School of
Information Technology and Electrical Engineering

Speaker: Chunyi Liu
Seminar Date: Fri, 06/12/2019 - 13:30
Venue: 45-413; Mansergh Shaw Building, Conference Room
Host: Prof Feng Liu

Seminar Type:  PhD Thesis Review

Abstract: 

Magnetic resonance based electrical properties tomography (MR-EPT) is a recently proposed method to reconstruct electrical properties (EPs) distribution during magnetic resonance (MR) scanning. The main idea of MR-EPT is determining the unknown EPs through the measured B1 data based on the Helmholtz equation. The obtained EPs maps can be used for local specific absorption rate (SAR) assumption, which helps to control the RF energy deposit into the patient according to the safety regulations. In addition, the obtained EPs maps can also provide important information for diagnostic purposes and therapeutic monitoring.

 

Current MR-EPT methods are associated with two main problems. (1) Unrealistic assumptions were applied to simplify the complex central equation. A widely used assumption is known as the ‘local homogeneity assumption’ which assumes EPs are locally constant or vary slowly in the region of interest (ROI). This assumption would result in significant artefacts at the transition regions with varied EPs. (2) The high order derivatives of the measured B1 field are needed to be computed in existing methods, which causes the augment of measurement noise. Therefore, it is necessary to obtain B1 field with high SNR and this is difficult to achieve by current B1 mapping technology.

In my thesis, two novel algorithms were proposed to solve the main problems in current MR-EPT methods and the application on the patient specific SAR estimation for pregnant woman scanning were investigated.

  1. A modified finite difference scheme were proposed in the first algorithm. It involves the gradient of EPs in the central equation to solve the problem caused by the ‘local homogeneity assumption’. In addition, due to the increase on the nodes in each discretised central equation, the algorithm can be more robust against noise and eliminate the high frequency oscillation to some extent. This algorithm was validated by both simulation and experimental results. Obvious improvements can be observed when compared with the results of existing methods.
  2. Inspired by the new expression proposed in the first algorithm, the author found that it can be further improved by applying the divergence theorem with introducing a new vector field and a more general expression of the central equation was derived. For the applications where the structural information can be obtained as prior information, the EPs can be calculated in a fast and reliable manner to compare with literature value for diagnosis. For the applications where the structural information is not available, a generalized D-EPT algorithm was proposed to seek a global solution of EPs. In this algorithm, both the ‘local homogeneity assumption’ and the calculation of high order derivatives were avoided and this is a very potential algorithm to put forward the application of MR-EPT in clinical applications.
  3. Based on the proposed D-EPT algorithm, a patient specific SAR estimation method was proposed and applied for the pregnant women scanning. Due to the movement of foetus in the amniotic fluid, it is not suitable to estimate the local SAR distribution with standard human body model when scanning the pregnant patient. Our proposed algorithm obtains a real-time SAR distribution from the measured data, which is more accurate and consistent with the scanned patient. This method can help to keep foetus safe during MR scanning and extend the application of MRI for pregnant scanning. 

Biography: 

Chunyi Liu is a current PhD candidate in the MRI group at School of Information Technology and Electrical Engineering, which leaded by Prof. Sturat Crozier. His principle supervisor is Prof. Feng Liu. Before that, he received his B.E and M.S. degree in physical electronics from Southeast University, Nanjing, China. His research work focuses on the electrical properties tomography in high/ultra-high field MRI system. His investigations can help to extend the applications of the high/ultra-high field MRI system and put forward its clinical applications.