The open design of the Evo MRI scanner with horizontal main field direction and vertical pole plates permits unique positional changes of patients which are, among many other applications, of great interest in upright RT.  In this context, the development of fast imaging and reconstruction techniques are of high relevance. The PhD student will be directly involved in the development process with the help of senior  engineers and scientists. The academic support will be provided by the department of mathematics, laboratory of “Methods for Image and Data Analysis” (MIDA) at the University of Genoa with substantial courses on the mathematical foundations of MR image analysis, reconstruction, and filtering. 

An important part of upright RT and cancer treatment is the characterization of the various organs and tissues in different positions. This is made evident in an example of cross-sectional liver scans during breathing (Figure 1). It applies, for example, to the dynamic variations during breathing of patients in upright positions which are not yet well studied by conventional imaging methods.  The differences between supine and standing positions are much smaller but nevertheless utmost important for accurate tumor delineation. This requires faster 3-dimensional MR data acquisitions with corresponding reconstruction and interpolation techniques.  

Figure 1: Sagittal cross section of the liver-lung interface at two different breathing states. 

Project tasks 

• Get to know the existing possibilities in MR imaging by using the MROpen EVO scanner for generating time-resolved MR imaging data, creating scan protocols and writing sequences.  

• Use, implement and optimize conventional and sub-sampling imaging techniques. 

• Explore time-resolved image reconstruction for better morphologic data and pseudo real-time displays, in the face of under-sampled data and low signal to noise acquisition.  

• Devise different imaging protocols to generate ground-truth data for machine learning for clinical MR applications. 

• Provide  a useful machine learning imaging database available for related projects. 

• Characterization of the hardware and software requirements corresponding to a successful use of MR imaging for upright RT. 

• Collaborate with all other students on their respective objectives, specifically at PSI, TheraPanacea and HZR. Planned secondments are at: 

• • TheraPanacea (Paris, France, 3 weeks) for AI training to provide background for the imaging reference data (ground-truth dataset) 

• • PSI (Villingen, Switzerland, 3 weeks) for insights into clinical supine MRI and in-room setup imaging 

• • HZDR (Dresden, Germany, 3 month) for testing devised sequences at the open-bore in-isocenter MRI