Comparison of 3D-RP and 3D-OPOSEM reconstructions of the ECAT HRRT PET data

Positron emission tomography (PET) is a powerful tool for quantifying the physiological processes through dynamic PET studies. In PET reconstruction, statistical image reconstruction methods such as ordered subset expectation maximization (OSEM) are widely used. However, it has been shown that iterative reconstruction methods produce bias for low count statistics data which is typical in dynamic PET studies. This bias problem has been published previously also for the ECAT high resolution research tomograph (HRRT, CTI PET Systems, Knoxville, TN, USA) PET scanner. Alternatively, dynamic PET data can be reconstructed analytically (for example by using 3D Reprojection, 3D-RP method). In this study, we compare 3D-RP and 3D-OPOSEM reconstructed dynamic HRRT PET images. For the 3D-RP reconstruction, the missing parts (caused by the octagonal detector design in the HRRT gantry) were estimated by the 3D DCT-domain gap-filling with a dedicated DCT-domain filter and the 3D transradial bicubic interpolation methods. We compared the reconstruction methods with two scans of one subject performed with the [¹¹C]-raclopride tracer. We assessed the methods visually by calculating the ratio images between the binding potential images obtained by different reconstruction methods. Moreover, regional time activity curves (TACs) and areas under the TACs of the employed reconstruction schemes were compared. We observed that for the cerebellum, caudate and putamen regions, 3D-RP reconstruction produced higher activity concentration than the 3D-OPOSEM reconstructions. On the other hand, the results obtained by the 3D-RP after different gap-filling methods were similar.