TY - JOUR T1 - <strong>Base-to-apex noise gradient with dedicated cardiac and standard SPECT camera designs</strong> JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1708 LP - 1708 VL - 59 IS - supplement 1 AU - Sarah Cuddy-Walsh AU - R. Glenn Wells Y1 - 2018/05/01 UR - http://jnm.snmjournals.org/content/59/supplement_1/1708.abstract N2 - 1708Objectives: Image noise impacts the detectability of disease. Increased attenuation with depth in a patient leads to decreased sensitivity and thus increased noise in the detected signal. Also, dedicated cardiac multi-pinhole single photon emission computed tomography (SPECT) cameras have spatially variant sensitivity across the field-of-view. The camera sensitivity decreases with increasing distance from the collimators and this may augment the gradient in the noise magnitude over the length of the heart. Our aim was to measure the noise gradient across the length of the heart and to compare its size for dedicated cardiac and standard parallel-hole SPECT camera designs. Methods: Supine rest studies with (342 ± 20) MBq of 99mTc-tetrofosmin from 27 patients (17 male) ranging from 45-102 kg with normal images (summed stress score &lt;4) were retrospectively re-evaluated. All patients were injected once and imaged with both a standard parallel-hole SPECT/CT (13 min, 180° acquisition in 3° steps) and dedicated cardiac SPECT (5 min). For each image, 392 statistically different sets of projections were produced using non-parametric bootstrapping. The standard-camera images were reconstructed with 2 iterations of ordered-subset expectation maximization (OSEM) using 10 subsets with (AC) and without (NC) attenuation correction, similar to our standard clinical processing. The dedicated cardiac SPECT images were reconstructed using 40 iterations AC maximum a posteriori expectation maximization (MAP-EM) with a lighter one-step-late Green’s (OSLG) prior (α=0.51, β=0.3), 60 iterations AC MAP-EM OSLG and a heavier prior (α=0.7, β=0.5), or 40 iterations NC MAP-EM OSLG (α=0.51, β=0.3). No post-filtering was applied. Mean and standard deviation (noise) images were calculated over the 392 reconstructed images and converted to 17-segment cardiac polarmaps. Noise was expressed as a percent of the maximum segment in the mean image. The gradient in the heart was evaluated as a ratio between the average noise in the 6 basal segments and noise in the apical segment. Results: In NC images, the average noise in cardiac images was (9±5)% and (16±3)% for the dedicated cardiac and standard cameras respectively. The base-to-apex noise ratios were small in the NC case (1.09±0.14 and 1.01±0.08) but significantly different (p=0.02) between the two cameras. Improving the estimate of the true underlying distribution by AC kept the average noise at approximately the same level but increased the base-to-apex ratio. For dedicated cardiac images reconstructed using 40 iterations with lighter regularization, the gradient from base to apex was (1.4±0.1) and for images reconstructed using 60 iterations with heavier regularization the gradient was (1.6±0.2). The standard-camera AC images also had a gradient from base to apex (1.3±0.1), though significantly smaller than those observed from the dedicated cardiac camera (p=0.018 and p=2×10-9 compared with light and heavy regularization). Conclusion: Base-to-apex gradients in image noise magnitude exist in the field-of-view of both dedicated cardiac and standard parallel-hole camera designs. These gradients are larger for multi-pinhole dedicated cardiac camera designs. Research Support: This work is supported by the M. Hildred Blewett Fellowship of the American Physical Society, www.aps.org, the Queen Elizabeth II Graduate Scholarship in Science and Technology, the Kiwanis Club of Ottawa Medical Foundation and Dr. Marwah, and an NSERC grant RGPIN 2016&amp;#8209;05658. ER -