As with any stereotactic technique, the cranium must be rigidly fixed with a frame which associates a reference coordinate system to an intracranial target, providing an accurate localization of the target utilizing only CT or additionally either MRI or angiographic data. All imaging data are either digitized first (angiographic data) or directly electronically transferred to the treatment planning workstation. We are currently using a Hewlett Packard Visualize C3000 treatment planning workstation. Either the BRW stereotactic headring (with or without the MRI-compatible adaptor) or the Gill-Thomas-Cosman (GTC) relocatable frame is utilized for rigid head fixation. The X-Knife software always utilizes CT data as a primary imaging modality due to the high spatial fidelity of CT, and the BRW localizer is utilized for the CT scan. MRI data, obtained before or after CT and without the frame, can be fused to CT data utilizing an image fusion algorithm developed by Radionics16. Fused images have the advantage of both high spatial fidelity and anatomic detail. Angiographic data are obtained utilizing the angiographic localizer, and treatment planning for AVMs is primarily based on digitized biplanar stereotactic angiographic data, although we are currently exploring the use of MR angiography as an additional image data file which promised to remove ambiguities created with 2-D angiographic images.

The case of interest is selected as a folder which invariably includes CT data and additional fused or angiographic data sets if obtained. Treatment planning commences with the default CT file or in the case of an AVM with the digitized stereotactic angiogram. In the CT-based cases, typically tumors, the 9 fiducial rod coordinates are first manually identified in one axial partial-screen CT image and subsequently autodetected throughout the axial CT file and reconstructed as a 3-dimensional reformatted image in “X-Knife space”, the 3-dimensional treatment-planning workspace also featured as a larger partial-screen image. If treatment planning involves an AVM, the fiducial coordinates in the angiographic localizer are identified. For CT-based cases, if fusion data are desired for treatment planning, one axial fused image slice is featured and the same 9 fiducial rod coordinates are again manually identified, autodetected throughout the remainder of the axial slice file, reconstructed and integrated into the pre-existing image established from the CT file. The lesion is then identified by the neurosurgeon in either CT or fused axial slices as are any critical dose-limiting normal structures or “anatomes”, and all contoured structures are reformatted, each color-coded, as three-dimensional structures in X-Knife space. At our institution, AVMs are identified and manually contoured on the cut film with a wax pencil by the neurosurgeon prior to digitization so the transferred images feature the nidus in each plane. Due to the constraints imposed by two-dimensional stereotactic angiography, other institutions have based AVM treatment planning on either CT or MRI data.

Treatment planning then commences. This task is customarily performed by the medical physicist but may include the neurosurgeon and /or the radiation oncologist. Treatment planning is constrained by the relationship between the volume of the lesion and a safe dose considered effective in the treatment of the lesion. This relationship was originally established by Kjellberg and serves as a fundamental treatment planning principle in radiosurgery. The ideal treatment plan involves a dose which is perfectly configured to the target (dose conformality) and homogeneous (dose homogeneity), although dose inhomogeneity remains a variable suggested but not unambiguously related to treatment-related morbidity. Recent measurements have been set forth by the Radiation Therapy Oncology Group (RTOG) which serve to assess dose conformality and homogeneity. These include the ratio of the prescription isodose volume to target volume as a reflection of conformality (the PITV ratio), and the ratio of maximum dose to prescribed dose ratio as a reflection of homogeneity (the MDPD ratio). Neither of these objectives (ideally in each case a ratio of 1) can be practically achieved due to the comparatively small collimator sizes and the limitations of circular collimators, particularly when utilized for the treatment of large irregularly-shaped lesions such as skull base meningiomas or large AVMs, examples of which are discussed below. A more formal and detailed description of target volumes and dose distributions within them can be found in a recent report furnished by the International Commission of Radiation Units and Measurements.

Tools provided by the X-Knife software maximize the liklihood of optimizing these important treatment planning objectives, thereby maximizing the therapeutic outcome and minimizing any treatment-related morbidities. As illustrations, we will provide examples of simple and complex treatment plans. A more detailed account of LINAC radiosurgery dosimetry can be found in a recent chapter by Kooy and colleagues.