|
EPR Imaging Toolbox collection user manual |
iradon_qiao_fbp3d 3D filtered backprojection single stage image reconstruction from an arbitrary set of 1D projections Syntax: image = iradon_qiao_fbp3d(P, radon_pars, recon_pars) Description: P - projections (float, MxN, M - points in projection, N - number of projections) radon_pars - Radon transformation parameters (structure) ~.G - gradient vector (float, G/cm, Nx3, N - number of projections) ~.w - projection weight factor (float, Nx1, N - number of projections) recon_pars - reconstruction parameters (structure) ~.nBins - image size in voxels (int) ~.Filter - projection filtration 1: 2rd derivative 2: 3-points derivative method 3: 5-points derivative method 4: RL filter 5: SL filter, 6: hamming filter 7: two-ramp filters method ~.FilterCutOff - Filter cut off, part of full bandwidth (float, 0 to 1) ~.Interpolation - interpolation method, (int, 0/1/2) 0: zero-rank interpolation method; 1: linear interpolation method; 2: spline interpolation method. Note that for GPU reconstruction, you can select only 0 or 1. ~.zeropadding_times - zeropadding factor for two-ramp filter (int, >= 1) ~.CodeFlag - reconstruction code (string, 'CPU'/'GPU') ~.display - verbosity settings (int, 0/1, 0: no display, 1: intermediate results); ~.tasksliced - break task in multiple executions to save memory, GPU only (int, 0/1) image - reconstructed object (float, NxNxN, N - number of voxels in each dimension) Example: % Generate equal solid angle layout of projections fbp_struct.nAz = 36; fbp_struct.nPolar = 36; fbp_struct.imtype = iradon_GetFBPImageType('XYZ'); fbp_struct.MaxGradient = 1; % maximum gradient (G/cm) fbp_struct.angle_sampling = 'UNIFORM_SPATIAL_FLIP'; pars = iradon_FBPGradTable(fbp_struct); % Radon transformation parameters radon_pars.un = pars.G; % unit vectors of the gradients radon_pars.size = 5; % projection spatial support (cm) radon_pars.nBins = 64; % length of the projection spatial support % Phantom parameters phantom.r = 2.2; % radius of the sphere (cm) phantom.offset = [0.1,-0.1,0]; % offset of the sphere (cm) % use radon_c2d_sphere to generate analytic projections P = radon_c2d_sphere(phantom, radon_pars); radon_pars.G = pars.G; % determine projection weights radon_pars.w = iradon_vor_area_3d(radon_pars.G); recon_pars.size = 5; % reconstructed image size (cm) recon_pars.nBins = 128; % reconstructed image dimensions 128x128x128 recon_pars.Filter = 6; % hamming filter recon_pars.FilterCutOff = 1.0; recon_pars.Interpolation = 1; % 1 is linear interpolation method recon_pars.InterpFactor = 2; recon_pars.CodeFlag = 'GPU'; recon_pars.display=0; recon_pars.tasksliced=1; % [0/1] use 1 to break task in multiple executions % call the reconstruction program and display result image = iradon_qiao_fbp3d(P, radon_pars, recon_pars); ibGUI(image); Legend: EPR-IT functions; MATLAB functions; comments.
[1] Z.W. Qiao, G. Redler, B. Epel, H.J. Halpern, Comparison of parabolic filtration methods for 3D filtered back projection in pulsed EPR imaging, J. Magn. Reson., 248 (2014) 42-53 DOI 10.1016/j.jmr.2014.08.010. See also: radon_c2d_sphere, iradon_InterpToUniformAngle, iradon_vor_area_3d |