Multi-arc, non-coplanar circular-cone stereotactic radiosurgery on a linear accelerator
Everything in linac-based stereotactic radiosurgery (SRS) depends on knowing exactly where the target is. Thin-slice CT and MRI, fused into a single stereotactic coordinate system, let clinicians localize a small brain metastasis or other lesion to sub-millimeter precision before a single beam is planned — the foundation the entire steep-gradient technique is built on.
Linac-based radiosurgery emerged in the early 1980s when Italian physicians Osvaldo Betti and Federico Colombo independently adapted standard linear accelerators — machines built for fractionated cancer treatment — to deliver single, highly focused, high-dose treatments using multiple arcs and a rigid stereotactic head frame. This built on the conceptual radiosurgery framework Lars Leksell had described decades earlier for a dedicated multi-source unit.
Early linac radiosurgery relied on an invasive frame screwed to the skull under local anesthesia, fixing the head in the same coordinate space for imaging and treatment. Modern practice increasingly uses frameless immobilization: a custom thermoplastic mask combined with daily cone-beam CT or stereoscopic X-ray surface tracking, which achieves comparable sub-millimeter accuracy without surgical pins — enabling multi-fraction SRS courses that a rigid frame could never support.
SRS targets are frequently only a few millimeters to a few centimeters across, and the dose gradient outside them falls off extremely steeply — often from 80% to 20% of prescription dose within just a few millimeters. Any imaging or registration error directly eats into that margin, risking either a geographic miss of the tumor or unnecessary irradiation of adjacent eloquent brain.
To minimize this risk, planning uses 1mm (or near-1mm) slice CT for electron-density and geometric fidelity, fused with contrast-enhanced T1 MRI (and sometimes MRA or high-resolution CISS/FIESTA sequences) for superior soft-tissue and vascular contrast. Automated rigid or deformable image registration algorithms align these datasets into one stereotactic coordinate frame, and the fusion is visually verified slice-by-slice before contouring begins.
Once fused, all imaging is expressed in a single stereotactic (x, y, z) coordinate system referenced to the treatment isocenter. Because SRS targets are typically small and well-circumscribed, the gross tumor volume (GTV) and planning target volume (PTV) are often identical or carry only a 0–1mm margin — unlike conventional radiotherapy, which adds several millimeters for setup uncertainty and microscopic spread.
This tight target definition is what makes the subsequent steps — cone selection and non-coplanar arc planning — possible: the whole treatment is only as accurate as this first localization step.
To sculpt the beam into a small, sharp-edged circle, linac radiosurgery historically relies on secondary tertiary collimators — fixed circular cones of tungsten or brass mounted beneath the standard jaws. Choosing the right cone diameter for a given target is a direct trade-off between undertreating the tumor and overtreating the healthy brain immediately around it.
Circular cones were the original tool used by Betti, Colombo, and later Lutz and Winston-Lutz to convert a general-purpose linac into a radiosurgery device. Each cone is a solid tungsten or brass insert that attaches below the collimator jaws and defines a clean, round field with minimal beam-edge blur (penumbra) — because the aperture sits close to the target and the field is small, scatter and transmission outside the circle are minimal.
A typical cone set spans roughly 5mm to 40mm in diameter, in fixed steps, letting planners match the beam to a wide range of small, roughly spherical lesions.
Planners select the smallest cone diameter that still geometrically encompasses the target with only a small margin — often near zero for well-defined lesions, since SRS dose gradients are so steep that even 1–2mm of unnecessary margin measurably increases normal-brain dose. A cone that is too large irradiates a disproportionate volume of healthy tissue around a small target; a cone that is too small under-covers the lesion, risking a marginal recurrence.
Because a single circular cone produces only a round cross-section, cones are best suited to targets that are roughly spherical. Irregularly shaped or larger lesions are usually treated instead with dynamic conformal arcs shaped by a micro-multileaf collimator (mMLC), or with multiple overlapping isocenters ("shots") if cones must be used.
Fixed cones remain attractive for small, singular, round targets because their penumbra can be sharper than an mMLC field of the same size — mMLC leaves introduce small transmission and "tongue-and-groove" leakage between leaves that cones simply do not have. Cones are also mechanically simple and inexpensive.
Modern dedicated SRS platforms increasingly favor mMLC-based dynamic conformal arcs for irregular, multiple, or larger targets, since the leaf pattern can reshape on the fly as the gantry rotates and as different targets are treated in a single session — something a fixed cone cannot do without a physical cone change.
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Fixed circular cones | Small, single, round lesions (<2–2.5cm) | Solid tungsten/brass aperture, 5–40mm, changed manually per case | Sharpest penumbra, simple, very reproducible |
| Micro-MLC (mMLC) | Irregular, larger, or multiple lesions | 2.5–5mm leaves reshape field continuously during arc rotation | Dynamic conformal shaping, no cone changes needed |
| Multiple cone "shots" | Irregular lesions treated with cones only | Several overlapping spherical dose shots fill an irregular volume | Uses simplest hardware, at cost of planning complexity |
A single arc delivered with the couch fixed at one angle sweeps the beam through only one plane, concentrating entrance and exit dose in a band of normal brain that every part of that arc shares. Rotating the couch between arcs — non-coplanar delivery — scatters those entrance/exit paths across many different planes so no point outside the target sees more than a fraction of the total dose.
If every arc is delivered with the couch at the same angle, all beams share the same rotational plane — they enter and exit the head along the same great circle. Every voxel of normal brain in that plane receives contribution from every arc, producing an elongated "hot streak" of unwanted dose running through otherwise healthy tissue, even though the target itself may look well covered on axial images.
By re-positioning the couch to a different angle before each arc, the beam paths spread out over many different orientations in 3D — much like sampling many great circles on a sphere rather than repeating one. Outside the small region where all paths cross (the target), any single point in normal brain is now traversed by only one or two arcs instead of all of them, dramatically lowering integral dose to healthy tissue while keeping the target fully covered.
Non-coplanar, multi-arc delivery only works if the gantry, collimator, and couch all rotate about the exact same point in space — the mechanical isocenter — regardless of angle. The Winston-Lutz test, described by Lutz, Winston, and Maleki in 1988, verifies this directly: a small radio-opaque ball bearing is placed at the presumed isocenter and imaged or irradiated from a series of gantry, couch, and collimator angles. Any offset between the BB's shadow and the field center reveals mechanical wobble.
Because cone-based SRS delivers beam with no live imaging feedback during treatment, this pre-treatment mechanical QA — typically required to demonstrate agreement within 1mm, and often under 0.5mm on modern dedicated units — is what makes it safe to trust that beams from many different couch/gantry combinations truly converge on the same point.
A Winston-Lutz deviation of even 1mm on a small cone can shift the effective beam center by a clinically meaningful fraction of the target diameter — which is why daily or per-fraction mechanical QA is standard practice for SRS programs.
Planners typically use 4–7 arcs, balancing plan quality against beam-on and setup time. Each arc is defined by a fixed couch angle plus a gantry start and stop angle; angles are chosen to maximize angular separation between arc planes and, where possible, to avoid entering through the eyes, optic apparatus, or brainstem.
Gamma Knife achieves an inherently non-coplanar, near-spherical geometry automatically, using 192–201 fixed cobalt-60 sources arranged in a hemispherical helmet. Linac-based SRS must recreate that spherical sampling manually, one couch kick and arc at a time — which is exactly why deliberate non-coplanar arc selection is such a central planning step for linac SRS specifically.
During treatment, the gantry rotates continuously through each planned arc range while the beam fires without interruption, rather than stopping at discrete angles. Dose from each arc adds vectorially to a growing three-dimensional distribution, and dedicated SRS platforms combine high dose rates with sub-millimeter positioning to deliver this efficiently and accurately.
Unlike step-and-shoot delivery, dynamic arc therapy fires the beam continuously as the gantry sweeps through its prescribed range — for example, a 180° or larger partial arc — so dose is deposited along a continuous swath rather than at a handful of static angles. Once one arc finishes, the couch rotates to the next planned angle and the gantry sweeps again.
Each arc contributes its own thin sliver of dose passing through the target; after all planned arcs are delivered, these slivers sum into a single composite 3D dose distribution that is high and uniform at the target and falls away steeply outside it, exactly because each arc's low-dose entrance/exit path lies in a different direction from every other arc's.
Purpose-built or SRS-optimized linac platforms — such as BrainLab/Varian Novalis, Varian TrueBeam STx, and Varian Edge — combine sub-millimeter mechanical accuracy with features that make multi-arc, non-coplanar delivery practical and safe: high-dose-rate flattening-filter-free (FFF) beams that shorten beam-on time, 6-degree-of-freedom robotic couches that correct both translational and rotational setup errors, and integrated stereoscopic kV imaging (e.g., ExacTrac) for real-time or near-real-time position verification.
High dose rate matters clinically: shorter beam-on time reduces the total treatment session length, which in turn reduces the chance of patient motion degrading the sub-millimeter accuracy the whole technique depends on.
Because SRS margins are so tight, position is checked before and often during delivery. Rigid stereotactic frames mechanically guarantee position for single-fraction treatments; frameless workflows instead pair a custom thermoplastic mask with optical surface-tracking systems (such as AlignRT or Catalyst) or periodic cone-beam CT / stereoscopic X-ray imaging, gating or interrupting the beam automatically if the patient moves beyond a sub-millimeter threshold.
This continuous verification is essential across a multi-arc session that may last many minutes, ensuring the target remains within tolerance for every one of the non-coplanar arcs, not just at the start of treatment.
The final measure of an SRS plan is how tightly the high-dose region conforms to the target and how quickly dose falls off just beyond it. A well-executed non-coplanar multi-arc plan produces a compact, near-spherical dose cloud with a steep gradient — very different from the elongated dose "streak" a single coplanar arc alone would leave in normal brain.
Classic single-fraction SRS delivers a high ablative dose — commonly 18–24 Gy — in one session, and is best suited to small targets (roughly under 2–3cm) that sit at a safe distance from critical structures such as the optic apparatus or brainstem. For larger lesions, or those adjacent to such structures, hypofractionated stereotactic radiotherapy (SRT) spreads a somewhat lower total dose over 3–5 fractions (for example, 25 Gy in 5 fractions or 27 Gy in 3 fractions), exploiting the radiobiological benefit of fractionation to reduce the risk of radionecrosis while still achieving high local control.
Linac platforms handle both regimens naturally, since fractionation is simply a matter of scheduling repeat sessions with the same frameless immobilization and daily image guidance.
Plan quality is quantified with two standard metrics. The Conformity Index (CI) is the ratio of the volume receiving the prescription dose to the target volume itself — an ideal plan has CI near 1.0, meaning the high-dose region exactly matches the target with no excess. The Gradient Index compares the volume of the half-prescription isodose to the volume of the full-prescription isodose — a low value means dose falls off quickly just outside the target.
Non-coplanar arcs improve both metrics simultaneously: because entrance/exit paths from different arcs rarely overlap outside the target, the composite dose cloud stays compact and drops off steeply in every direction. A single coplanar arc, by contrast, can still produce an acceptable-looking axial conformity while hiding a long, low-conformity streak of dose running through normal brain in its rotation plane — a limitation only visible when the distribution is examined in 3D.
Gamma Knife's 192–201 fixed cobalt-60 sources are arranged in a hemispherical helmet, giving it an inherently, near-perfectly spherical non-coplanar geometry and one of the steepest achievable dose gradients — ideal for very small, single-fraction targets treated with a rigid frame. Linac-based SRS instead builds non-coplanarity through sequential couch and gantry combinations, which requires the careful arc planning and mechanical QA described above, but in exchange offers substantially more flexibility: irregular or larger targets can be shaped with dynamic conformal mMLC arcs, multiple targets can be treated in a single setup, hypofractionated regimens are straightforward, and dedicated SRS-capable linacs are far more widely available than dedicated Gamma Knife units.
In practice, the two technologies are complementary rather than competing: Gamma Knife excels for small, discrete, single-fraction lesions, while linac SRS/SRT extends precision radiosurgery to larger, irregular, multiple, or fractionation-requiring cases.
With a well-designed non-coplanar arc arrangement, the 50% isodose line can fall within roughly 3–5mm of the target edge — compared with a much longer, asymmetric extension of unwanted dose along the rotation plane when only a single coplanar arc is used.