Hi!
I’m using SIMIND and Pytomography for my masters thesis and i ran into some questions.
I simulated 128 projections with 128x128 matrices and im using Pytomography’s OSEM to reconstruct the volume. In the projections I have a total of 31 million counts in all projections (i used tc99m) but when i reconstruct I end up with 536 thousand counts. The projections were already activity scaled by index 25 but were later time scaled to 15s per projection time.
Here is my code:
ct_path= os.path.join(path, ‘mesa_tc99m.hct’)
header_path= os.path.join(path, ‘mesa_tc99m_tot_w2.h00’)
spect= simind.get_projections(header_path)
activity=488 #MBq
rescaled_spect= torch.poisson(spect*15)
total_counts= np.sum(rescaled_spect.cpu().numpy())
object_meta, proj_meta= simind.get_metadata(header_path, distance=‘mm’)
lower_path= os.path.join(path,‘mesa_tc99m_tot_w1.h00’)
upper_path= os.path.join(path,‘mesa_tc99m_sca_tot_w1.h00’)
lower_width= simind.get_energy_window_width(lower_path)
upper_width= simind.get_energy_window_width(upper_path)
photopeak_width= simind.get_energy_window_width(header_path)
lower_projection= simind.get_projections(lower_path)
upper_projection= simind.get_projections(upper_path)
scatter_estimation= simind.compute_EW_scatter(lower_projection, upper_projection, lower_width,
upper_width, photopeak_width)
scatter_rescaled_tew= torch.poisson(scatter_estimation*15)
ac_map= simind.get_attenuation_map(ct_path)
att_transform= SPECTAttenuationTransform(ac_map)
psf_meta = simind.get_psfmeta_from_header(header_path)
psf_transform = SPECTPSFTransform(psf_meta)
system_matrix= SPECTSystemMatrix(
obj2obj_transforms = [att_transform, psf_transform],
proj2proj_transforms=,
object_meta= object_meta,
proj_meta=proj_meta
)
likelihood= PoissonLogLikelihood(
system_matrix=system_matrix,
projections=rescaled_spect,
additive_term=scatter_rescaled_tew
)
recon_algorithm= OSEM(likelihood)
reconstructed_spect= recon_algorithm(n_iters=3,n_subsets=8)
total_recon_counts= reconstructed_spect.sum().item()
i wonder what’s wrong here, since I incorporated attenuation, psf and scatter corrections. This happens with other isotopes, not only with technetium. I’m running simind v8.0.