Magnetic resonace imaging is an important component of modern radiotherapy. The current project aims at development of methods for efficient and robust integration of MR into the radiotherapy workflow.
MR has since long been recognized as important for accurate radiotherapy because of the superior soft tissue contrast and possibilities to image physiological tissue properties such as diffusion, perfusion and vessel permeability, but up to recently only as compliment to CT. Developments the last years; most importantly - improved geometric accuracy and wider scanner boars, has changed the view of the future role of MR in radiotherapy. Today, there is a strong interest for MR only workflows, both to avoid geometrical errors when MR and CT images are registered and to save resource.
The overall goal of the project is to develop a method to extract computed tomography (CT) equivalent information from magnetic resonance (MR) images. This method can replace the CT in radiotherapy and PET applications with MR imaging. MR has several advantages compared to CT for both these applications. -MR does not expose the patient to ionizing radiation, which enables continuous imaging of the patient during a PET/MR acquisition for motion correction purposes or in 4D imaging as part of preparation for radiotherapy. -MR provides a superior soft tissue contrast which adds great value both in radiotherapy, when the physician defines the volume to be treated; and in PET application, when the tracer uptake is localized in relation to the patient anatomy. -The options for functional imaging with MR are numerous which is of high interest both scientifically in a PET/MR scanner where the physiological imaging can be related to the uptake for specific tracers; and in radiotherapy where the functional imaging can be of importance when defining the extension of the tumor volume. To be able to take full advantage of the MR in radiotherapy and in a PET/MR application it is necessary eliminate the dependence of the CT scanner, i.e. find an alternative way to estimate the electron density- or Hounsfield Unit (HU)-map. The project is centered around a novel method for indirect estimation of a Hounsfield unit map, i.e. a substitute-CT (s-CT) image set, based on MR images of a patient. We have recently published a proof of concept publication in Medical Physics, which is based on ultra short echo time images and a Gaussian mixture regression (GMR) model. Specific aims of the proposed project are: 1.Further development of the Gaussian mixture regression method. An important step is to investigate how to use relative instead of absolute voxel values as input to the GMR model. This would reduce the sensitivity to the coil placement and is a necessary step for a method that is easily portable between individual MR scanners. 2.Optimization of the UTE sequence for s-CT purposes. The two focus areas in the sequence optimization is contrast and to reduce motion sensitivity. A secondary goal is to extend the field of view for the UTE sequence. 3. Clinical testing of the method and creation of robust and efficient methods for quality control.