Improved Cerebral Arteriovenous Malformation Obliteration With 3-Dimensional Rotational Digital Subtraction Angiography for Radiosurgical Planning: A Retrospective Cohort Study

The new era of artificial intelligence in neuroradiology: current research and promising tools

Radiology has a number of characteristics that make it an especially suitable medical discipline for early artificial intelligence (AI) adoption. These include having a well-established digital workflow, standardized protocols for image storage, and numerous well-defined interpretive activities. The more than 200 commercial radiologic AI-based products recently approved by the Food and Drug Administration (FDA) to assist radiologists in a number of narrow image-analysis tasks such as image enhancement, workflow triage, and quantification, corroborate this observation. However, in order to leverage AI to boost efficacy and efficiency, and to overcome substantial obstacles to widespread successful clinical use of these products, radiologists should become familiarized with the emerging applications in their particular areas of expertise. In light of this, in this article we survey the existing literature on the application of AI-based techniques in neuroradiology, focusing on conditions such as vascular diseases, epilepsy, and demyelinating and neurodegenerative conditions. We also introduce some of the algorithms behind the applications, briefly discuss a few of the challenges of generalization in the use of AI models in neuroradiology, and skate over the most relevant commercially available solutions adopted in clinical practice. If well designed, AI algorithms have the potential to radically improve radiology, strengthening image analysis, enhancing the value of quantitative imaging techniques, and mitigating diagnostic errors.

Nidus Compacity Determined by Semi-Automated Segmentation is a Strong Quantitative Predictor of Brain Arterio-Venous Malformation Cure

BACKGROUND AND OBJECTIVE

A compact nidus is a well-known feature of good outcome after treatment in brain arteriovenous malformations (bAVM). This item, included in the "Supplementary AVM grading system" by Lawton, is subjectively evaluated on DSA. The present study aimed to assess whether quantitative nidus compacity along with other angio-architectural bAVM features were predictive of angiographic cure or the occurrence of procedure-related complications.

MATERIALS AND METHODS

Retrospective analysis of 83 patients prospectively collected data base between 2003 to 2018 having underwent digital subtraction 3D rotation angiography (3D-RA) for pre-therapeutic assessment of bAVM. Angio-architectural features were analyzed. Nidus compacity was measured with a dedicated segmentation tool. Univariate and multivariate analyses were performed to test the association between these factors and complete obliteration or complication.

RESULTS

Compacity was the only significant factor associated with complete obliteration in our predictive model using logistic multivariate regression; the area under the curve for compacity predicting complete obliteration was excellent (0.82; 95% CI 0.71-0.90; p < 0.0001). The threshold value maximizing the Youden index was a compacity > 23% (sensitivity 97%; specificity 52%; 95% CI 85.1-99.9; p = 0.055). No angio-architectural factor was associated with the occurrence of a complication.

CONCLUSION

Nidus high compacity quantitatively measured on 3D-RA, using a dedicated segmentation tool is predictive of bAVM cure. Further investigation and prospective studies are warranted to confirm these preliminary results.

Frameless Co-Registration of Biplane 2D Digital Subtraction Angiography Whole Frames and 3D Rotational Angiography-Based Cone-Beam Computed Tomography Angiogram on Dedicated Software for Stereotactic Radiosurgery of Cranial Vascular Malformations

Purpose: Given its high spatial resolution and vasculature selectivity, the cone-beam computed tomography (CT) angiography (CBCTA) image acquired by selective 3D rotational angiography (3DRA) is the most suitable 3D image for the target definition of stereotactic radiosurgery (SRS) for intracranial arteriovenous malformations (AVMs) and dural arteriovenous fistulas (DAVFs). Furthermore, the relatively low temporal resolution of 3DRA-based CBCTA can be complemented by the stereotactic co-registration of orthogonally paired 2D dynamic digital subtraction angiography (2D-DSA). The integration of 2D-DSA, which is usually limited to one or a few frames for each projection, into CBCTA and/or planning CT can be achieved only by catheter-directed angiography on the day of SRS via a dedicated image localizer under rigid frame fixation to the skull, which imposes substantial burdens on patients. This study aimed to demonstrate a novel, convenient, and significantly less invasive method for the frameless co-registration of biplane 2D-DSA whole frames and CBCTA on commercially available dedicated software, namely, Brainlab^® Elements (Brainlab AG, Munich, Germany), and present its prerequisite for successful image fusion.

Technical Report: Elements have afforded the following functionality: A 3D vasculature image is automatically extracted as a floating image from any 3D image series containing vascular details and then subsequently co-registered manually and automatically to a selected frame pair of 2D-DSA with a six-degree-of-freedom rigid registration. As a preclinical feasibility study, two anonymous image datasets from patients harboring cerebral AVM and transverse-sigmoid (TS) DAVF were used to verify the accuracy and practicality of Elements for the frameless co-registration of 2D/3D vascular images, particularly on the assumption of clinical workflow for the target delineation of SRS planning. The use of ordinary unsubtracted CBCTA resulted in the insufficient extraction of abutting vessels or vessels that are in close proximity to bony structures, particularly in the case of TS-DAVF, where the fistulous pouch and the affected venous sinuses were adjacent to the cranial bone. By contrast, the amount and selectivity of vasculatures and the accuracy of subsequent image fusion were significantly improved from the subtracted CBCTA. The integration of CBCTA into dynamic 2D-DSA allowed the simultaneous review of both image information by sharing any concerning point and 2D or 3D structures under a common 3D coordinate.

Conclusions: Elements enable the clinically useful frameless co-registration of biplane 2D-DSA whole frames into CBCTA, for which the routine acquisition of both subtracted and unsubtracted CBCTA axial images for ordinary diagnostic purposes is an indispensable prerequisite for successful image fusion and further widespread application. This frameless integration of the 2D/3D angiogram would dramatically enhance both the frame-based and frameless SRS workflow and circumstances by allowing users to forward SRS planning well in advance before SRS, along with the omission of invasive angiography on the day of SRS, and would broaden the implementation of frameless SRS. Furthermore, the comprehensive alternating interactive review of the 2D/3D integrated angiogram leads to a more in-depth quasi-4D understanding of the affected angioarchitectures compared with the separate viewing of each image.

Hemodynamic changes during the obliteration process for cerebral arteriovenous malformations after radiosurgery

OBJECTIVE

The process of cerebral arteriovenous malformation (AVM) obliteration following radiosurgery is poorly understood. Authors of this retrospective study aimed to assess the changes in AVM hemodynamics after stereotactic radiosurgery (SRS) by using 3D flow magnetic resonance imaging (MRI) to elucidate the process of AVM obliteration.

METHODS

Twenty-four patients with AVMs treated with SRS between July 2015 and December 2017 were included in this study and classified into two groups depending on the duration of AVM obliteration: group A, obliteration within 3 years (n = 15); and group B, obliteration taking more than 3 years or no obliteration (n = 9). Blood flow (ml/min) in the largest feeding artery was measured before and after SRS by using time-averaged 3D flow MRI. The decreasing rate of blood flow in the feeding artery after SRS was calculated as the percent change from baseline blood flow. A Wilcoxon rank-sum test was used to compare the decreasing blood flow rate between the two groups at 4 and 12 months after SRS.

RESULTS

For the entire cohort, the mean decrease in blood flow in the feeding artery from baseline was 29% at 4 months and 71% at 12 months after SRS. In general, blood flow after SRS decreased faster in group A and slower in group B. The decreasing rates in blood flow at 4 and 12 months after SRS were significantly different between the two groups (p = 0.02 and < 0.001, respectively).

CONCLUSIONS

Tracking changes in AVM hemodynamics after SRS may be useful for assessing the progress of AVM obliteration and the therapeutic effects of SRS, possibly contributing to the prediction of subsequent obliteration outcome.

Effects of cone versus multi-leaf collimation on dosimetry and neurotoxicity in patients with small arteriovenous malformations treated by stereotactic radiosurgery

PURPOSE/OBJECTIVE

Linear accelerator (LINAC) based stereotactic radiosurgery (SRS) for arteriovenous malformations (AVMs) is delivered with cone or multileaf collimators (MLCs), and favorable dosimetry is associated with reduced radionecrosis in normal brain tissue. This study aims to determine whether cones or MLCs has better dosimetric characteristics, to predict differences in toxicity.

METHODS

All patients treated for AVMs using LINAC SRS from 2003-2017 were examined retrospectively. Demographic data, volumes of normal tissue exposed to 12Gy (V12Gy[cc]) and 4Gy (V4Gy[cc]), maximal dose, and dose gradient were analyzed. Univariate and multivariate analyses were used to evaluate relationships between collimator type, dosimetric parameters, and toxicity. Propensity score matching was used to adjust for AVM size.

RESULTS

Compared to MLC, cones were independently associated with reduced V12Gy[cc] after propensity score matching (p=0.008) and reduced neurotoxicity (p=0.016). Higher V12Gy[cc] (p=0.0008) and V4Gy[cc] (p=0.002) were associated with increased neurotoxicity.

CONCLUSIONS

Treating AVMs with cone-based SRS over MLC-based SRS may improve dosimetry and reduce toxicities.