Radiation Physics: Stereotactic

History of Stereotactic Irradiation

The field of stereotactic irradiation owes its origins to parallel development in two disparate medical disciplines dating back over a century: those of stereotactic surgery and external beam radiation. In the late 1800s, surgeons and anatomists began exploring quantitative methods for studying the structure and function of the brain. The methodology known as stereotaxis relies on external fiducials to define a three-dimensional frame of reference in which targeting can be performed. Historically, the fiducials consisted of a head frame that was rigidly attached to the skull. Because the skull and cranial structures remained fixed with respect to the frame, the external fiducials served as an appropriate surrogate for intracranial localization. The stereotactic method was pioneered by Horsley and Clarke, who devised an instrument for producing lesions at exact locations within the brains of nonhuman subjects. Other early efforts included those of Mussen and Kirschner. Routine clinical application of stereotactic surgery in humans is credited to the team of Speigel and Wycis. Lars Leksell subsequently developed a frame based on polar rather than Cartesian coordinates, providing maximum flexibility in choosing entry point and trajectory. The commercial Leksell frame (Elekta AB, Stockholm, Sweden) remains identical in function to the 1949 device. Other developments in stereotactic frames include the efforts of Talairach, Narabayashi, Riechert, and Mundinger, and Brown, Roberts, Todd, and Wells. Brown's modifications to the original Wells-Todd device resulted in the Brown-Roberts-Wells (BRW) coordinate system that is available commercially from Integra Radionics (Burlington, MA) and Brainlab (Feldkirchen, Germany). Examples of commercial stereotactic localization devices are shown in eFig. 7.1 . In the era of image guidance, the skull serves as its own set of fiducials, and head frames are no longer required.

The discovery of x-rays by Wilhelm Roentgen in late 1895 is well known. Throughout subsequent decades, efforts to increase the penetration ability of x-rays produced with conventional tubes met with modest success, with tube potentials limited to approximately 200 kV. In the early 1950s, Leksell realized that external forms of energy, such as ultrasound and ionizing radiation, could be coupled to a stereotactic frame. In 1951, Leksell proposed the term stereotactic radiosurgery to refer to the delivery of a single high dose of radiation accurately directed to an intracranial target. The principles of dose localization require irradiation “through a large number of small portals…which all meet and cross in the structure in question.” Leksell also appreciated the limitations imposed by low-energy x-rays; his subsequent efforts, as well as those of groups located in Boston, Massachusetts, and Berkeley, California, utilized beams of protons and other light ions. In the mid-1960s, Leksell's colleagues at the Karolinska Institute in Stockholm, Sweden, began efforts on what would become the first dedicated radiosurgery device. The initial Gamma Knife, installed in Sofiahemmet Hospital in 1967, used 179 rectangular-collimated ⁶⁰ Co sources; subsequent commercial versions were implemented with 201 circular-collimated sources. The Perfexion unit, consisting of 192 cobalt sources arranged in eight independent sectors, was introduced by Elekta in 2007. Each sector can move to one of 4 locations to choose between blocked, 4-, 8-, and 16-mm circular collimation. eFig. 7.2 shows the evolution of the commercial Gamma Knife, from the Model U to the present Perfexion and Icon.

The development of technology for linear accelerator-based radiosurgery began in the early 1980s, with the first radiosurgery patient treated on a modified medical linear accelerator (LINAC) in 1982. All early LINAC systems used hardware modifications designed to facilitate stereotactic frames and improve targeting accuracy to overcome shortcomings with the treatment couch. Subsequent improvement in LINAC couches allowed the stereotactic frame to be affixed directly to the couch top. Subsequent developments included LINACs dedicated for radiosurgery, including the CyberKnife (Accuray, Sunnyvale, CA), 600SR (Varian Medical Systems, Palo Alto, CA), Novalis (BrainLAB AG, Heimstetten, Germany), and C-arm multi-rotation-axis LINAC (Mitsubishi Electric Ltd., Tokyo, Japan). Modern LINACs, such as the TrueBeam STx and EDGE (Varian Medical Systems, Palo Alto, CA), and Axesse and Versa HD (Elekta AB, Stockholm, Sweden) have localization and dosimetric characteristics that make them very well suited for radiosurgery applications. eFig. 7.3 shows the evolution of the radiosurgery LINACs, including the original Betti device, an early floor stand, the original CyberKnife, the original Novalis, and current offerings from Varian and Elekta.

Principles of Stereotactic Irradiation

There are several key principles that make SRS distinct from conventional radiotherapy. First, the procedure involves the delivery of an ablative dose of radiation that overwhelms the capacity of the irradiated cells to survive. The biological processes associated with high-dose delivery are quite different from those occurring at ~ 2 Gy per fraction; an increasing number of investigators have alluded to these threshold effects . The ablative intent implies that the approach is meant to treat gross disease exclusively and not microscopic extension. The accompanying need for tight margins places demands on both physical and dosimetric accuracy. The superior contrast resolution provided by magnetic resonance (MR) necessitates its use in most cranial indications, albeit with careful consideration of potential spatial distortions.

Second, the accuracy and precision of both spatial and dosimetric localization are essential. Prior to the advent of computed tomography (CT), stereotactic localization was performed using a pair of projection radiographs. Fiducials located on the entrance and exit surfaces of the localization device facilitated calculation of both target location and magnification from a pair of radiographs ( Fig. 7.1 ). Studies have demonstrated that the target point can be determined to within 0.3 mm over a large range of source-to-target and source-to-image distance and projection angles using such a technique. The obvious shortcoming of the approach is that it cannot adequately reconstruct a target volume.

The principle of stereotactic localization based on tomographic images is illustrated in Fig. 7.2 . The superior-inferior dimension (z) is defined by measuring the in-plane distance between diverging fiducials (y) and applying simple geometry:

With modern imaging and head frames, localization accuracy on the order of 1 to 2 mm is achievable. There are numerous sources of geometric uncertainty, including mechanical and radiation device isocentricity, imaging resolution, target delineation, and the stereotactic frame itself. The report of American Association of Physicists in Medicine (AAPM) Task Group 42 specifies an overall localization uncertainty of 2.4 mm (1 σ), provided that the planning images use a sufficiently thin-slice thickness.

Dosimetric localization requires the delivery of many beams from many directions, overlapping only at the target of interest. Ideally, the intensity of any individual beam is sufficiently low that little damage occurs along the beam path, except at the point that all beams intersect. Conformality of the high dose and target volumes is essential, as is compactness of the overall dose distribution. Because the intermediate dose volume increases rapidly as a function of target size, single-fraction cranial applications are largely limited to lesions smaller than 4 cm.

Dosimetry of Radiosurgery Beams

Dosimetric integrity—that is, the ability to accurately deliver the intended dose distribution to the desired volume—is a critical element of radiosurgery. Dosimetric integrity begins with the accurate measurement of photon beam characteristics. In general, SRS planning systems require the measurement of depth dose characteristics (percent depth dose [PDD] or tissue maximum ratio [TMR]), off-axis profiles, and relative output factors. For systems equipped with a micro-MLC, measurement of leaf transmission and dynamic leaf gap is also required. For commissioning Monte Carlo (MC) algorithms, additional beam measurements, including central and off-axis profiles and output factors in air, are often required. In-air scans require a brass build-up cap suitable for the beam energy used.

Considerations in Small-Field Dosimetry

The measurement of beam characteristics (relative output factors, PDDs, and profiles) in small fields is challenging owing to their inherent steep dose gradients. A misunderstanding of detector requirements or misinterpretation of measured beam data can lead to injurious or even life-threatening consequences. There is no accepted criterion defining a “small field,” but, in general, a small field can be labeled so depending on the electron range in the irradiated medium, the detector dimensions, and/or partial obstruction of the radiation source through collimation.

Lateral electronic equilibrium becomes compromised when the beam radius is comparable in size to the maximum electron range. The maximum electron range depends on the photon beam energy and the irradiated medium composition (i.e., density and atomic number). At this point, electrons leaving the central portion of the field are insufficiently replaced by electrons scattered from the surrounding medium. Output factors for small fields are difficult to measure in these conditions owing to spectral differences causing stopping power ratios and correction factors to be unclear. AAPM Task Group 65 recommends a general rule-of-thumb for the lateral electron range being about one-third of the forward range, and in a 6-MV x-ray beam, electron disequilibrium has been shown to occur for radii <1.0 cm. Below a certain field size, collimating devices can partially block the radiation source from the detector's field of view. Clearly, the output from the partially blocked source will be reduced compared to fields in which the detector views the complete source. The partially obscured focus also leads to the field size, as defined by the 50% isodose, to reduce more rapidly than the collimated aperture.

The measurement of small-field beam characteristics requires a small detector. The rationale for selection of an appropriately small detector size and the consequences of using too large a detector differ for relative output factor, depth dose, and beam profile measurements. Relative output factors are commonly defined as the ratio of dose for a field size of interest (a small field) over a reference field size (a larger field) at a fixed depth in medium. Relative output factors are specified on the central axis. The detector of choice must be placed on the central axis and must only intercept central axis fluence. This is easily achieved for a large reference field size with a flat beam profile around the central axis, but much greater consideration is necessary for a small field with steep dose gradients in close proximity to the central axis. The cross-sectional dimensions of the detector cannot be larger than the flat portion of the beam profile (crossline and inline) around the central axis. Appropriate detector selection for a relative output factor measurement of a 5 × 5 cm ² field normalized to a 10 ×10 cm ² reference field is illustrated in Fig. 7.3A . The air ionization chamber fits within the flat portion of the beam profile around the central axis for both fields. Fig. 7.3B illustrates an inappropriate detector selection for a relative output factor measurement of a 0.5-cm diameter circular field normalized to a 10 ×10 cm ² reference field. The air ionization chamber is much longer than the flat portion of the beam profile around the central axis of the 0.5-cm diameter field and the chamber intercepts the shoulder and tail of the beam profile in addition to the central axis. This results in an output factor measurement that is much smaller than reality and, if used clinically, will result in an overdose to patients. Relative output factors measured with a Farmer-type air ionization chamber and a silicon diode for square fields from 6 to 80 mm are presented in Table 7.1 for comparison. Output factors must be measured for each system independently, and values in Table 7.1 should not be used as a standard. When selecting a detector to measure relative output factors, one should always confirm that the detector dimensions are at least as small as the flat portion of the beam profile around the central axis. For SRS and SBRT beams, AAPM Task Group 101 recommends use of a detector with a maximum dimension of 1 mm.

TABLE 7.1

Example of Relative Output Factors Measured in Small Photon Fields ^a

Square Field Size (mm)	Output Factor
	Stereotactic Diode	Farmer-Type Chamber
6.0	0.599	0.126
12.0	0.775	0.420
18.0	0.824	0.647
42.0	0.907	0.900
60.0	0.945	0.940
80.0	0.988	0.975
100.0	1.000	1.000

a The Farmer-type ionization is much larger than the beam, averaging over the shoulder and tail of the beam profile in addition to the central axis, and resulting in grossly underestimated output factor.

PDD measurements for small fields require small detectors because, by definition, PDD describes the change in relative dose with depth on the central axis only. As in the case of relative output factor measurements, a detector wider than the flat portion of the beam profile (see Fig. 7.3B ) at any depth will give an inaccurate PDD by including the central axis fluence as well as the shoulder and tail regions. Fig. 7.4 shows PDD scans for a 6-mm diameter circular field scanned with a 0.6-cc Farmer-type air ionization chamber (30013, PTW), a 0.015-cc pinpoint air ionization chamber (31014, PTW), and a silicon diode (SFD, DEB050, IBA Dosimetry). Larger chambers overmeasure the PDD at depth owing to integration of dose lateral to the central axis.

Small detectors are required to accurately measure beam profiles of small fields to resolve steep dose gradients in the penumbra region. Fig. 7.5A (crossline) and Fig. 7.5B (inline) show beam profiles for a 6.0-mm diameter circular field measured with a Farmer-type air ionization chamber (dashed) and a stereotactic diode (solid). The larger air ionization chamber smears out the profile (penumbra broadening) because it integrates ionization over its 0.6-cc (6-mm diameter, 24-mm length) measurement volume while the diode integrates over its 0.6-mm cross-section. The shape of a beam profile ultimately affects the apparent dose coverage of a target. Overcorrecting the shoulder region to achieve target coverage will increase the dose delivered to surrounding tissues.