Over the last three decades medical imaging has developed to be an important tool for medical condition diagnosis, treatment planning and surgery. The insight provided by information contained in medical images has become essential in providing the community with 'a better life' and 'a healthy lifestyle'. Medical imaging aims to promote life and prolong life expectancy within the wider community by increasing our knowledge of healthy and diseased processes. Medical imaging tools to better delineate and identify information are necessary to impart clarity to the medical profession. High field Magnetic Resonance Imaging (MRI) has had much development in the last five years and Susceptibility Weighted Imaging (SWI) has emerged as an image post-processing technique. Mostly the magnitude of the complex data is illustrated as intensities in images, but in SWI, the additional information contained in the phases is used to enhance what is captured in magnitude images. The extracted additional level of detail can be used in traumatic brain injury, stroke and other brain disorders, to better delineate the spatial severity of the problem. Assignment 1 is related to how SWI can be used to enhance magnetic resonance images. In MRI the presence of a magnetic field induces the relaxation of protons. This relaxation in time is captured as a signal through the generation of an echo. The signal is time series data
and is referred to as the Free Induction Decay (FID). In this assignment it is assumed that proton (1H - hydrogen) relaxation data has been acquired and is stored as signal magnitude and phase. The provided data are in image space, and generated after data derived from the FID was 2D Fourier transformed. The actual raw data collected by the scanner can be computed by taking the inverse 2D Fourier transform of the complex image intensities. As part of this assignment you are provided with '2dseq' files containing magnitude and phase information and 'acqp' file consisting of data acquisition parameters. These files were generated on a Bruker 16.4T research scanner using Paravision 4.0. The echo signals were generated using a T2 * weighted gradient echo (i.e. transverse relaxation) image acquisition sequence. The assignment requires that the data is read, processed and susceptibility weighted images are reconstructed. You are required to submit your m-files and answers on blackboard. Each question must be a function in a separate m-file executed and saved as A1QX.m (X=1, 2, 3 or 4). You should save your answers and plots, combine these into a PDF file and submit it along with your m- files. Clearly mark which question you are answering and label all plots. You are not required to write lengthy answers, simply provide what is asked. You will not only receive marks for correct answers, but also for clarity of your presentation. This means that your algorithms should use indenting, explanatory comments for variables and complicated lines of code. You should use your research and searching skills to investigate any terms that you do not understand.
Using the supplied read_2dseq.m file read the provided files into MATLAB (useful Matlab commands: abs, image, axis, caxis, imagesc, colormap, figure, subplot, title). The results should be illustrated as figure 1, using four subplots and gray scale. The following questions should be illustrated as separate plots in figure 1. For slice number 100, illustrate the following:
(a) Normalised magnitude image
(b) Scaled phase imaged in [-Π, Π]
(c) Real part of the image
(d) Imaginary part of the image
You need to filter slice 100 phase images to identify microscopic phase changes contained amongst macroscopic phase variations (Note: file QualityGuidedUnwrap2D.m has been supplied to you).
(a) Low-pass filters the images using an 8*8 boxcar filter. (Note: Boxcar method smooths the variations through averaging, effectively reducing the amount of high frequency variations. Furthermore, it helps if the phases are first unwrapped).
(b) Illustrate the microscopic phase variations. (Note: Difference between original phase and new phase is high-pass filtered phase).