You are here:

Work package 03

Image management

Develop reliable non-linear registration techniques to align all CT, MRI, and PET scans acquired during the course of a treatment. Use these registrations to develop the concept of adaptive dose delivery plans in radiotherapy to account for the changes in morphology and tumour biology encountered during repeated treatment/imaging sessions.
Description of work
Task 3.1. Non-linear registration of CT, CBCT, MRI, and PET images and autosegmentation. Regularize non-linear, image intensity based registration algorithms with biomechanical models derived from contoured organs in the reference scan to ensure that only physically plausible deformations can occur. Autosegmentation will be validated in clinical images (repetitive imaging during EBRT and 4D image acquisitions).
Task 3.2. Statistical modelling of daily anatomical variation. Using statistical modelling (principal component analysis (PCA)) of e.g. the initial week of treatment, predict/model the expected variation in subsequent sessions. Couple the PCA to the model of biomechanical deformation (task 3.1) for correct point matching.
Task 3.3. Developing strategies for adaptive image guided RT taking organ deformation and biology into account. 1) Plan-of-the-day strategy: using the PCA (task 3.2), precompute dose plans for a number of likely organ deformations and choose the best match in each subsequent treatment session. 2) Adaptive strategy: develop a tool for calculation of cumulated delivered doses by using non-rigid deformation combined with continuous adaptation of the dose distribution according to organ deformations and biology (dose painting).
Deliverables (projected outcome)
D3.1. Robust non-linear, multi-modality image registration regularized by biomechanics.
D3.2. Establish procedures to adapt dose plans to organ deformations and tumour biology.
D3.3. Plan-of-the-day scenario based on PCA and non-linear registration.