The effect of attenuation on lesion detection in PET oncology

Using simulation and patient studies, we compare the lesion detection in whole-body PET oncology with and without performing attenuation correction (AC and non-AC). In addition, we discuss the prediction of lesion distortion in non-AC PET images. In all of the simulation studies, the OSEM algorithm is used to reconstruct 2D noise-free data. Lesions of different size, density, and uptake are simulated in the lungs. Lesions with the density of soft tissue and different uptake are simulated in the mediastinum and abdomen. The patient studies are performed using an ADAC dual-head hybrid SPECT-PET system. The lesion-to-background ratio (LBR) in the reconstructed images is used for quantitative analyses. Simulation studies and a total of 40 patient studies show that the LBR in AC images is higher in lungs but lower in the mediastinum and abdomen than in non-AC images. In addition, simulation studies show that, in non-AC images, for given lesion and lung uptakes, the LBR increases when the lesion density decreases and spherical lesions appear to be elongated in the direction with minimal attenuation and squeezed in the direction with maximal attenuation.