Background
The measurement of cerebral blood flow (CBF) by means of single photon emission computed tomography (SPECT) is a complimentary method to diagnose a brain death. It is difficult to prove a zero-flow state of the brain in the tomographic image because of an image reconstruction noise. We aimed to improve a reliability of the SPECT CBF measurement in a brain death state by utilizing the ordered subset expectation maximization (OSEM) algorithm for image reconstruction.
Materials and Methods
Brain phantom consisting of the skull part and intracranial part (cylindrical or Hoffman phantom) was employed. First, the skull part was filled with 12.7% solution of copper sulfate dissolving 2.58×10−1 MBq/ml 99 mTc-O4. The cylindrical intracranial part was filled with physiological saline (Phantom 1 mimicking intracranial zero-flow state). Second, the cylindrical intracranial part was replaced by Hoffman phantom filled with 1.12×10−2 MBq/ml 99 mTc-O4 for gray matter compartment and 5.59×10−3 MBq/ml 99 mTc-O4- for white matter compartment (Phantom 2 mimicking 5% CBF or 2.5% CBF compared with that of skull). Third, the 16 cm-diameter cylindrical phantom was filled with saline solution dissolving 4.26×10−2 MBq/ml sodium pertechnetate (99 mTc-O4-). (Phantom 3 to test the system sensitivity). The Gamma View SPECT 2000 H system (Hitachi Medical Co, Tokyo) was used for the scanning. The acquisition parameters were as follows; 20% symmetrical window (centered on 99mTc 140 keV photopeak), high resolution collimator, 128 8 seconds-acquisition, 64×64 image matrix, Butterworth filter (order 10, cutoff frequency 0.50 cycle/cm) and the Chang's method with attenuation coefficient of 0.08 cm−1. The scan data were reconstructed by means of the filtered back projection (FBP) algorithm or the ordered subset expectation maximization (OSEM) algorithm. The 10 cm-diameter oval region of interest (ROI) was placed on the intracranial part of the phantom image for count reading.
Result
In the Phantom 1, mean count and standard deviation was 0.367 ± 1.04 and 0.00867 ± 0.165 for the FBP and the OSEM (iteration 4; subset 8), respectively, indicating the OSEM value was much close to zero. In the Phantom 2, we were able to visually detect phantom brain structures such as cortical and central gray matter only when the OSEM was employed. In the Phantom 3, mean count/pixel was 239±33 and 144±15 for the FBP and OSEM algorithm, respectively.
Discussion
The present study demonstrated that zero-flow state or low-flow state of the brain could be more reliably visualized by means of the OSEM algorithm. Although the system sensitivity of the FBP appeared higher than that of the OSEM, this may be due to the inclusion of more noise in the FBP method. Since the OSEM algorithm allowed incorporation of effects of attenuation and scatter, modeling of statistical noise, and non-negative count assumption, it refined an accuracy of the tomographic activity distribution. For the evaluation of brain death state using a brain perfusion SPECT, the OSEM algorithm should be utilized for image reconstruction.