Stereotactic Radiosurgery · Technical Foundations

Platforms & Delivery

Cobalt, gantry linac, robotic, and particle systems: one concept, different physics

Every radiosurgery platform exists to do the same thing: concentrate an ablative dose on a stereotactic target with a steep gradient. They differ in radiation source, how they shape and aim the beams, whether the patient is frame-fixed or image-guided, and whether they can treat the body as well as the head. This page compares them on those axes, platform-neutral, so that a choice can be made on target size, location, fractionation, and resources rather than on brand.

Timoteo Almeida, MD, MSc, PhD, FCNS | Departments of Neurological Surgery and Radiation Oncology

Orientation

It is tempting to rank platforms, but for most indications the evidence does not support one delivery technology as categorically superior; well-commissioned systems achieve clinically comparable local control when used within their design envelope, even though measured conformity, gradient, and normal-brain dose do differ modestly between them in planning studies. What genuinely differs is fit to the problem: a dedicated cobalt unit excels at many small, complex, or multiple cranial targets in a single frame-fixed session; a gantry linac is a versatile generalist that also reaches the body; a robotic linac brings frameless motion tracking; and particle beams offer a physical dose-distribution advantage at a cost in access. Knowing each system's geometry explains its strengths far better than marketing does.

Part I

Gamma Knife (Cobalt-60)

1.How it works and where it shines

The Gamma Knife is a dedicated cranial radiosurgery device in which many fixed cobalt-60 sources are arrayed so their gamma beams converge on a single focal point. In the Perfexion, Icon, and current Esprit lineage, 192 sources are grouped into eight movable sectors (24 sources each) that index over 4-, 8-, and 16-mm collimators or are blocked, allowing each sector to be shaped or weighted independently. A treatment is built from multiple overlapping isocenters ("shots") of differing collimator sizes, producing a dose cloud with a very steep gradient outside the target. In multi-platform planning comparisons that steep falloff and the low normal-brain dose are the platform's most consistently reproduced dosimetric signature, while conformity indices are comparable to or slightly worse than a high-definition MLC linac or a robotic system. The steep gradient is the main reason the platform suits small, irregular, or critically located cranial and skull-base targets. For multiple metastases specifically the comparative picture is less settled: a randomized comparison of Gamma Knife and Edge-based radiosurgery found equivalent local control, and cohort and planning data on radionecrosis and normal-brain dose favor neither platform consistently. Perfexion is frame-based; Icon and Esprit add cone-beam CT, thermoplastic-mask fixation, and intrafraction motion management, enabling image-guided frameless and hypofractionated cranial treatment while retaining frame-based workflows. The trade-offs are that the platform is dedicated to the head, cobalt sources decay and require periodic replacement, and complex many-isocenter plans can be time-intensive.

Practice preference (not a universal rule) For complex and benign cranial work (vestibular schwannomas, skull-base meningiomas, multiple small metastases, and functional targets) my preference is Gamma Knife, for its many-isocenter conformality and steep gradient near cranial nerves and brainstem. This is a preference, not evidence of platform superiority; comparable results are achievable on a well-commissioned linac, and the right choice depends on the target, the fractionation needed, and local expertise.

Part II

LINAC-Based Systems (Cranial and Body)

2.The versatile generalist

A modern medical linear accelerator delivers radiosurgery with a megavoltage photon beam shaped by a high-resolution multileaf collimator and delivered as dynamic arcs (VMAT), often in flattening-filter-free mode for high dose rate and short treatment times. Localization is image-guided and frameless: cone-beam CT, stereoscopic kV imaging with surface tracking (e.g., ExacTrac Dynamic, which combines stereoscopic kV X-ray with thermal-surface tracking), and optical surface monitoring for setup and intrafraction tracking, typically with a thermoplastic mask for cranial and a body frame/vac-bag for trunk targets. Representative configurations include TrueBeam STx and Edge-class systems, historically marketed with Brainlab as Novalis. For multiple cranial targets, semi-automated non-coplanar planning packages such as Varian HyperArc and Brainlab Elements deliver several metastases from a single isocenter, which shortens delivery substantially; the trade-off is that rotational setup error translates into larger positional offsets as target distance from isocenter increases, so six-degree-of-freedom correction and tight rotational tolerances are required. The defining advantage is versatility: the same machine treats cranial SRS, hypofractionated cranial SRT, and, crucially, spine and body SBRT, and it is the most widely available platform. The trade-off is that maintaining sub-millimeter accuracy depends on rigorous, ongoing QA (the discipline Winston and Lutz established for LINAC radiosurgery).

Practice preference (not a universal rule) For spine SBRT my preference is a gantry LINAC platform (e.g., Edge-class), for its arc delivery, integrated image guidance and intrafraction tracking, and ability to handle the cord-constraint geometry of spinal targets. Again a preference, not a verdict on other systems: robotic and other linac platforms deliver spine SBRT effectively, and the hub stays vendor-neutral.

Part III

Robotic Radiosurgery

3.Frameless, tracked, non-isocentric

The CyberKnife mounts a compact 6 MV X-band linac on a robotic manipulator, delivering many non-coplanar, non-isocentric beams under continuous image guidance. Its signature capability is real-time tracking: six-degree-of-freedom skull tracking for cranial targets, Xsight for spine without fiducials, and Synchrony correlating respiratory motion with fiducial or lung-based imaging for mobile body targets, entirely frameless. Beams are shaped by fixed cones, a variable Iris aperture, or the InCise multileaf collimator introduced with the M6 generation. This makes it well suited to spinal lesions, targets adjacent to critical structures, and mobile body sites, and it gained FDA clearance for intracranial treatment in 1999 and for extracranial treatment in 2001. On the current S7 generation with MLC collimation and the VOLO optimizer, intracranial sessions typically run about 10 to 25 minutes. The remaining trade-offs are platform-specific planning and a workflow that differs substantially from gantry-linac practice.

Part IV

Particle (Proton/Helium) Radiosurgery

4.The Bragg-peak advantage

Charged-particle radiosurgery exploits the Bragg peak: protons (and historically helium ions) deposit most of their energy at a defined depth and essentially no exit dose beyond it, giving a dosimetric advantage for sparing tissue distal to the target. Historically applied to AVMs, skull-base tumors, and pituitary lesions, and attractive in pediatric cases for integral-dose reduction, particle radiosurgery remains limited to a small number of centers by cost and complexity, and a clear clinical-outcome superiority over photon radiosurgery has not been established for most cranial indications.

Part V

Comparison and Emerging Systems

5.Side by side

Newer dedicated cranial systems include the self-shielded, cobalt-free ZAP-X, which aims a linac source gyroscopically with self-shielding for outpatient siting (FDA cleared 2017). The table summarizes the families on the axes that actually drive selection.

Radiosurgery platforms compared on selection-relevant axes (representative, not exhaustive; vendor-neutral).
Platform	Source / beam	Localization	Sites	Characteristic strengths
Gamma Knife	192 cobalt-60 sources, fixed convergent beams	Frame; Icon/Esprit add CBCT, mask, and motion management	Head	Many-isocenter conformality; steep gradient; small/multiple/complex cranial targets
Gantry LINAC	MV photons, high-definition MLC (2.5 mm central leaves) or add-on micro-MLC, VMAT (often FFF)	CBCT, stereoscopic kV and surface (e.g., ExacTrac Dynamic); mask/body frame	Cranial and body	Versatility; widely available; cranial SRS/SRT and spine/body SBRT
Robotic (CyberKnife)	6 MV X-band linac on robotic arm; non-isocentric	Frameless image guidance; real-time tracking	Cranial and body	Motion tracking; lesions near critical structures; spine/body
Particle (proton)	Protons/ions; Bragg peak	Image-guided; frame or frameless	Cranial and body	No exit dose; integral-dose sparing; pediatric, skull base, AVM
ZAP-X	Linac source, gyroscopic, self-shielded (no cobalt)	kV image guidance; mask	Cranial	Self-shielded outpatient siting; cobalt-free dedicated cranial

Platform choice, in one principle Match the machine to the target, not the target to the machine. Single small or complex cranial lesion in one session → a dedicated cobalt unit is excellent; a body or spine target → a linac platform; cranial fractionation → either a mask-capable cobalt unit or a linac, decided by target size, geometry, and local workflow; mobile body target or heavy intrafraction motion → robotic tracking; integral-dose-critical pediatric/skull-base case → consider protons. Within its envelope, each well-commissioned platform performs comparably, so experience and QA matter more than brand.

Key points

All platforms realize one concept (stereotactic, steep-gradient, ablative dose); selection is by target size/location, fractionation, motion, and resources, not brand.
Gamma Knife: 192 cobalt-60 sources, 8 sectors, 4/8/16 mm collimators; many-isocenter cranial conformality; Perfexion is frame-only, while Icon and Esprit add CBCT, mask fixation, and motion management; cranial only.
Gantry LINAC (TrueBeam/Novalis/Edge-class): micro-MLC VMAT, FFF, CBCT/ExacTrac/surface guidance; the versatile generalist that also does spine/body SBRT.
CyberKnife: robotic, frameless, non-isocentric, real-time tracking; strong for spine, critical-structure-adjacent, and mobile body targets.
Proton/particle: Bragg-peak (no exit dose), useful for pediatric/skull-base/AVM; limited by cost/access; no clear cranial outcome superiority shown.
Within its design envelope, each well-QA'd platform performs comparably; the hub presents practice preferences (Gamma Knife for complex/benign cranial; Edge-class for spine) as preferences, not verdicts.

References

Adler JR Jr, Chang SD, Murphy MJ, et al. The CyberKnife: a frameless robotic system for radiosurgery. Stereotact Funct Neurosurg. 1997;69(1–4 Pt 2):124–128. PubMed
Benedict SH, Yenice KM, Followill D, et al. Stereotactic body radiation therapy: the report of AAPM Task Group 101. Med Phys. 2010;37(8):4078–4101. PubMed
Winston KR, Lutz W. Linear accelerator as a neurosurgical tool for stereotactic radiosurgery. Neurosurgery. 1988;22(3):454–464. PubMed
Petti PL, Rivard MJ, Alvarez PE, et al. Recommendations on the practice of calibration, dosimetry, and quality assurance for gamma stereotactic radiosurgery: Report of AAPM Task Group 178. Med Phys. 2021;48(7):e733–e770. AAPM
Cirino E, Benedict SH, Dupre PJ, et al. AAPM-RSS Medical Physics Practice Guideline 9.b: SRS-SBRT. J Appl Clin Med Phys. 2025;26(4):e14624. PubMed
Hanvey S, Hackett P, Winch L, et al. A multi-centre stereotactic radiosurgery planning study of multiple brain metastases using isocentric linear accelerators with 5 and 2.5 mm width multi-leaf collimators, CyberKnife and Gamma Knife. BJR Open. 2024;6(1):tzae003. BJR Open
Grishchuk D, Dimitriadis A, Sahgal A, et al. ISRS technical guidelines for stereotactic radiosurgery: treatment of small brain metastases (≤1 cm in diameter). Pract Radiat Oncol. 2023;13(3):183–194. PubMed

Educational synthesis for neurosurgery and radiation-oncology trainees; platform descriptions are technical/representative and vendor-neutral, not endorsements. Vendor-neutral platform and QA references verified against PubMed/AAPM during review.