Ricerca Progetti nazionali, regionali, internazionaliNAZIONALIPRIN 2022InTrEPID: In vivo 3D dosimetry

InTrEPID: In vivo 3D dosimetry in radiotherapy Treatments with EPID

Over the past two decades, radiotherapy (RT) has seen a steep increase in technological complexity of treatment preparation and treatment execution, calling for the development of more and better quality assurance (QA) tools and procedures. In-vivo dosimetry (IVD) has emerged as a very powerful tool for treatment verification in conjunction with electronic portal imaging devices (EPIDs). However, EPIDs present several drawbacks like non-water equivalence and cumbersome calibration procedures.

The aim of InTrEPID is to use modern machine learning (ML) techniques to develop a novel 3D IVD method to overcome current EPID limitations and provide real-time and effective treatment QA. Specific patient checks usually include pretreatment verification performing dose measurements before the treatment starts together with linac QA and IVD. IVD can catch delivery errors during patient treatment, assist in treatment adaptation, and record the actual dose delivered to the patient. Unfortunately, it is not routinely performed in the RT centres due to drawbacks of commercial systems. Currently, IVD is technologically feasible using EPID. Amorphous silicon EPIDs are also widely used due to their intrinsic characteristics: large imaging area, high spatial resolution, high dynamic range and real time acquisition capability. However, the use of these

devices as dosimeters requires a calibration procedure to establish a relation between the pixel intensity value and the dose distribution in the patient, or the energy fluence map which are difficult to assess. ML can help to overcome these limitations, as they are successfully employed in many pattern recognition problems in radiation therapy applications.

The InTrEPID project will develop a 3D IVD method which exploits the information encoded in EPID images acquired during the RT treatment, combined with the patient anatomy, to reconstruct the 3D dose map delivered to the patient. The proposed procedure is fully data driven, being based on advanced deep learning methods. This approach will allow bypassing the complex calibration steps needed to overcome the non-water equivalence of the a-Si EPID flat panel. Furthermore it will avoid explicit patient scattering estimations and complicated commissioning steps.

For the first time we will demonstrate the potential of ML to reconstruct the 3D dose distributions of intensity modulated treatments. This study will investigate dose discrepancies observed during in vivo dosimetry, establish tolerance levels for IVD for advanced treatments, and investigate the detectability of introduced errors. InTrEPID will be a fully automated and easy-to-use procedure for single subject 3D dose map predictions from EPID images and it will help to verify the consistency of the treatment delivery with the treatment prescribed by the Radiation Oncologist, in real-time, for every patient, during every fraction, fully automatically without any additional workload.

 

Data di avvio 28 Settembre 2023

Data di completamento 28 Settembre 2025

Total cost  €136528,00

 

Progetto 2022CWXR8K finanziato all’interno del Bando PRIN 2022 di cui al Decreto Direttoriale n. 104 del 02/02/2022 nell’ambito del Piano Nazionale di Ripresa e Resilienza, Missione 4 – Componente 2. Dalla Ricerca all’Impresa - Investimento 1.1 Fondo per il Programma Nazionale della Ricerca (PNR) e Progetti di Ricerca di Rilevante Interesse Nazionale (PRIN), finanziato dall’Unione europea – NextGenerationEU – CUP B53D23004090006