Protocol for Evaluating a Planning Algorithm for Gamma Knife Radiosurgery

The process of developing Gamma Knife radiosurgery treatment plans is today very dependent on the level of human expertise resulting in a great heterogeneity of the intrinsic qualities of the treatment planning and the quality of care delivered in radiosurgery. The existing reverse planning systems, although they have progressed considerably in recent years, produce results that are still lower than those achieved by an expert. Conventionally, an experienced dosimetric planner will act mainly on the coordinates of the position of the isocenters, on the size of the collimators, sector by sector, and on the irradiation time of each isocenters. In theory, the combination of these variables provides access to billions of combinatorics whose diversity far exceeds the computational capabilities of the human mind. The dosimetric planner therefore uses a very small part of the spectrum of possible patterns by always reproducing a limited number of empirical solutions. The company Intuitive Therapeutics has developed a new mathematical algorithm which can automatically test in a very short time all the combinatorial possibilities and converge very quickly to the solution that best meets the clinical, anatomical and dosimetric objectives defined by the neurosurgeon. The quick processing of the system also allows the operator to modify the constraints to refine the proposed result in real time. The demonstration of the reality of the performances of this algorithm would give the ability to even inexperienced users to develop high performance planning for the benefit of the patient in terms of optimizing the efficacy / toxicity ratio of the radiosurgery treatment results The primary objective is to evaluate comparatively the quality of the schedules produced by the algorithm developed by the company Innovative Therapeutics to those produced by an expert who carried out more than 15000 dosimetric plannings and radiosurgical interventions. The main criterion of comparison is the Paddick index. The secondary criteria for comparison are:

• Compliance index
• selectivity index
• Gradient index
• Maximum, minimum, average dose at risk structures
• Dose distribution in the target volume
• Treatment time (at equal source activity)
• Time of realization of the planning It was chosen to treat patients with vestibular Schwannoma OR multiple brain metastases (\> 5) treated in single session by Gamma Knife OR para-optic meningioma in hypo-fractionated treatment on Gamma Knife with restraint mask with inclusion of visual paths in the target volume planning. The aim of the study is to show at least the non-inferiority of this new method compared to the expert user based on the Paddick Index. This index has continuous values between 0 and 1, 0 being the worst case and 1 being the ideal solution. In order to define the sample size needed for each pathology, a pilot phase is required. This phase can be performed retrospectively using treatments already defined by the expert user. This pilot phase will allow us to identify the difference that can be expected between the index values and the variability of this difference. Based on these values we will be able to determine the size of the sample allowing us to statistically test the non-inferiority, or even the superiority of this new device. The number of cases to be included during the pilot phase should be at least ten cases and a maximum of thirty cases. The choice of the number of cases to be integrated during this pilot phase will depend on the homogeneity of the differences obtained on the first cases. These values will allow us to calculate the size of the samples necessary for the study of non-inferiority as well as for the study of superiority. Depending on the calculated sizes samples we will make a choice to ensure that this study takes place in the expected duration.