Comparison of an image registration technique based on normalized mutual information with a standard method utilizing implanted markers in the staged radiosurgical treatment of large arteriovenous malformations

Volume-Staged Stereotactic Radiosurgery for Intracranial Arteriovenous Malformations: Outcomes Based on an 18-Year Experience

BACKGROUND

Radiation-based treatment options of large intracranial arteriovenous malformations (AVM) must balance the likelihood of obliteration with the risk of adverse radiation effects (ARE).

OBJECTIVE

To analyze the efficacy and risks of volume-staged stereotactic radiosurgery (VS-SRS) for AVM.

METHODS

Retrospective study of 34 AVM patients having VS-SRS between 1997 and 2012. A median of 2 stages (range, 2-4) was used to treat a median AVM volume of 22.2 cm 3 (range, 7.4-56.7). The median AVM margin dose was 16 Gy (range, 14-18); the median radiosurgery-based AVM score was 2.81 (range, 1.54-6.45). The median follow-up after VS-SRS was 8.2 years (range, 3-13.3).

RESULTS

Nidus obliteration was noted in 18 patients (53%) after VS-SRS. The rate of obliteration was 14% at 3 years, 54% at 5 years, and 75% at 7 years. Six patients (18%) had 11 bleeds after VS-SRS. Two patients (6%) remained neurologically stable, 2 (6%) patients had significant deficits, and 2 patients (6%) died. The actuarial risk of a first bleed after VS-SRS was 6% at 1 year, 12% at 3 years, and 19% at 7 years. Six patients (18%) underwent repeat SRS; all achieved nidus obliteration for an overall cure rate of 71%. Two patients (6%) had a permanent ARE after VS-SRS or repeat SRS.

CONCLUSION

VS-SRS permitted large volume intracranial AVM to be treated with a low rate of ARE. Further study is needed on dose escalation and decreasing the treatment volume per stage to determine if this will increase the rate of obliteration with this technique.

Use of image registration and fusion algorithms and techniques in radiotherapy: Report of the AAPM Radiation Therapy Committee Task Group No. 132

Image registration and fusion algorithms exist in almost every software system that creates or uses images in radiotherapy. Most treatment planning systems support some form of image registration and fusion to allow the use of multimodality and time-series image data and even anatomical atlases to assist in target volume and normal tissue delineation. Treatment delivery systems perform registration and fusion between the planning images and the in-room images acquired during the treatment to assist patient positioning. Advanced applications are beginning to support daily dose assessment and enable adaptive radiotherapy using image registration and fusion to propagate contours and accumulate dose between image data taken over the course of therapy to provide up-to-date estimates of anatomical changes and delivered dose. This information aids in the detection of anatomical and functional changes that might elicit changes in the treatment plan or prescription. As the output of the image registration process is always used as the input of another process for planning or delivery, it is important to understand and communicate the uncertainty associated with the software in general and the result of a specific registration. Unfortunately, there is no standard mathematical formalism to perform this for real-world situations where noise, distortion, and complex anatomical variations can occur. Validation of the software systems performance is also complicated by the lack of documentation available from commercial systems leading to use of these systems in undesirable 'black-box' fashion. In view of this situation and the central role that image registration and fusion play in treatment planning and delivery, the Therapy Physics Committee of the American Association of Physicists in Medicine commissioned Task Group 132 to review current approaches and solutions for image registration (both rigid and deformable) in radiotherapy and to provide recommendations for quality assurance and quality control of these clinical processes.

A simple and efficient methodology to improve geometric accuracy in gamma knife radiation surgery: implementation in multiple brain metastases

PURPOSE

To propose, verify, and implement a simple and efficient methodology for the improvement of total geometric accuracy in multiple brain metastases gamma knife (GK) radiation surgery.

METHODS AND MATERIALS

The proposed methodology exploits the directional dependence of magnetic resonance imaging (MRI)-related spatial distortions stemming from background field inhomogeneities, also known as sequence-dependent distortions, with respect to the read-gradient polarity during MRI acquisition. First, an extra MRI pulse sequence is acquired with the same imaging parameters as those used for routine patient imaging, aside from a reversal in the read-gradient polarity. Then, "average" image data are compounded from data acquired from the 2 MRI sequences and are used for treatment planning purposes. The method was applied and verified in a polymer gel phantom irradiated with multiple shots in an extended region of the GK stereotactic space. Its clinical impact in dose delivery accuracy was assessed in 15 patients with a total of 96 relatively small (<2 cm) metastases treated with GK radiation surgery.

RESULTS

Phantom study results showed that use of average MR images eliminates the effect of sequence-dependent distortions, leading to a total spatial uncertainty of less than 0.3 mm, attributed mainly to gradient nonlinearities. In brain metastases patients, non-eliminated sequence-dependent distortions lead to target localization uncertainties of up to 1.3 mm (mean: 0.51 ± 0.37 mm) with respect to the corresponding target locations in the "average" MRI series. Due to these uncertainties, a considerable underdosage (5%-32% of the prescription dose) was found in 33% of the studied targets.

CONCLUSIONS

The proposed methodology is simple and straightforward in its implementation. Regarding multiple brain metastases applications, the suggested approach may substantially improve total GK dose delivery accuracy in smaller, outlying targets.

Validation of accuracy in image co-registration with computed tomography and magnetic resonance imaging in Gamma Knife radiosurgery

The latest version of Leksell GammaPlan (LGP) is equipped with Digital Imaging and Communication in Medicine (DICOM) image-processing functions including image co-registration. Diagnostic magnetic resonance imaging (MRI) taken prior to Gamma Knife treatment is available for virtual treatment pre-planning. On the treatment day, actual dose planning is completed on stereotactic MRI or computed tomography (CT) (with a frame) after co-registration with the diagnostic MRI and in association with the virtual dose distributions. This study assesses the accuracy of image co-registration in a phantom study and evaluates its usefulness in clinical cases. Images of three kinds of phantoms and 11 patients are evaluated. In the phantom study, co-registration errors of the 3D coordinates were measured in overall stereotactic space and compared between stereotactic CT and diagnostic CT, stereotactic MRI and diagnostic MRI, stereotactic CT and diagnostic MRI, and stereotactic MRI and diagnostic MRI co-registered with stereotactic CT. In the clinical study, target contours were compared between stereotactic MRI and diagnostic MRI co-registered with stereotactic CT. The mean errors of coordinates between images were < 1 mm in all measurement areas in both the phantom and clinical patient studies. The co-registration function implemented in LGP has sufficient geometrical accuracy to assure appropriate dose planning in clinical use.

Image registration accuracy of GammaPlan: a phantom study

OBJECT

The authors evaluated the accuracy of the automatic image coregistration function implemented in the Leksell GammaPlan treatment planning software (Version 4C with MultiView Extension and Version 8.0).

METHODS

The authors used a phantom with 9 landmarks (tips of thin cylindrical acrylic rods) evenly distributed in the treatment space. Two sets of images of the phantom were taken with both CT and MR imaging systems. The first image was obtained with the phantom aligned with the scanner's axis and the second scan was made by intentionally shifting and rotating the phantom relative to the scanner's axis. The authors attempted image registration of 2 CT image sets, CT and MR image sets, and 2 MR image sets. The accuracy of image registration was evaluated by measuring the x, y, and z coordinate values of the landmarks on each image set after 2 image sets were coregistered. The authors calculated the differences of the x, y, and z values and the distance, d, between corresponding landmarks in 2 image sets. To minimize interobserver dependence of coordinate measurements, 2 physicists did measurements independently.

RESULTS

The distances, d, averaged over the 9 landmarks, were 2.63 +/- 1.64 and 0.95 +/- 0.25 mm for CT-CT and MR-MR image registrations, respectively. When the CT images of the air-filled phantom and MR images were coregistered, however, the algorithm performed poorly: d = 13.8 +/- 1.23 mm. To remedy this, the authors undertook a 2-step process by first performing landmark-based registration of the 2 image sets and subsequently applying the automatic registration. With this approach, the mean distance drastically improved: d = 0.74 +/- 0.31 mm. When the water-filled phantom was used for CT scans, the registration accuracy of CT and MR image sets was acceptable without the 2-step registration process: d = 1.18 +/- 0.36 mm.

CONCLUSIONS

The accuracy of automatic registration of image sets from the same modality was within the voxel size of the scanned images. The accuracy of CT-MR image registration strongly depended on whether the phantom for CT scans was filled with air or water. This indicates the significant effect of the amount of common data available for a mutual information-based algorithm on the accuracy.

Anatomic Landmarks Versus Fiducials for Volume-Staged Gamma Knife Radiosurgery for Large Arteriovenous Malformations

PURPOSE

The purpose of this investigation was to compare the accuracy of using internal anatomic landmarks instead of surgically implanted fiducials in the image registration process for volume-staged gamma knife (GK) radiosurgery for large arteriovenous malformations.

METHODS AND MATERIALS

We studied 9 patients who had undergone 10 staged GK sessions for large arteriovenous malformations. Each patient had fiducials surgically implanted in the outer table of the skull at the first GK treatment. These markers were imaged on orthogonal radiographs, which were scanned into the GK planning system. For the same patients, 8-10 pairs of internal landmarks were retrospectively identified on the three-dimensional time-of-flight magnetic resonance imaging studies that had been obtained for treatment. The coordinate transformation between the stereotactic frame space for subsequent treatment sessions was then determined by point matching, using four surgically embedded fiducials and then using four pairs of internal anatomic landmarks. In both cases, the transformation was ascertained by minimizing the chi-square difference between the actual and the transformed coordinates. Both transformations were then evaluated using the remaining four to six pairs of internal landmarks as the test points.

RESULTS

Averaged over all treatment sessions, the root mean square discrepancy between the coordinates of the transformed and actual test points was 1.2 +/- 0.2 mm using internal landmarks and 1.7 +/- 0.4 mm using the surgically implanted fiducials.

CONCLUSION

The results of this study have shown that using internal landmarks to determine the coordinate transformation between subsequent magnetic resonance imaging scans for volume-staged GK arteriovenous malformation treatment sessions is as accurate as using surgically implanted fiducials and avoids an invasive procedure.

Less Common Indications for Stereotactic Radiosurgery or Fractionated Radiotherapy for Patients with Benign Brain Tumors

Microsurgical resection remains the mainstay of treatment for truly benign brain tumors that can be safely resected because of the potential for permanent cure with most histologic findings, including most of the histologic findings discussed in this article. Physicians must keep in mind the indolent nature of many of the benign brain tumors and realize that many patients are likely to live out normal life spans if tumor control is achieved. Therefore, it is not sufficient simply to consider local tumor control rates and short-term toxicity risks when choosing between surgery, stereotactic radiosurgery, and fractionated radiotherapy. Patients need to be apprised of all therapeutic options and to make their decisions with all information required to evaluate the risks and benefits. For benign brain tumors, these decisions may have consequences that last for decades.

A method to implement full six-degree target shift corrections for rigid body in image-guided radiotherapy

Treatment position setup errors often introduce temporal variations in the position of target relative to the planned external radiation beams. The errors can be introduced by the movement of a target relative to external setup marks or to other relevant landmarks that are used to position a patient for radiotherapy. Those variations can cause dose deviations from the planned doses and result in suboptimal treatments where part of the target is not fully irradiated or a critical structure receives more than desired radiation doses. Clinically available technology for image-guided radiotherapy can detect variations of target position. In this study, a method has been developed to correct for target position variations and restore the original beam geometries relative to the target. The technique involves three matrix transformations: (1) transformation of beams from the machine coordinate system to the patient coordinate system as in the patient geometry in the approved dosimetric plan; (2) transformation of beams from the patient coordinate system in the approved plan to the patient coordinate system that is identified at the time of treatment; (3) transformation of beams from the patient coordinate system at the time of treatment in the treatment patient geometry back to the machine coordinate system. The transformation matrix used for the second transformation is determined through the use of image-guided radiotherapy technology and image registration. By using these matrix transformations, the isocenter shift, the gantry, couch and collimator angles of the beams for the treatment, adjusted for the target shift, can be derived. With the new beam parameters, the beams will possess the same positions and orientations relative to the target as in the plan for a rigid body. This method was applied to a head phantom study, and it was found that the target shift was fully corrected in treatment and excellent agreement was found in target dose coverage between the plan and the treatment.

Dosimetric accuracy of a staged radiosurgery treatment

For large cerebral arteriovenous malformations (AVMs), the efficacy of radiosurgery is limited since the large doses necessary to produce obliteration may increase the risk of radiation necrosis to unacceptable levels. An alternative is to stage the radiosurgery procedure over multiple stages (usually two), effectively irradiating a smaller volume of the AVM nidus with a therapeutic dose during each session. The difference between coordinate systems defined by sequential stereotactic frame placements can be represented by a translation and a rotation. A unique transformation can be determined based on the coordinates of several fiducial markers fixed to the skull and imaged in each stereotactic coordinate system. Using this transformation matrix, isocentre coordinates from the first stage can be displayed in the coordinate system of subsequent stages allowing computation of a combined dose distribution covering the entire AVM. The accuracy of this approach was tested on an anthropomorphic head phantom and was verified dosimetrically. Subtle defects in the phantom were used as control points, and 2 mm diameter steel balls attached to the surface were used as fiducial markers and reference points. CT images (2 mm thick) were acquired. Using a transformation matrix developed with two frame placements, the predicted locations of control and reference points had an average error of 0.6 mm near the fiducial markers and 1.0 mm near the control points. Dose distributions in a staged treatment approach were accurately calculated using the transformation matrix. This approach is simple, fast and accurate. Errors were small and clinically acceptable for Gamma Knife radiosurgery. Accuracy can be improved by reducing the CT slice thickness.

Does new magnetic resonance imaging technology provide better geometrical accuracy during stereotactic imaging?

Object.The authors sought to compare the accuracy of stereotactic target imaging using the Siemens 1T EXPERT and 1.5T SYMPHONY magnetic resonance (MR) units.

Methods.A water-filled cylindrical Perspex phantom with axial and coronal inserts containing grids of glass rods was fixed in the Leksell stereotactic frame and subjected to MR imaging in Siemens 1T EXPERT and Siemens 1.5T SYMPHONY units. Identical sequences were used for each unit. The images were transferred to the GammaPlan treatment planning system. Deviations between stereotactic coordinates based on MR images and estimated real geometrical positions given by the construction of the phantom insert were evaluated for each study. The deviations were further investigated as a function of the MR unit used, MR sequence, the image orientation, and the spatial position of measured points in the investigated volume.

Conclusions.Larger distortions were observed when using the SYMPHONY 1.5T unit than those with the EXPERT 1T unit. Typical average distortion in EXPERT 1T was not more than 0.6 mm and 0.9 mm for axial and coronal images, respectively. Typical mean distortion for SYMPHONY 1.5T was not more than 1 mm and 1.3 mm for axial and coronal images, respectively. The image sequence affected the distortions in both units. Coronal T2-weighted spin-echo images performed in subthalamic imaging produced the largest distortions of 2.6 mm and 3 mm in the EXPERT 1T and SYMPHONY 1.5T, respectively. Larger distortions were observed in coronal slices than in axial slices in both units, and this effect was more pronounced in SYMPHONY 1.5T. Noncentrally located slice positions in the investigated volume of the phantom were associated with larger distortions.

Does new magnetic resonance imaging technology provide better geometrical accuracy during stereotactic imaging?

Object. The authors sought to compare the accuracy of stereotactic target imaging using the Siemens 1T EXPERT and 1.5T SYMPHONY magnetic resonance (MR) units.

Methods. A water-filled cylindrical Perspex phantom with axial and coronal inserts containing grids of glass rods was fixed in the Leksell stereotactic frame and subjected to MR imaging in Siemens 1T EXPERT and Siemens 1.5T SYMPHONY units. Identical sequences were used for each unit. The images were transferred to the GammaPlan treatment planning system. Deviations between stereotactic coordinates based on MR images and estimated real geometrical positions given by the construction of the phantom insert were evaluated for each study. The deviations were further investigated as a function of the MR unit used, MR sequence, the image orientation, and the spatial position of measured points in the investigated volume.

Conclusions. Larger distortions were observed when using the SYMPHONY 1.5T unit than those with the EXPERT 1T unit. Typical average distortion in EXPERT 1T was not more than 0.6 mm and 0.9 mm for axial and coronal images, respectively. Typical mean distortion for SYMPHONY 1.5T was not more than 1 mm and 1.3 mm for axial and coronal images, respectively. The image sequence affected the distortions in both units. Coronal T2-weighted spin-echo images performed in subthalamic imaging produced the largest distortions of 2.6 mm and 3 mm in the EXPERT 1T and SYMPHONY 1.5T, respectively. Larger distortions were observed in coronal slices than in axial slices in both units, and this effect was more pronounced in SYMPHONY 1.5T. Noncentrally located slice positions in the investigated volume of the phantom were associated with larger distortions.

Registration of two-dimensional cardiac images to preprocedural three-dimensional images for interventional applications

PURPOSE

To evaluate the accuracy and efficiency of rigid-body registration of two-dimensional fast cine and real-time cardiac images to high-resolution and SNR three-dimensional preprocedural reference volumes for application during MRI-guided interventional procedures.

MATERIALS AND METHODS

Mutual information (MI) and correlation ratio (CR) similarity measures were evaluated. The dependence of registration accuracy and efficiency on different resolution and SNR parameters, and also on cardiac-phase differences was evaluated in a porcine model. Two-dimensional images were initially misoriented at distances (d) of 2-10 mm, and rotations of +/-5 degrees about all axes. Registration error and computation time were evaluated, and performance was also assessed visually.

RESULTS

The maximum registration error using MI (<2.7 mm and <3.6 degrees ) occurred for d = 10 mm, misrotation of +/-5 degrees , and relative SNR = 1. The computation time was 15 seconds for MI and 10 seconds for CR.

CONCLUSION

Registration accuracy was not highly dependent on the relative timing, within the cycle, between the two-dimensional and three-dimensional images. Registration using CR was faster than that using MI, although accuracy was marginally higher with MI. J.