Jonathan Orban de Xivry

Evaluation of nonrigid registration models for interfraction dose accumulation in radiotherapy.

PURPOSE Interfraction dose accumulation is necessary to evaluate the dose distribution of an entire course of treatment by adding up multiple dose distributions of different treatment fractions. This accumulation of dose distributions is not straightforward as changes in the patient anatomy may occur during treatment. For this purpose, the accuracy of nonrigid registration methods is assessed for dose accumulation based on the calculated deformations fields. METHODS A phantom study using a deformable cubic silicon phantom with implanted markers and a cylindrical silicon phantom with MOSFET detectors has been performed. The phantoms were deformed and images were acquired using a cone-beam CT imager. Dose calculations were performed on these CT scans using the treatment planning system. Nonrigid CT-based registration was performed using two different methods, the Morphons and Demons. The resulting deformation field was applied on the dose distribution. For both phantoms, accuracy of the registered dose distribution was assessed. For the cylindrical phantom, also measured dose values in the deformed conditions were compared with the dose values of the registered dose distributions. Finally, interfraction dose accumulation for two treatment fractions of a patient with primary rectal cancer has been performed and evaluated using isodose lines and the dose volume histograms of the target volume and normal tissue. RESULTS A significant decrease in the difference in marker or MOSFET position was observed after nonrigid registration methods (p < 0.001) for both phantoms and with both methods, as well as a significant decrease in the dose estimation error (p < 0.01 for the cubic phantom and p < 0.001 for the cylindrical) with both methods. Considering the whole data set at once, the difference between estimated and measured doses was also significantly decreased using registration (p < 0.001 for both methods). The patient case showed a slightly underdosed planning target volume and an overdosed bladder volume due to anatomical deformations. CONCLUSIONS Dose accumulation using nonrigid registration methods is possible using repeated CT imaging. This opens possibilities for interfraction dose accumulation and adaptive radiotherapy to incorporate possible differences in dose delivered to the target volume and organs at risk due to anatomical deformations.

Diffeomorphic registration of images with variable contrast enhancement

Nonrigid image registration is widely used to estimate tissue deformations in highly deformable anatomies. Among the existing methods, nonparametric registration algorithms such as optical flow, or Demons, usually have the advantage of being fast and easy to use. Recently, a diffeomorphic version of the Demons algorithm was proposed. This provides the advantage of producing invertible displacement fields, which is a necessary condition for these to be physical. However, such methods are based on the matching of intensities and are not suitable for registering images with different contrast enhancement. In such cases, a registration method based on the local phase like the Morphons has to be used. In this paper, a diffeomorphic version of the Morphons registration method is proposed and compared to conventional Morphons, Demons, and diffeomorphic Demons. The method is validated in the context of radiotherapy for lung cancer patients on several 4D respiratory-correlated CT scans of the thorax with and without variable contrast enhancement.

Investigating CT to CBCT image registration for head and neck proton therapy as a tool for daily dose recalculation

PURPOSE Intensity modulated proton therapy (IMPT) of head and neck (H&N) cancer patients may be improved by plan adaptation. The decision to adapt the treatment plan based on a dose recalculation on the current anatomy requires a diagnostic quality computed tomography (CT) scan of the patient. As gantry-mounted cone beam CT (CBCT) scanners are currently being offered by vendors, they may offer daily or weekly updates of patient anatomy. CBCT image quality may not be sufficient for accurate proton dose calculation and it is likely necessary to perform CBCT CT number correction. In this work, the authors investigated deformable image registration (DIR) of the planning CT (pCT) to the CBCT to generate a virtual CT (vCT) to be used for proton dose recalculation. METHODS Datasets of six H&N cancer patients undergoing photon intensity modulated radiation therapy were used in this study to validate the vCT approach. Each dataset contained a CBCT acquired within 3 days of a replanning CT (rpCT), in addition to a pCT. The pCT and rpCT were delineated by a physician. A Morphons algorithm was employed in this work to perform DIR of the pCT to CBCT following a rigid registration of the two images. The contours from the pCT were deformed using the vector field resulting from DIR to yield a contoured vCT. The DIR accuracy was evaluated with a scale invariant feature transform (SIFT) algorithm comparing automatically identified matching features between vCT and CBCT. The rpCT was used as reference for evaluation of the vCT. The vCT and rpCT CT numbers were converted to stopping power ratio and the water equivalent thickness (WET) was calculated. IMPT dose distributions from treatment plans optimized on the pCT were recalculated with a Monte Carlo algorithm on the rpCT and vCT for comparison in terms of gamma index, dose volume histogram (DVH) statistics as well as proton range. The DIR generated contours on the vCT were compared to physician-drawn contours on the rpCT. RESULTS The DIR accuracy was better than 1.4 mm according to the SIFT evaluation. The mean WET differences between vCT (pCT) and rpCT were below 1 mm (2.6 mm). The amount of voxels passing 3%/3 mm gamma criteria were above 95% for the vCT vs rpCT. When using the rpCT contour set to derive DVH statistics from dose distributions calculated on the rpCT and vCT the differences, expressed in terms of 30 fractions of 2 Gy, were within [-4, 2 Gy] for parotid glands (D(mean)), spinal cord (D(2%)), brainstem (D(2%)), and CTV (D(95%)). When using DIR generated contours for the vCT, those differences ranged within [-8, 11 Gy]. CONCLUSIONS In this work, the authors generated CBCT based stopping power distributions using DIR of the pCT to a CBCT scan. DIR accuracy was below 1.4 mm as evaluated by the SIFT algorithm. Dose distributions calculated on the vCT agreed well to those calculated on the rpCT when using gamma index evaluation as well as DVH statistics based on the same contours. The use of DIR generated contours introduced variability in DVH statistics.

Phantom based evaluation of CT to CBCT image registration for proton therapy dose recalculation

The ability to perform dose recalculation on the anatomy of the day is important in the context of adaptive proton therapy. The objective of this study was to investigate the use of deformable image registration (DIR) and cone beam CT (CBCT) imaging to generate the daily stopping power distribution of the patient. We investigated the deformation of the planning CT scan (pCT) onto daily CBCT images to generate a virtual CT (vCT) using a deformable phantom designed for the head and neck (H & N) region. The phantom was imaged at a planning CT scanner in planning configuration, yielding a pCT and in deformed, treatment day configuration, yielding a reference CT (refCT). The treatment day configuration was additionally scanned at a CBCT scanner. A Morphons DIR algorithm was used to generate a vCT. The accuracy of the vCT was evaluated by comparison to the refCT in terms of corresponding features as identified by an adaptive scale invariant feature transform (aSIFT) algorithm. Additionally, the vCT CT numbers were compared to those of the refCT using both profiles and regions of interest and the volumes and overlap (DICE coefficients) of various phantom structures were compared. The water equivalent thickness (WET) of the vCT, refCT and pCT were also compared to evaluate proton range differences. Proton dose distributions from the same initial fluence were calculated on the refCT, vCT and pCT and compared in terms of proton range. The method was tested on a clinical dataset using a replanning CT scan acquired close in time to a CBCT scan as reference using the WET evaluation. Results from the aSIFT investigation suggest a deformation accuracy of 2-3 mm. The use of the Morphon algorithm did not distort CT number intensity in uniform regions and WET differences between vCT and refCT were of the order of 2% of the proton range. This result was confirmed by proton dose calculations. The patient results were consistent with phantom observations. In conclusion, our phantom study suggests the vCT approach is adequate for proton dose recalculation on the basis of CBCT imaging.

Evaluation of the radiobiological impact of anatomic modifications during radiation therapy for head and neck cancer: can we simply summate the dose?

BACKGROUND AND PURPOSE Adaptive strategies in radiotherapy (RT) require the knowledge of the total dose given to every organ of the body. Because of anatomical changes and setup errors non-rigid registration is necessary to map the different dose fractions to a common reference. This study evaluates practically if the accumulation of all of these registered dose fractions must take radiobiology into account in a classical clinical setting. MATERIALS AND METHODS Ten patients with head and neck tumors treated by chemo-RT were used. Contrast-enhanced CT scans were acquired prior and during RT following delivery of mean doses of 14.2, 24.5, 35.0 and 44.9 Gy and the planned pre-treatment helical tomotherapy sinograms were applied on the per-treatment CTs to create a series of per-treatment dose distributions corresponding to each per-treatment CT image. In order to calculate the cumulative dose distribution, the per-treatment dose maps were non-rigidly deformed by using the deformation map computed by a non-rigid registration. The deformed dose maps were then summed in two ways: one while taking radiobiology into account and one without. These two strategies were compared using clinical surrogates in the target volumes (TV) and in surrounding organs at risk (OAR). RESULTS The differences between the strategies, while statistically significant (p<0.05), are clinically irrelevant. In the OARs, the mean differences stay in the 0.01-0.07 Gy range for the total dose. In the targets, all mean differences stay in the 0.001-0.012 Gy range. However, some local high difference spots appear leading to punctual errors as high as 2.5 Gy. CONCLUSION If using current radiotherapy practices and clinical recommendations based on dose surrogates computed globally on OARs and TVs, one does not need to take radiobiological effects into account while accumulating total dose as these lead to very small differences compared to a simple accumulation technique consisting of a linear sum of the dose fractions. However, care must be taken if other adaptive strategies, based on local rather than global information, are used.

Effect of inter-subject variation on the accuracy of atlas-based segmentation applied to human brain structures

Large variations occur in brain anatomical structures in human populations, presenting a critical challenge to the brain mapping process. This study investigates the major impact of these variations on the performance of atlas-based segmentation. It is based on two publicly available datasets, from each of which 17 T1-weighted brain atlases were extracted. Each subject was registered to every other subject using the Morphons, a non-rigid registration algorithm. The automatic segmentations, obtained by warping the segmentation of this template, were compared with the expert segmentations using Dice index and the differences were statistically analyzed using Bonferroni multiple comparisons at significance level 0.05. The results showed that an optimum atlas for accurate segmentation of all structures cannot be found, and that the group of preferred templates, defined as being significantly superior to at least two other templates regarding the segmentation accuracy, varies significantly from structure to structure. Moreover, compared to other templates, a template giving the best accuracy in segmentation of some structures can provide highly inferior segmentation accuracy for other structures. It is concluded that there is no template optimum for automatic segmentation of all anatomical structures in the brain because of high inter-subject variation. Using a single fixed template for brain segmentation does not lead to good overall segmentation accuracy. This proves the need for multiple atlas based solutions in the context of atlas-based segmentation on human brain.

Impact of motion induced artifacts on automatic registration of lung tumors in Tomotherapy

PURPOSE Tomotherapy MV-CT acquisitions of lung tumors lead to artifacts due to breathing-related motion. This could preclude the reliability of tumor based positioning. We investigate the effect of these artifacts on automatic registration and determine conditions under which correct positioning can be achieved. MATERIALS AND METHODS MV-CT and 4D-CT scans of a dynamic thorax phantom were acquired with various motion amplitudes, directions, and periods. For each acquisition, the average kV-CT image was reconstructed from the 4D-CT data and rigidly registered with the corresponding MV-CT scan in a region of interest. Different kV-MV registration strategies have been assessed. RESULTS All tested registration methods led to acceptable registration errors (within 1.3 ± 1.2 mm) for motion periods of 3 and 6 s, regardless of the motion amplitude, direction, and phase difference. However, a motion period of 5 s, equal to half the Tomotherapy gantry period, induced asymmetric artifacts within MV-CT and significantly degraded the registration accuracy. CONCLUSIONS As long as the breathing period differs from 5 s, positioning based on averaged images of the tumor provides information about its daily baseline shift, and might therefore contribute to reducing margins, regardless of the registration method.

An overview of alignment issues for in-vivo image guided proton therapy

Protontherapy is based on physical properties of ion beams which allow the delivery of high radiation doses at very precise location in the body of the patient. The treatment planning aims at maximizing the delivery in the target volume while avoiding any organs at risk. The treatment is generally planned prior the treatment, and the patient is aligned in the treatment room on the basis of fiducial markers. However, the alignment of the patient may suffer from lack of precision and moreover, the body of the patient may vary between the time of imaging for planning and the time of treatment in the protontherapy room. More precise protontherapy and adaptive treatment which can track modifications of the body and the treatment of mobile tumors require the design of in vivo imaging systems to be deployed in the treatment room. The goal of this paper is to overview the present and future development of in-vivo image guided protontherapy and to give some image processing related challenges. The technique mostly used today is to take 2 orthogonal X-ray views of the patient. It requires an efficient 2D-3D coregistration procedure but is quite easy to deploy. Cone Beam CT is a next step which allows the capture of an in-vivo 3-D view on which the 3-D planning can be registered. The ultimate goal is to develop 4-D imaging techniques suited for the treatment of mobile tumors, for the cases of lung cancer. The development of new detectors will allow to validate the treatment by an “a posteriori” validation of the dose delivery in the body.

Feasibility and preliminary validation of 2D/3D image registration using fixed 2-D X-ray devices in image-guided radiotherapy

In radiotherapy, fixed 2-D X-ray imaging devices have several advantages compared to gantry-mounted systems, such as less geometrical deformations and the possibility to monitor 3-D markers motion in real-time. However, there is a lack of studies concerning the geometry of these systems. For example, in the case of a non-orthogonal geometry, the effect of the angle between the X-ray axes has not been investigated yet. In the first part of this study, the optimal angle was analyzed theoretically. Results showed that 60° between the axes still enables displacements of the order of 0.35 mm to be detected. In a second step, the performance of the registration method for such oblique configuration was evaluated on phantom data sets. It was found that using images separated by 60° rather than 90° required more than twice as much the number of iterations to obtain sufficient accuracy (i.e. 0.7 mm and 0.5°).

Respiratory motion variations from skin surface on lung cancer patients from 4D CT data

In radiation therapy of thorax and abdomen regions, knowing how respiratory motion modifies tumor position and trajectory is crucial for accurate dose delivery to tumors while avoiding healthy tissue and organs at risk. Three types of variations are studied: motion amplitudes measured from the patients skin surface and internal tumor trajectory, internal/external correlations and tumor trajectory baseline shift. Four male patients with lung cancer with three repeated 4D computed tomography (4DCT) scans, taken on different treatment days, were studied. Surfaces were extracted from 4DCT scans by segmentation. Motion over specific regions of interest was analyzed with respect to the motion of the tumor center of mass and correlation coefficient was computed. Tumor baseline shifts were analyzed after rigid registration based on vertebrae and surface registration. External amplitude variations were observed between fractions. Correlation coefficients of internal trajectories and external distances are greater than 0.6 in the abdomen. This correlation was observable and significant for all patients showing that the external motion is a good surrogate for internal movement on an intra-fraction basis. However for the inter-fraction case, external amplitude variations were observed between fractions and no correlation was found with the internal amplitude variations. Moreover, baseline shifts after surface registration were greater than those after vertebrae registration and the mean distance between surfaces after registration was not correlated to the magnitude of the baseline shift. These two observations show that, with the current representation of the external surface, inter-fraction variations are not detectable on the surface.