Embedded Files
The Zhao MRI lab is focused on innovating novel approaches to improve magnetic resonance imaging of human and animal brains. The Zhao Lab has developed machine learning methods for investigation of functional and structural connectivity in human and animal brains, and for other biophysical and biomedical applications such as cell tracking and cancer detection/treatment. The research topics includes:
The Zhao MRI lab is focused on innovating novel approaches to improve magnetic resonance imaging of human and animal brains. The Zhao Lab has developed machine learning methods for investigation of functional and structural connectivity in human and animal brains, and for other biophysical and biomedical applications such as cell tracking and cancer detection/treatment. The research topics includes:
- Structurally informed Functional Connectivity via Bayesian Framework
- Cellular MR imaging: Tracking stem cells labeled with iron oxide nanoparticles
- Machine learning and functional MRI
- Therapeutic treatment of cancer using SPIO nanoparticles
- MR phase gradient mapping
- Dynamic contrast enhanced MRI
- Magnetic resonance spectroscopy
Structurally informed Functional Connectivity via Bayesian Framework
Structurally informed Functional Connectivity via Bayesian Framework
Human or animal’s brain network can be described using either structural- or functional-connectivity (SC and FC), but most methods often treat FC and SC as separate domains, even though they are inherently linked. Structural pathways shape the functional communication between brain regions and dynamic patterns of brain activity may also influence the structure of neural connections over time. The lack of integration between these two entities limits our ability to fully understand the brain’s complexities. We predict that an integrated FC and SC approach could potentially better detect alterations in FC caused by neurological conditions such as traumatic brain injury (TBI). Developing the relationship between both entities can involve models ranging from simple to highly complex, contingent upon the specific objectives to be achieved. A Bayesian framework with Gaussian graphical models to generate a structurally informed FC (siFC) is found to be the best choice. The figure below shows the differential networks using typical FC and siFC methods.
Cellular MR Imaging: Tracking stem cells labeled with iron oxide nanoparticles
Cellular MR Imaging: Tracking stem cells labeled with iron oxide nanoparticles
MRI-based cell tracking using super-paramagnetic iron oxide (SPIO) particles provide an excellent means of cell monitoring in vivo. The magnetic nanoparticles function as T2 or T2* contrast agent and they change the transverse relaxation time of protons in surrounding water. SPIOs are ideal for in vivo cell tracking because they are non-radioactive, non-toxic, do not require viral transfection, and provide a detectable intracellular signal. Figure 1: Implanted neural stem cells (pointed by red arrowheads) labeled with SPIOs into CNS of a chicken embryo, tracked by using a 7 Tesla MR scanner. Figure 2: Implanted magnetic nanoparticles in chick embryos, monitored by using a 7 Tesla MR scanner.
Figure 1
Figure 2
Machine Learning: Functional and structural connectivity in human and animal brains
Machine Learning: Functional and structural connectivity in human and animal brains
Zhao Lab has been focusing on functional connectivity of brains by using Blood-Oxygenation Level Dependent (BOLD)-based functional MRI. In addition to a traditional sparse Dictionary Learning (sDL), convolutional neural networks (CNN) and restricted Boltzmann machine (RBM) are two deep learning methods that Zhao lab has investigated in detecting functional connectivity of brains in temporal and spatial domains. Figure: Six resting state networks detected in pig brains that are homologous with those in human brains. Representative images of the averaged RSN maps (orange on left) for six RSNs and the corresponding reference pig RSN atlas (yellow on right).
Cancer Therapeutics using SPIO Nanoparticles
Cancer Therapeutics using SPIO Nanoparticles
A. Tumor Detection
A. Tumor Detection
SPIO nanoparticles, as T2 contrast agents, can be used for early detection of tumors. However, conventional gradient-echo based pulse sequences generate negative contrast (darker spot, see figure (a) below) due to signal loss caused by shortened relaxation time by the nanoparticles. By collaborating with the investigators (Michael Garwood et al) of the University of Minnesota, the Zhao group used a new approach, sweep imaging with Fourier transform (SWIFT) sequence integrated with saturation pulses to enhance the tumor by generating positive contrast (brighter spot, see figure (b)) and suppressing surrounding tissues (see figure (c)). Figure: Tumors grown in mouse flank, and axial (cross-sectional) images acquired using (a) gradient echo sequence, (b) SWIFT sequence, (c) SWIFT with saturation pulses.
B. Magnetic Therapeutic Treatment using Hyperthermia
B. Magnetic Therapeutic Treatment using Hyperthermia
SPIO nanoparticle induced hyperthermia using an alternating magnetic field can be applied for treatment of various cancers. A hyperthermia system for remote heating of iron oxide nanoparticles was developed using alternating magnetic fields in Zhao’s lab to treat human head and neck cancer using a mouse xenograft model. Figure (Left) : Illustration of treatment of tumors using an alternating magnetic field. Figure3 Middle : A cross section of mouse flank (tumor enclosed with the red line). Figure3 Right : Histology result after tumor was treated with hyperthermia.
C. Drug Delivery
C. Drug Delivery
Controlled drug release is targeted by the application of SPIO nanoparticles as drug carriers. A “therapeutic cocktail” treatment is being investigated in Zhao’s lab that uses magneto-thermal effects to treat tumor cells. Small IONPs coated with polymer brushes and targeting ligands release drugs as a result of magneto-thermal effect. Molecular thermometer measures the localized coating temperature using the temperature dependent fluorescence intensity of different dyes. This is currently a collaborative work with Jason Locklin and Jin Xie, Professors of Chemistry at the UGA.
Magnetic Resonance Phase Gradient Mapping
Magnetic Resonance Phase Gradient Mapping
Monitoring of temperature in real time during thermal treatment of tumor (e.g. high intensity focused ultrasound (HIFU) or hyperthermia) is crucial for evaluating treatment effect. MR phase gradient mapping (PGM), a novel technique developed in the Zhao’s lab, is capable of achieving this goal. Figure (Left): (a) A view of the magnitude image of the high intensity focused ultrasound (HIFU) data set. The ROI used to estimate the baseline phase map is highlighted. (b) The unwrapped baseline phase map. (c) The unwrapped post-heating phase map. The distribution of temperature estimations from the standard PRF-shift MR thermometry, Rieke’s reference less method and the proposed method implemented using the standard basis is shown in the 2nd row. Figure (Right): Estimations of the internal temperature in the HIFU data set.
Dynamic Contrast Enhancement MRI
Dynamic Contrast Enhancement MRI
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a noninvasive imaging technique that has been widely studied as a cancer imaging tool. The Zhao’s group has developed several methods for DCE-MRI data analysis, such as pharmacokinetic parameter ratios based on a reference region model. Figure: Seven canine brain tumors (pointed by double arrows) enhanced by DCE-MRI.
MR Spectroscopy
MR Spectroscopy
Multinuclear MR spectroscopy is another area that the Zhao group has focused on. Different from MR imaging, analysis of a MR spectrum provides information on the number and type of chemical entities in a molecule. Working with UGA kinesiology scientists, the Zhao group has developed a noninvasive technique for measuring muscle oxidative capacity using 31P (Phosphorus) magnetic resonance spectroscopy.
Page updated
Report abuse