A round burgundy crest emblem with simple gold ring detailing, plain and formal in style
A wide horizontal photograph of mountain scenery near Banff with pale turquoise sky, dark evergreen slopes, and a soft muted mood
A round burgundy crest emblem mirroring the first, with plain gold ring detailing and a formal quiet mood

Annual Scientific Meeting of the Canadian Organization of Medical Physicists — July 9–12, 2014, The Banff Centre, Banff, Alberta, Canada

Accuracy Considerations for Stereotactic Localization in Cranial SRS

The use of stereotactic radiosurgery for intracranial targets depends on a chain of geometric and imaging decisions that determine where the beam actually converges. Sub-millimetre deviations can shift dose away from the planned isocentre, with measurable consequences for adjacent organs at risk. Localisation systems have evolved from rigid frame-based approaches to image-guided platforms that combine cone-beam CT, surface imaging, and optical tracking. Each component carries its own uncertainty budget, and the cumulative error defines the margin a clinician can apply in practice.

Medical physicists evaluating the accuracy of stereotactic localization systems for cranial SRS must balance vendor claims against commissioning data and long-term quality assurance. Independent audits, end-to-end tests, and Winston-Lutz-style analyses remain foundational, even as modern platforms automate much of the workflow. The discussion that follows draws on widely available evidence and on the experience presented at professional gatherings such as the annual scientific program reviewed in Banff.

In Australia, cranial SRS is performed at a relatively concentrated set of metropolitan centres, including Peter MacCallum Cancer Centre in Melbourne, the Crown Princess Mary Cancer Centre in Sydney, and Royal Brisbane and Women's Hospital. The geographic spread means that physicists in Hobart, Darwin, and Cairns often rely on remote collaboration with credentialed specialists, and audits traverse several jurisdictions with different state-level radiation safety frameworks. The Therapeutic Goods Administration regulates hardware approvals under the Therapeutic Goods Act 1989, and ACPSEM sets the professional competency expectations that underpin safe practice.

Imaging Chain Uncertainty and Its Quantification

The first major contributor to localisation accuracy is the imaging chain that produces the planning study. CT slice spacing, kVp selection, reconstruction kernel, and the presence or absence of intravenous contrast each influence the apparent position of a target. Magnetic resonance registration adds another layer of uncertainty, particularly when thin-slice volumetric sequences are not acquired isotropically. Distortion correction algorithms vary in handling gradient nonlinearity, so physicists should test these on their own scanners rather than relying on vendor defaults.

A practical approach involves placing fiducials at multiple off-axis positions and comparing reconstructed coordinates against physical truth. Repeat scans across several days reveal geometric accuracy and reproducibility, which often matters more clinically than single-shot precision. The traceability standards underpinning Australian dosimetry calibrations are not generally available for geometric phantoms, so physicists rely on consensus guidelines from ACPSEM and international equivalents.

Documenting imaging uncertainty and combining it with registration uncertainty yields a useful pre-treatment sanity check. Many Australian departments record these values in a structured table reviewed at each plan sign-off, a practice encouraged through remote multidisciplinary meetings linking Perth, Sydney, and Brisbane. The habit also surfaces trends that might otherwise go unnoticed between annual audits.

Hardware Performance of Couch and Optical Tracking Systems

Modern linear accelerators used for cranial SRS rely on a tightly integrated motion stack: a six-degree-of-freedom couch, kV and MV imaging arms, and often an optical surface monitor. Each subsystem introduces small but uncorrelated errors that combine into total system uncertainty. Periodic Winston-Lutz testing at multiple gantry, couch, and collimator angles captures the dominant contribution, but only if repeated for every cone or MLC setting used clinically.

Surface tracking has its own limitations. Surface reconstruction is sensitive to patient positioning aids, blankets, and even the lighting in the treatment room. At centres such as Peter MacCallum and Liverpool Hospital, physicists have observed that sweating, hot palms, and nasal cannula tubing can degrade the surface image. The optical system is best treated as a confirmation tool that complements, rather than replaces, X-ray-based verification.

Mechanical wear, sagging couch tops, and tolerance drift in encoder resolution mean that hardware performance is not static. Annual testing captures slow drift, but smaller drift between services often requires additional checks. Many sites schedule a brief Winston-Lutz test at the start of each single-fraction SRS session, a habit that has spread from larger centres to regional units in Newcastle and Wollongong.

End-to-End Testing as a Practical Baseline

End-to-end tests reproduce the full clinical pathway from imaging through planning, setup, and delivery, and they remain the most defensible demonstration of localisation performance. The Hidden Target test, the Modified Winston-Lutz test, and a phantom with multiple small targets offer complementary information, and most Australian centres now use a combination. The Royal Australian and New Zealand College of Radiologists faculty guidelines, where they intersect with medical physics practice, often echo the result rather than prescribe the method.

The realism of end-to-end testing depends on the maturity of the planning system. Some MR-only systems introduce additional uncertainty when synthetic CTs are generated, and a separate phantom test should be designed to capture this. Across the Tasman, Australasian centres have collaborated on shared phantom designs, encouraged by the limited number of cranial SRS programmes in the region. Smaller centres sometimes pool together to receive an independent audit, with the cost shared and the data anonymised.

End-to-end testing pays off most when automated. Several vendors offer scripted workflows that run during off-hours and email a summary to the physicist, a feature that has become popular in newer centres in Adelaide and Perth. Reporting should be standardised so that trends can be tracked over time, with control limits set on the basis of initial baseline performance rather than vendor recommendations. The resulting dataset also supports cross-centre benchmarking.

Independent Audits and Cross-Validation

External audit programmes offer a useful complement to internal quality assurance, because they identify blind spots that arise from a single team repeating the same workflow. The IAEA postal audit programme, the European equivalent, and locally developed Australian programmes each have a role. ACPSEM's certification programme requires evidence of external audit participation for the relevant scope of practice, and many centres in Australia maintain that participation as a matter of routine.

Cross-validation between institutions can be done cheaply by exchanging anonymised patient plans and recalculating them on independent planning systems. Cranial SRS is well suited to this exercise, because the targets are small and the dose distributions are sensitive to setup uncertainty. Several Australian centres participate in the work of the ACPSEM Special Interest Group in Radiation Oncology Physics, comparing benchmark calculations and reviewed plans.

Independent audit data should be reviewed alongside internal data, and any discrepancy of more than a defined threshold should trigger an investigation. The Therapeutic Goods Administration's post-market surveillance expectations, combined with state-level radiation safety requirements, mean that documentation should be retained for several cycles. Audit findings also feed into the centre's broader quality management programme and inform future commissioning decisions.

Practical Recommendations for Evaluating Localisation Accuracy

A structured approach to localisation accuracy pays off most when it is aligned with the clinical workflow.

  • Document imaging, mechanical, and registration uncertainty budgets separately and in combination.
  • Perform end-to-end tests for each treatment modality, including specialised deliveries such as HyperArc or multiple brain metastases.
  • Schedule a brief Winston-Lutz check at the start of each SRS session, with documented corrective actions for any out-of-tolerance result.
  • Participate in an external audit programme that includes on-site or postal dosimetry and geometric checks.
  • Maintain a rolling trend analysis of QA results, with control limits set on the basis of initial baseline performance rather than vendor recommendations.

For Australian centres, additional considerations include alignment with ACPSEM's ROEMP certification requirements, compliance with state-level radiation safety legislation, and engagement with national audit programmes where available. A shared culture of open data exchange, supported by professional networks and facilitated by remote collaboration tools, can lift the standard of practice across the country.

For physicists, clinical physicists, and biomedical engineers interested in the broader discussion of localisation accuracy and cranial SRS, the annual scientific meeting archive remains a useful resource, with abstracts and presentations covering commissioning methodology, audit results, and emerging platforms. Reviewing the program details can help practitioners tailor a working session to local needs, especially for those preparing for credentialing, refreshing a QA programme, or commissioning a new delivery system.