of average prostate and SV volumes was 1.1 and 1.2, respectively. On average (interobserver and intraobserver), the local delineation variability, in terms of standard deviations [Deltasigma = radical(sigma2MVCT-sigma2KVCT)], increased by 0.32 cm from KVCT to MVCT.

CONCLUSIONS

Although MVCT was inferior to KVCT for prostate delineation, the application of MVCT in prostate radiotherapy remains useful.

Deformable structure registration of bladder through surface mapping

Cumulative dose distributions in fractionated radiation therapy depict the dose to normal tissues and therefore may permit an estimation of the risk of normal tissue complications. However, calculation of these distributions is highly challenging because of interfractional changes in the geometry of patient anatomy. This work presents an algorithm for deformable structure registration of the bladder and the verification of the accuracy of the algorithm using phantom and patient data. In this algorithm, the registration process involves conformal mapping of genus zero surfaces using finite element analysis, and guided by three control landmarks. The registration produces a correspondence between fractions of the triangular meshes used to describe the bladder surface. For validation of the algorithm, two types of balloons were inflated gradually to three times their original size, and several computerized tomography (CT) scans were taken during the process. The registration algorithm yielded a local accuracy of 4 mm along the balloon surface. The algorithm was then applied to CT data of patients receiving fractionated high-dose-rate brachytherapy to the vaginal cuff, with the vaginal cylinder in situ. The patients' bladder filling status was intentionally different for each fraction. The three required control landmark points were identified for the bladder based on anatomy. Out of an Institutional Review Board (IRB) approved study of 20 patients, 3 had radiographically identifiable points near the bladder surface that were used for verification of the accuracy of the registration. The verification point as seen in each fraction was compared with its predicted location based on affine as well as deformable registration. Despite the variation in bladder shape and volume, the deformable registration was accurate to 5 mm, consistently outperforming the affine registration. We conclude that the structure registration algorithm presented works with reasonable accuracy and provides a means of calculating cumulative dose distributions.