Ponentes
Descripción
Tumors located near the optic pathways pose a particular challenge due to the risk of optical toxicities. For this reason, they are often treated with proton therapy. Such patients with paraoptic tumors represented more than half the whole patient population, suggesting referral biases to proton therapy toward difficult to treat tumors.
The PIOTox study aims to conduct a voxel-scale analysis of optical toxicities induced by proton therapy [1]. For this study, patients with paraoptic head-neck/skull-base/CNS tumors undergoing proton therapy were consecutively included. A prospective database of 240 patients from the Cancer Center Baclesse comprises information on patient radiotherapy, such as proton therapy planning, millimetric CT scans, and comprehensive optic structure delineation.
This database is complemented by patients' results from ophthalmologic examinations (field of view, visual evoked potential, and thickness measure of RNFL) conducted at the University Hospital of Caen before radiotherapy and at subsequent time points (1 month, 1 year, 2 years, etc.). This multicentric database is organized and anonymized using the ArDCore software [2]. The objective is to predict the future visual state of the patient based on the initial patient condition and the planned radiotherapy. For this purpose, preliminary work on paraclinical data is conducted (age effect correction, calibration, noise analysis, quality assurance). Subsequently, a model of the optic nerve is built on CT scans to enable the use of the dose deposited in each voxel as a feature for the future use of machine learning algorithms. The development of the geometric model of the optic nerve was carried out using the ESPADON package [3] deployed in the R language.
As eye rotations and gaze may affect the dose to functional optic subunits (ESTRO 2024), the visual outcome predictions, has also highlighted that eye rotations and gaze may affect the dose to functional optic subunits.
[1] Thariat J, SEQ-RTH22 INCa 16863
[2] Combes, S. & Bacry, E. & Fontbonne, Cathy. (2020). Health Data Hub; plateforme des données de santé en France, application à l’oncologie radiothérapie. Cancer/Radiothérapie. doi: 10.1016/j.canrad.2020.07.003.
[3] Espadon, an R package for automation, exploitation and processing of DICOM files in medical physics and clinical research », C. Fontbonne, J.-M. Fontbonne, N. Azemar, Phys. Med. 109 (2023). doi: 10.1016/j.ejmp.2023.102580
