ESTRO 2024 - Abstract Book

S4036

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2024

Material/Methods:

A prospective cohort of 240 patients with paraoptic head and neck or CNS tumors, receiving proton therapy as part of their radiotherapy, underwent systematic assessment of visual acuity, visual field perimetry, visual evoked potentials, optical coherence tomography (OCT) +/- angio-OCT, and slit-lamp cataract assessment. Patients’ sagittal planes were determined based on points equidistant from both eye centroids. The patients’ Frankfurt planes were defined using right and left landmarks at lowest orbital floors and ear canal roofs [4]. Patients’ planes were defined automatically using machine learning based on a training set and homothetic morphing transformation of the skull shape/curvatures was applied to all patients of the set [5]. The intersection of these two planes defined the head orientation. The optical axes were defined on each CT as vectors passing through eye and lens centroids. Gaze directions were characterized as the angles between the optical axes and head orientation. The macula is at the intersection between the retina and optical axis, the papilla at the intersection of the retina and optic nerve. The retinal dose map was obtained from the projection of the surface covering a 30° angle around the optical axis at the posterior eye. This surface corresponds to the patient's sensitivity map of the visual field perimetry.

Interpatient variability in head and gaze positions were calculated in the 240-patient database; 21 patients (65 CTs) were then randomly selected to assess interfraction gaze variations and impact on retinal dose.

Results:

In the 240-patient cohort, mean patient head orientation (expressed in degrees) in the machine coordinate system were values of -0.02° (normal distribution, SD, [range] (2.64, [-13.60;9.13]) in the right-left direction, i.e. well-aligned for most. They were 2.44° (9.97, [-17.70;33.43]) in the foot-head direction. Head position was generally well aligned, but with a larger inter-individual dispersion in the foot-head direction. Patient gaze direction in the patient coordinate system were normally distributed with mean (SD, range) values of - 0.70° (11.40, [-38.49;34.21]) in the right-left direction, i.e. a straight ahead gaze for most but with moderate inter individual variability and significant outliers in the 30° order of magnitude range. Corresponding values were -13.02° (13.73, [-49.03;29.40]) in the foot-head direction. Gaze was predominantly directed to the feet with over 10° inter individual variability and important outliers in the 30° order of magnitude range. In the 21-patient sample, median angle was -1.02° (distribution not normal, mean:-1.48, [range:-28.02;20.91]) in the right-left direction. Corresponding values in foot-head direction were -13.88 (-15.23, [-49.03;18.44]). Patients had dramatically distinct patterns of gaze variations between fractions (on repeat CTs), especially in the foot-head direction. Median dose difference in mean dose to the retina between CTs compared with the planning CT was clinically unsignificant -0.07 Gy (0.83, [-16.37;19.38]), yet 4 patients had more than 5 Gy difference. Corresponding median differences in dose to the macula were -0.032, (-0.41, [-16.44 ; 18.74]) with 5 patients having more than 5 Gy difference. Corresponding values to the papilla were -1.18, (-0.09, [-0.43;7.61]) with 6 patients having more than 5 Gy difference. In the most critical case, dose deposit exceeded normal tissue tolerance dose [6].

Conclusion: