Assessment of heart-to-contralateral lung ratios in virtual planar images derived from SPECT using...

A new AutoPlanar imaging technique can generate virtual planar images from SPECT data. We aimed to determine whether AutoPlanar images can provide accurate heart-to-contralateral lung ratio (H/CL) from ^99mTc-pyrophosphate (PYP) scintigrams and subsequently to identify optimal reconstruction conditions for these images.

This retrospective study assessed 41 patients with suspected transthyretin cardiac amyloidosis using ^99mTc-PYP scintigraphy followed by planar and SPECT imaging three hours later. We reconstructed SPECT images using ordered subset expectation maximization (OSEM) and ordered subset conjugate-gradient minimizer (OSCGM) combined with resolution recovery (RR), attenuation (AC), and scatter (SC) correction. Cardiac tracer uptake on planar and AutoPlanar images was visually assessed using a 4-point scale (0 to 3) and H/CL ratios were compared between the planar and AutoPlanar methods.

Visual assessment demonstrated good to excellent agreement, with exact agreement ranging from 73.2% to 97.6%. The H/CL ratios derived from AutoPlanar and planar images were highly correlated (r ≥ 0.97). Bland-Altman analysis demonstrated small negative biases (- 0.087 to - 0.027); the 95% limits of agreement were - 0.130 to 0.051 for Flash3D+ with ACRR and - 0.123 to 0.061 for Flash3D+ with ACSCRR. Equivalence testing demonstrated statistical equivalence between planar and AutoPlanar H/CL ratios with ACRR and ACSCRR, whereas equivalence was not demonstrated with RR.

The H/CL ratios derived from AutoPlanar and planar images were comparable. AutoPlanar images were more strongly influenced by correction conditions than by the reconstruction method, with Flash3D+ using ACRR or ACSCRR showing favorable overall results.