Comparison of imaging modalities for the accurate delineation of arteriovenous malformation, with reference to stereotactic radiosurgery

Comparison of imaging modalities for the accurate delineation of arteriovenous malformations (AVM) and evaluation of setup accuracy with reference to non-invasive LINAC-based stereotactic radiosurgery (SRS)

AIMS

To compare the accuracy of nidus delineation using magnetic resonance angiography (MRA) to digital subtraction angiography (DSA) and to evaluate setup accuracy of non-invasive frame SRS treatments.

SETTINGS AND DESIGN

A prospective observational study of 16 patients who underwent non-invasive frame LINAC-based SRS for brain AVMs.

MATERIALS AND METHODS

The nidus was separately delineated using DSA and MRA after co-registration onto CT simulation images and compared with respect to their volume and maximum diameters. During treatment, the setup errors observed in x-, y-, and z-directions were recorded.

STATISTICAL ANALYSIS

Paired t-test (to compare volume and maximum diameter). Wilcoxon signed-rank test (for setup accuracy).

RESULTS

The mean volume of nidus contoured in MRA was 4.16 cc compared to 3.11 cc in DSA (P 0.297). The mean maximum diameters using MRA and DSA, respectively, in antro-posterior, cranio- caudal, and transverse diameters were 21.97 cc vs. 19.46 cc (P 0.2380), 6.59 cc vs. 9.63 cc (P 0.161), and 18.87 cc vs. 16.81 cc (P 0.178). But these modalities can potentially misinterpret the nidus volume, warranting caution for use of either modality alone. The mean translational shift observed in the x-, y-, and z-directions were 0.06 mm, 0.13 mm, and 0.13 mm, respectively, when couch was brought to neutral position after clockwise couch rotation and 0.07, 0, and 0, respectively, after counterclockwise couch rotation.

CONCLUSION

This study could not demonstrate any statistically significant differences in nidus delineation between MRA and DSA. Setup accuracy achieved with non-invasive thermoplastic mask-based immobilization is within acceptable limits for SRS.

Arterial Spin Labeling for Pediatric Central Nervous System Diseases: Techniques and Clinical Applications

Dynamic susceptibility contrast (DSC) and arterial spin labeling (ASL) are techniques used to evaluate brain perfusion using MRI. DSC requires dynamic image acquisition with a rapid administration of gadolinium-based contrast agent. In contrast, ASL obtains brain perfusion information using magnetically labeled blood water as an endogenous tracer. For the evaluation of brain perfusion in pediatric neurological diseases, ASL has a significant advantage compared to DSC, CT, and single-photon emission CT/positron emission tomography because of the lack of radiation exposure and contrast agent administration. However, in ASL, optimization of several parameters, including the type of labeling, image acquisition, background suppression, and postlabeling delay, is required, because they have a significant effect on the quantification of cerebral blood flow (CBF).In this article, we first review recent technical developments of ASL and age-dependent physiological characteristics in pediatric brain perfusion. We then review the clinical implementation of ASL in pediatric neurological diseases, including vascular diseases, brain tumors, acute encephalopathy with biphasic seizure and late reduced diffusion (AESD), and migraine. In moyamoya disease, ASL can be used for brain perfusion and vessel assessment in pre- and post-treatment. In arteriovenous malformations, ASL is sensitive to detect small degrees of shunt. Furthermore, in vascular diseases, the implementation of ASL-based time-resolved MR angiography is described. In neoplasms, ASL-derived CBF has a high diagnostic accuracy for differentiation between low- and high-grade pediatric brain tumors. In AESD and migraine, ASL may allow for accurate early diagnosis and provide pathophysiological information.

Analysis of the targeting uncertainty of a stereotactic frameless radiosurgery technique for arteriovenous malformation

In order to target arteriovenous malformations (AVM) in a frameless approach, registration of two-dimensional (2D) digital-subtracted-angiographs (DSA) with three-dimensional (3D) computed tomography (CT) is required. Targeting accuracy and delineation of a frameless 2D-DSA and 3D-CT image registration tool based on bony anatomy of the skull was evaluated. This frameless approach assures accurate target localization and can be used in a clinical setting.

Planning evaluation of C-arm cone beam CT angiography for target delineation in stereotactic radiation surgery of brain arteriovenous malformations

PURPOSE

Stereotactic radiation surgery (SRS) is one of the therapeutic modalities currently available to treat cerebral arteriovenous malformations (AVM). Conventionally, magnetic resonance imaging (MRI) and MR angiography (MRA) and digital subtraction angiography (DSA) are used in combination to identify the target volume for SRS treatment. The purpose of this study was to evaluate the use of C-arm cone beam computed tomography (CBCT) in the treatment planning of SRS for cerebral AVMs.

METHODS AND MATERIALS

Sixteen consecutive patients treated for brain AVMs at our institution were included in this retrospective study. Prior to treatment, all patients underwent MRA, DSA, and C-arm CBCT. All images were coregistered using the GammaPlan planning system. AVM regions were delineated independently by 2 physicians using either C-arm CBCT or MRA, resulting in 2 volumes: a CBCT volume (VCBCT) and an MRA volume (VMRA). SRS plans were generated based on the delineated regions.

RESULTS

The average volume of treatment targets delineated using C-arm CBCT and MRA were similar, 6.40 cm(3) and 6.98 cm(3), respectively (P=.82). However, significant regions of nonoverlap existed. On average, the overlap of the MRA with the C-arm CBCT was only 52.8% of the total volume. In most cases, radiation plans based on VMRA did not provide adequate dose to the region identified on C-arm CBCT; the mean minimum dose to VCBCT was 29.5%, whereas the intended goal was 45% (P<.001). The mean volume of normal brain receiving 12 Gy or more in C-arm CBCT-based plans was not greater than in the MRA-based plans.

CONCLUSIONS

Use of C-arm CBCT images significantly alters the delineated regions of AVMs for SRS planning, compared to that of MRA/MRI images. CT-based planning can be accomplished without increasing the dose to normal brain and may represent a more accurate definition of the nidus, increasing the chances for successful obliteration.

Volumetric analysis of intracranial arteriovenous malformations contoured for CyberKnife radiosurgery with 3-dimensional rotational angiography vs computed tomography/magnetic resonance imaging

BACKGROUND

Accurate target delineation has significant impact on brain arteriovenous malformation (AVM) obliteration, treatment success, and potential complications of stereotactic radiosurgery.

OBJECTIVE

We compare the nidal contouring of AVMs using fused images of contrasted computed tomography (CT) and magnetic resonance imaging (MRI) with matched images of 3-dimensional (3-D) cerebral angiography for CyberKnife radiosurgery (CKRS) treatment planning.

METHODS

Between May 2009 and April 2012, 3-D cerebral angiography was integrated into CKRS target planning for 30 consecutive patients. The AVM nidal target volumes were delineated using fused CT and MRI scans vs fused CT, MRI, and 3-D cerebral angiography for each patient.

RESULTS

The mean volume of the AVM nidus contoured with the addition of 3-D cerebral angiography to the CT/MRI fusion (9.09 cm(3), 95% confidence interval: 5.39 cm(3)-12.8 cm(3)) was statistically smaller than the mean volume contoured with CT/MRI fused scans alone (14.1 cm(3), 95% confidence interval: 9.16 cm(3)-19.1 cm(3)), with a mean volume difference of δ = 5.01 cm(3) (P = .001). Diffuse AVM nidus was associated with larger mean volume differences compared with a compact nidus (δ = 6.51 vs 2.11 cm(3), P = .02). The mean volume difference was not statistically associated with the patient's sex (male δ = 5.61, female δ = 5.06, P = .84), previous hemorrhage status (yes δ = 5.69, no δ = 5.23, P = .86), or previous embolization status (yes δ = 6.80, no δ = 5.95, P = .11).

CONCLUSION

For brain AVMs treated with CKRS, the addition of 3-D cerebral angiography to CT/MRI fusions for diagnostic accuracy results in a statistically significant reduction in contoured nidal volume compared with standard CT/MRI fusion-based contouring.

Computer-aided nidus segmentation and angiographic characterization of arteriovenous malformations

PURPOSE

Exact knowledge about the nidus of an arteriovenous malformation (AVM) and the connected vessels is often required for image-based research projects and optimal therapy planning. The aim of this work is to present and evaluate a computer-aided nidus segmentation technique and subsequent angiographic characterization of the connected vessels that can be visualized in 3D.

METHODS

The proposed method was developed and evaluated based on 15 datasets of patients with an AVM. Each dataset consists of a high-resolution 3D and a 4D magnetic resonance angiography (MRA) image sequence. After automatic cerebrovascular segmentation from the 3D MRA dataset, a voxel-wise support vector machine classification based on four extracted features is performed to generate a new parameter map. The nidus is represented by positive values in this parameter map and can be extracted using volume growing. Finally, the nidus segmentation is dilated and used for an automatic identification of feeding arteries and draining veins by integrating hemodynamic information from the 4D MRA datasets.

RESULTS

A quantitative comparison of the computer-aided AVM nidus segmentation results to manual gold-standard segmentations by two observers revealed a mean Dice coefficient of 0.835, which is comparable to the inter-observer agreement for which a mean Dice coefficient of 0.830 was determined. The angiographic characterization was visually rated feasible for all patients.

CONCLUSION

The presented computer-aided method enables a reproducible and fast extraction of the AVM nidus as well as an automatic angiographic characterization of the connected vessels, which can be used to support image-based research projects and therapy planning of AVMs.

Magnetic resonance imaging/magnetic resonance angiography fusion technique for intraoperative navigation during microsurgical resection of cerebral arteriovenous malformations

OBJECT

Microsurgical resection of arteriovenous malformations (AVMs) is facilitated by real-time image guidance that demonstrates the precise size and location of the AVM nidus. Magnetic resonance images have routinely been used for intraoperative navigation, but there is no single MRI sequence that can provide all the details needed for characterization of the AVM. Additional information detailing the specific location of the feeding arteries and draining veins would be valuable during surgery, and this detail may be provided by fusing MR images and MR angiography (MRA) sequences. The current study describes the use of a technique that fuses contrast-enhanced MR images and 3D time-of-flight MR angiograms for intraoperative navigation in AVM resection.

METHODS

All patients undergoing microsurgical resection of AVMs at the Dartmouth Cerebrovascular Surgery Program were evaluated from the surgical database. Between 2009 and 2011, 15 patients underwent surgery in which this contrast-enhanced MRI and MRA fusion technique was used, and these patient form the population of the present study.

RESULTS

Image fusion was successful in all 15 cases. The additional data manipulation required to fuse the image sets was performed on the morning of surgery with minimal added setup time. The navigation system accurately identified feeding arteries and draining veins during resection in all cases. There was minimal imaging-related artifact produced by embolic materials in AVMs that had been preoperatively embolized. Complete AVM obliteration was demonstrated on intraoperative angiography in all cases.

CONCLUSIONS

Precise anatomical localization, as well as the ability to differentiate between arteries and veins during AVM microsurgery, is feasible with the aforementioned MRI/MRA fusion technique. The technique provides important information that is beneficial to preoperative planning, intraoperative navigation, and successful AVM resection.

Comparisons of DSA and MR angiography with digital subtraction angiography in 151 patients with subacute spontaneous intracerebral hemorrhage

To exclude underlying vascular abnormalities in patients with spontaneous intracerebral hemorrhage, the traditional paradigm requires investigation using digital subtraction angiography (DSA) in both the acute and subacute phases. We investigated whether MRI and magnetic resonance angiography (MRA), in the subacute stage of intracerebral hematoma, had high positive predictive values (PPV) and negative predictive values (NPV) in screening for vascular abnormality in the routine clinical setting. In a regional neurosurgical center in Hong Kong, we retrospectively reviewed 151 patients investigated with both MRI and DSA for underlying structural vascular abnormalities during the subacute phase. Sensitivity, specificity, and intermodality agreement were assessed. A total of 70/151 (46%) vascular lesions accountable for the hemorrhage were found. Patients with vascular abnormalities tended to be younger (mean age+/-standard deviation [SD], 33+/-15years), less likely to be hypertensive (6.3%), and the lesion was more likely to be accompanied by intraventricular hemorrhage (22%). In terms of cerebral arteriovenous malformation and dural arteriovenous fistulas, MRI/MRA had a PPV of 0.98 and a NPV of 1.00. We concluded that MRI/MRA was able to detect most structural vascular abnormalities in the subacute phase in most patients and, thus, its use is recommended as the screening test.

Obliteration dynamics in cerebral arteriovenous malformations after cyberknife radiosurgery: quantification with sequential nidus volumetry and 3-tesla 3-dimensional time-of-flight magnetic resonance angiography

OBJECTIVE

To investigate the time-dependent obliteration of cerebral arteriovenous malformations (cAVM) after CyberKnife radiosurgery (CKRS) (Accuray, Inc., Sunnyvale, CA) by means of sequential 3-T, 3-dimensional (3D), time-of-flight (TOF) magnetic resonance angiography (MRA), and volumetry of the arteriovenous malformation (AVM) nidus.

METHODS

In this prospective study, 3D TOF MRA was performed on 20 patients with cAVMs treated by single-fraction CKRS. Three-dimensional TOF MRA was performed on a 3-T, 32-channel magnetic resonance scanner (Magnetom TIM Trio; Siemens Medical Solutions, Erlangen, Germany) with isotropic voxel size at a spatial resolution of 0.6 x 0.6 x 0.6 mm3. The time-dependent relative decay of the transnidal blood flow evidenced by 3D TOF MRA was referred to as "obliteration dynamics." Volumetry of the nidus size was performed with OsiriX imaging software (OsiriX Foundation, Geneva, Switzerland). All patients had 3 to 4 follow-up examinations at 3- to 6-month intervals over a minimum follow-up period of 9 months. Subtotal obliteration was determined if the residual nidus volume was 5% or less of the initial nidus volume. Stata/IC software (Version 10.0; Stata Corp., College Station, TX) was used for statistical analysis and to identify potential factors of AVM obliteration.

RESULTS

Regarding their clinical status, case history, and pretreatments, the participants of this study represent difficult-to-treat cAVM patients. The median nidus volume was 1.8 mL (range, 0.4-12.5 mL); the median minimum dose prescribed to the nidus was 22 Gy (range, 16-24 Gy) delivered to the 67% isodose line (range, 55-80%). CKRS was well tolerated, with complications in 2 patients. No further hemorrhages occurred after RS, except 1 small and clinically inapparent incident. The median follow-up period after RS was 25.0 months (range, 11.7-36.8 months). After RS, a statistically significant obliteration was observed in all patients. However, the obliteration dynamics of the cAVMs showed a pronounced variability, with 2 types of post-therapeutic behavior identified. cAVMs of Group A showed a faster reduction of transnidal blood flow than cAVMs in Group B. The median time to subtotal obliteration was 23.8 months for all patients, 11.6 months for patients in Group A, and 27.8 months for patients in Group B (P = 0.05). Logistic regression analysis revealed dose homogeneity and the circumscribed isodose to be the only variables (P < 0.01) associated with the obliteration dynamics in this study. The cumulative complete angiographic obliteration rate was 67% (95% confidence interval, 32-95%) 2 years after RS.

CONCLUSION

The use of sequential 3D TOF MRA at 3 T and nidus volumetry enables a noninvasive quantitative assessment of the dynamic obliteration process induced by CKRS in cAVMs. This method may be helpful to identify factors related to AVM obliteration after RS when larger patient cohorts become available.

A methodology for validating a 3D imaging modality for brain AVM delineation: application to 3DRA

A general methodology is described to validate a 3D imaging modality with respect to 2D digital subtracted angiography (DSA) for brain AVMs (BAVM) delineation. It relies on the assessment of the statistical compatibility of the radiosurgical target delineated in 3D with its delineations in 2D. This methodology is demonstrated through a preliminary evaluation of 3D rotational angiography (3DRA). Generally speaking, BAVM delineation cannot be performed on 3DRA alone. However, in our study, 3DRA showed similar performances to DSA for rather easy cases, and even better for three patients. Conversely, three problematic cases are identified and discussed.

Gamma knife surgery for arteriovenous malformations in the brain: integration of time-resolved contrast-enhanced magnetic resonance angiography into dosimetry planning

Object

The aim of this study was to develop an algorithm for the integration of time-resolved contrast-enhanced magnetic resonance (MR) angiography into dosimetry planning for Gamma Knife surgery (GKS) of arteriovenous malformations (AVMs) in the brain.

Methods

Twelve patients harboring brain AVMs referred for GKS underwent intraarterial digital subtraction (DS) angiography and time-resolved MR angiography while wearing an externally applied cranial stereotactic frame. Time-resolved MR angiography was performed on a 1.5-tesla MR unit (Achieva, Philips Medical Systems) using contrast-enhanced 3D fast field echo sequencing with stochastic central k-space ordering. Postprocessing with interactive data language (Research Systems, Inc.) produced hybrid data sets containing dynamic angiographic information and the MR markers necessary for stereotactic transformation. Image files were sent to the Leksell GammaPlan system (Elekta) for dosimetry planning.

Results

Stereotactic transformation of the hybrid data sets containing the time-resolved MR angiography information with automatic detection of the MR markers was possible in all 12 cases. The stereotactic coordinates of vascular structures predefined from time-resolved MR angiography matched with DS angiography data in all cases. In 10 patients dosimetry planning could be performed based on time-resolved MR angiography data. In two patients, time-resolved MR angiography data alone were considered insufficient. The target volumes showed a notable shift of centers between modalities.

Conclusions

Integration of time-resolved MR angiography data into the Leksell GammaPlan system for patients with brain AVMs is feasible. The proposed algorithm seems concise and sufficiently robust for clinical application. The quality of the time-resolved MR angiography sequencing needs further improvement.

Delineation of brain AVMs on MR-Angiography for the purpose of stereotactic radiosurgery

PURPOSE

To assess the dosimetric consequences of brain arteriovenous malformation (bAVM) delineation on magnetic resonance angiography (MRA) for the purpose of stereotactic radiosurgery.

METHODS AND MATERIALS

Three observers contoured a bAVM in 20 patients, using digital subtraction angiography (V(DSA)) and three-dimensional time-of-flight MRA (V(MRA)). Displacement between contours was calculated. Agreement and differences between observers and imaging modalities were assessed. A standardized treatment plan with dynamic conformal arcs was generated and dosimetric coverage of all contours and the volume of normal brain tissue within the high dose region was determined.

RESULTS

The generalized reliability coefficient was "fair" for target volume (0.79), but "poor" for displacement (0.35). V(MRA) was larger than V(DSA) (5.0 vs. 4.0 mL, p = 0.001). No difference in displacement was found (2.8 vs. 2.5 mm, p = 0.156). Dosimetric coverage of V(MRA) was 62.9% (95% CI, 56.9-68.8) when V(DSA) was used as planning target volume, and coverage of V(DSA) was 83.5% (95% CI, 78.1-88.8) when V(MRA) was used for planning (p < 0.001). The mean volume of normal brain within the 80% isodose was larger when the bAVM was delineated on MRA (0.7 vs. 1.0 mL (p = 0.02) for targets < or =3 mL and 3.7 vs. 7.0 mL (p = 0.01) for targets >3 mL).

CONCLUSIONS

Brain arteriovenous malformations delineated on MRA are larger and more randomly displaced. However, for bAVMs < or =3 mL, the difference in volume of normal brain tissue within the high-dose region does not seem to be clinically relevant. Therefore, MRA-images might be used as the sole imaging modality for the radiosurgical treatment of bAVMs < or =3 mL when the bAVM is located in a noneloquent position.

MR angiography fusion technique for treatment planning of intracranial arteriovenous malformations

PURPOSE

To develop an image fusion technique using elliptical centric contrast-enhanced (CE) MR angiography (MRA) and three-dimensional (3D) time-of-flight (TOF) acquisitions for radiosurgery treatment planning of arteriovenous malformations (AVMs).

MATERIALS AND METHODS

CE and 3D-TOF MR angiograms with disparate in-plane fields of view (FOVs) were acquired, followed by k-space reformatting to provide equal voxel dimensions. Spatial domain addition was performed to provide a third, fused data volume. Spatial distortion was evaluated on an MRA phantom and provided slice-dependent and global distortion along the three physical dimensions of the MR scanner. In vivo validation was performed on 10 patients with intracranial AVMs prior to their conventional angiogram on the day of gamma knife radiosurgery.

RESULTS

Spatial distortion in the phantom within a volume of 14 x 14 x 3.2 cm(3) was less than +/-1 mm (+/-1 standard deviation (SD)) for CE and 3D-TOF data sets. Fused data volumes were successfully generated for all 10 patients.

CONCLUSION

Image fusion can be used to obtain high-resolution CE-MRA images of intracranial AVMs while keeping the fiducial markers needed for gamma knife radiosurgery planning. The spatial fidelity of these data is within the tolerance acceptable for daily quality control (QC) purposes and gamma knife treatment planning.