Abstract
241749
Introduction: The role of choroid plexus (ChP) in cerebrospinal fluid secretion, circulation, and brain functioning is an active area of research . Studies have reported the presence of several different cell linings, receptors , and in some studies, mineral deposits within these structures. Imaging the ChP could serve as an essential tool for diagnostic as well novel therapeutic development and treatment purposes. However, the size and anatomical location of these structures deep within the mid-brain pose a challenge due to the spatial resolution achieved by the current clinical SPECT imaging systems. Herein, we demonstrate our preliminary results of imaging of the ChP structures embedded in the XCAT brain phantom using the simulation software developed to model the imaging performance of the AdaptiSPECT-C (ASC) , a brain-dedicated multi-pinhole (MPH) SPECT imaging system being built at the University of Arizona for both clinical use and novel drug discovery.
Methods: An anthropomorphic model of the ChP structures, adjoining the lateral ventricles were added to the digital XCAT brain phantom. The relative concentration of selected radiopharmaceuticals of interest can now be assigned to voxels representing these ChP structures in left and right hemispheres for creation of source distributions. The ASC system has 24 modular MPH gamma cameras arranged in 3 rings in a hemi-spherical manner to image a 21 cm spherical volume of interest (VOI), large enough to image the 99th percentile male human brain. The ability to control the aperture size of individual pinhole or completely close it during image acquisition to meet the sensitivity and resolution requirements of the imaging task enables the truly adaptive imaging capability to ASC.
Herein we compare the noise-free image-acquisition performance of the ChP without and with background (i.e., complete brain with exception to the four ventricles) activity assignment. ChP: background activity ratio was set to 10:1. Voxelized activity and attenuation distribution were generated with 0.25 mm voxel size. Noise-free projection image acquisition was simulated and reconstructed using the expectation maximization maximum likelihood (MLEM) algorithm . To avoid multiplexing, 4 oblique pinholes were used to irradiate each detector (i.e., a total of 96 pinholes) with an aperture diameter that was determined to provide 8mm spatial resolution for imaging at the center of the VOI. Images were reconstructed using an isotropic voxel size of 1 mm. Image quality was assessed both visually and using cuboidal region of interest specific normalized root mean square error (NRMSE) and structural similarity index metric (SSIM).
Results: Simulated projection images of the ChP without and with background activity assignment are compared in Fig.1. The sum slice ground truth as well as reconstructed images along the transverse, sagittal and coronal planes are compared in Fig. 2. Anatomical details of these structures are well preserved and visualized even when the background activity was used during simulated imaging. The NRMSE and SSIM in the reconstructed images after 100 iterations were 0.40, 0.99 and 0.43, 0.98 for the images without background activity present and with background activity respectively.
Conclusions: Through computer simulations, we present simulated imaging of the choroid plexus structures added to the digital XCAT brain phantom at a 1 mm isotropic reconstructed voxel size. For the noise-free case, with the selected aperture diameters, the reconstructed images both visually and by image quality assessment metrics show excellent agreement with the ground truth. Future work will include multigrid based MLEM reconstruction algorithm to provide sub-mm resolution for assessment by higher-spatial resolution imaging than the ASC and variation of the assigned activity to the organs within the field of view, i.e., skin, thyroid, and skull bone. Research Support: NIBIB/NIH Grant UG3 EB034686 and R01EB022521.