[Application of the IR Method Using a Normal Database Created by FBP Method with No Correction in ^99mTc Myocardial...

Japanese normal databases (NDB) for quantitative evaluation of ^99mTc myocardial perfusion SPECT have traditionally been created by the filtered back projection (FBP) method but are also applied to the iterative reconstruction (IR) method in some institutions. Although creating the NDB specifically for the IR method is desirable, it has not been done partly due to the ethical aspects of creating an institution-specific database. This study investigated the impact of using NDBs created by the FBP method on quantitative evaluations performed with the IR method.

We analyzed single-photon emission computed tomography (SPECT) images from patients undergoing myocardial perfusion study in two hospitals with three SPECT vendors. Images were processed with and without various corrections, and a polar map was created by QPS software using a 17-segment model to obtain %uptake and scores. The effects of the reconstruction and correction methods were evaluated to determine the difference in the number of local segmental analyses by coronary artery territories.

In most segments, quantitative values showed no significant difference across device types, with or without corrections. However, both choices of the SPECT device and correction methods influenced the scores in specific coronary artery regions. Application of attenuation correction resulted in lower defect scores in the right coronary artery region.

Applying FBP-derived NDB to the IR method generally yielded a consistent patient diagnosis in terms of defect score analysis. However, since the quantitative evaluation was partly affected by various correction methods and devices, recognizing these characteristics is crucial for enhancing diagnostic accuracy.