Global Navigation
Development and Processing of High-Resolution Cardiac Magnetic Resonance Imaging
- Principal Investigator: Jay Zweier, M.D.
Development of a fast high-resolution EPR imaging method using spinning magnetic field gradients
One of the fundamental limitations to the implementation of higher-dimensional spectroscopic imaging methods to biological systems is the prohibitively long data acquisition time. In order to increase the image acquisition speed and to enhance the image resolution, we developed a low-frequency (300 MHz, for imaging whole-body rodents) high-speed EPR imaging (EPRI) system using spinning magnetic field gradient (SMFG). The fast EPR imaging technique relies on the sensitivity of the imaging systems that is directly proportional to the EPR working frequency. In low frequency EPRI that is uniquely suited for the imaging of large living samples, the image quality is subject to the magnetic field inhomogeneity. In this case, the introduction of a spinning magnetic field gradient may induce additional image distortion such as geometric deformation that becomes more noticeable over the large FOV. We have thoroughly investigated the optimal spinning frequency (rate of gradient cycling) and number of steps of field sweep, for a given set of imaging parameters, so as to shorten the imaging time without sacrificing image quality. We have optimized both the hardware and software implementation of the SMFG technique to perform fast EPR imaging. We further validated the technique with phantoms (see Fig 1) and biological objects and compared to the standard stepped gradient approach. The spinning magnetic field gradient method should enable fast acquisition of multi-dimensional EPR images for mapping free radicals and oxygen in the heart. Further work is in progress to adapt this method to cardiac gated imaging.
Fig 1: High-speed EPR imaging of a free radical phantom. A phantom consisting of 7 tubes (id = 3 mm) filled with DPPH powder was used. Images A and B are the pictures of the phantom. Image C was acquired using discrete rotation of the gradient (conventional method) while image D was acquired using the SMFG fast imaging technique. Parameters used in C were: number of projections = 32; scan time = 2.6 s/projection; time constant = 10 ms. Total imaging time was 84 s. Parameters for fast imaging experiment (D) were: spinning frequency = 24 Hz; number of field sweep steps = 64; time constant = 0.64 ms. In both regular and fast imaging experiments, scan width = 4 mT; FOV = 70x70 mm2 and modulation amplitude = 0.05 mT. A total of 203 projections were used and the imaging time was 2.6s.
Development of software algorithm for 3D/4D spectroscopic imaging methods for mapping of oxygen concentration in tissues, in vivo
Imaging of oxygen concentration in tissues can be performed using EPR spectroscopic or spectral-spatial (SSI) imaging. The method requires the use of an oxygen-sensitive paramagnetic probe. A complete 3D reconstruction of oxygen image requires 4D imaging (3-spatial dimensions, and one spectral-dimension) of the oxygen probe. The challenges facing the 4D SSI for imaging oxygen concentration are: (i) huge data size; (ii) long acquisition time; and (iii) reliable and efficient algorithms to extract oxygen information from the line-shape data. We are currently developing strategies for high-speed acquisition of 4D images using spinning/sweeping magnetic field gradient. We are also implementing a direct reconstruction algorithm to avoid the extra time required in the projection reconstruction (PR) mode data acquisition and back-projection. Our goal is to address the data acquisition, reconstruction and processing of large data arrays (128/256 pixels/dimension) to enable high-resolution mapping of oxygen concentration.
PR spectral-spatial (or spectroscopic) images contain EPR spectrum for each voxel in an image data set. Because the spatial and spectral dimensions are fully separable, information about local line-width, and hence local oxygen content, can in principle be derived independently from local spin density. We have developed procedures for accurate evaluation of line-width and hence oxygen concentration in the voxels. We have used thefollowing approaches for the auomatic conversion of EPR spectral lineshapes to linewidth/oxygen concentration:
Computation of linewidth by lineshape fitting
In this approach (see Fig 2) the spectral function in each voxel is accurately simulated using either Lorentzian or a mixture of Gaussian and Lorentzian functions. The line-width data are then converted to pO2 values using standard curve or calibration parameters. The pO2 data are displayed as 2- or 3-dimensional image.
Fig 2. EPR images of embedded particulates of LiPc and oxygen mapping in a RIF-1 tumor. The images were obtained on day 11 after inoculation of the animal with LiPc and RIF-1 cells. A: Spatial image showing the distribution of the particulates in the tumor. B: Oxygen image, obtained from spectroscopic imaging, of the tumor in room air-breathing animal. C: Oxygen image of the tumor in carbogen-breathing animal. The images were reconstructed from 256 projections acquired in 24 min. The overall increase in the intensity in the case of carbogen-breathing animal demonstrates an increase in the tumor oxygenation. Note that the oxygen information is obtained only from the regions where the particulates are present.
Computation of line-width directly from spectral parameters, area under the curve (AUC) and amplitude
From this data set, spatial maps corresponding to local spin density and maximum EPR spectral line amplitude are generated (see Fig 3). A map of local EPR spectral line-width is then computed. Because line-width directly correlates with oxygen concentration, the line-width image provides a map of oxygenation. This method avoids a difficulty inherent in other oxygen content mapping techniques using EPR, that is, the unwanted influence of local spin probe density on the image. Further work is in progress to process 4D spectroscopic images.
Fig 3: Software interface showing the image of oxygen concentration in a murine tumor measured under in vivo conditions.