Research
Our research focuses on the following two interconnected areas:
- motion correction that eliminates artifacts in brain images and reduces errors in image-derived quantitative measures;
- Small vessel MRI methods that can monitor early blood vessel lesions underlying cerebral small vessel disease.
1. Motion correction
The early intervention and accurate diagnosis of brain diseases rely on high-quality MRI images. However, motion artifacts seriously affect the image quality and the accuracy of quantification results. Elimination of motion artifacts entails accurate measurement of head positions and modelling of artifact mechanisms which include:
- motion-induced k-space coordinate errors;
- B0 field changes;
- coil sensitivity changes.
All the above mechanisms can affect measured MRI signal. Our research aims to develop hybrid motion measurement techniques to achieve fast and reliable head motion measurement. Techniques under development include fat navigators that can provide stable motion parameter measurement and motion sensitive electromagentic induction coils that has higher temporal resolution.
Based on the measured motion parameters, we are devloping methods that can correct the above-mentioned motion effects in the clinical MRI sequences. To correct for the last two effects, we are developing physical and neural network models that can predict motion-induced B0 field and coil sensitivity changes from motion parameters.
2. Small vessel MRI
Cerebral small vessel disease is common among individuals above 60 years of age. However, current clinical MRI sequences can only detect tissue damages induced by the diseased small vessel, but not the vessel lesions themselves that occur at an earlier stage of the disease, which hampered the development of effective therapeautic strategies. The inability of detecting the small vessel lesons is largely due to lack of sufficient spatial resolution, poor signal to noise ratio for detecting such small structures, and paucity of quantitative studies due to the challenges in small vessel segmentation and the presence of severe partial volume effects. To overcome these limitations, a multitude of strategies are being pursued, including
- motion correction to ensure high quality images and consistent results during longitudinal studies;
- targeted imaging of small vessels with high in-plane resolution and thick slices perpendicular to the vessels to simultaneously achieve high SNR and high spatial resolution;
- Further improving SNR by scanning under ultra high magnetic field strengths such as 5 T and 7 T;
- Further improving SNR by utilizing custom designed high permittivity materials and wireless coil elements;
- developing deep learning based automatic segmentation models;
- building physical models that can derive physiologically meaningful results despite the presence of severe partial volume effects;
The overarching goal is to develop MRI methods that can accurately follow the pathogenesis of small vessel lesions prior to the development of tissue damage, which will pave the way for the development of more effective therapeautic strategies of cerebral small vessel disease.
3. Other research activities
We also encourage students to pursue other research directions that suit their interests and career development goals. Research projects unrelated to the above two topics include oxygenation and perfusion MRI of human kidneys, deep learning based quantification of tissue perfusion based on dynamic imaging data, and MRI study of the physiological effects of acupoint stimulation practiced in tranditional Chinese medicine.