SU-E-T-280: DSA as the Primary Image Modality in Frameless SRS for Treatment of Arteriovenous Malformations

Purpose: Stereotactic radiosurgery(SRS) is an effective alternative to microsurgical resection or embolization for arteriorvenous malformations (AVMs). Digital Subtraction Angiography(DSA) is the gold standard for diagnosis and characterization of vascular anatomy, but it requires a frame attached to the skull. Proton and the new image‐guidedLINACSRS systems are frameless. These frameless systems represent a major progress in radiotherapy — however, they prevent the usage of DSA. Without DSA as the primary image, the outcome may be less‐than‐optimal and sometimes even impossible to treat. This study intends to solve the problem. Methods: The solution is based on a conceptual breakthrough whereby, in contrast to the previous practice where a localizer was fixed to the skull during angiography, the patient can now be angiographed under standard clinical conditions without frame and localizer, while a newly‐designed localizer box is radiographed separately. Based on several fiducials implanted into the skull, a target volume on DSA can be transferred to the localizer system in 3D, then further transferred to contours in multi‐CT slices. Combined with other imaging modalities, a treatment can be planned and executed by a routine frameless SRS. Results: Phantom tests demonstrated that the uncertainty in transferring a point target to the localizer is less than 0.3 mm. The maximum transformation errors from the localizer to CT system are ∼0.3 mm. The overall uncertainty of placing a point target to CT is 0.4 mm. For volume targets the volume determined is accurate within < 6%. The algorithm and software are robust and accurate. The method has been successfully applied clinically. Conclusions: A new reliable method for frameless Stereotactic Radiosurgery has been developed, which allows using DSA as the primary imaging in AVM treatment. Implanted fiducials are required for the time being, but using non‐implanted fiducials is feasible. This work is partially supported by the JCRT Foundation.