A comparison of segmentation and emission subtraction for singles transmission in PET

Single photon transmission scans, using ¹³⁷Cs (662 keV gamma rays), are performed post-injection for clinical whole body PET studies. Two methods of processing the transmission data are investigated. The first method segments the transmission image into lung and soft tissue volumes. Average attenuation coefficients for 511 keV are applied and this transmission image is forward projected for attenuation correction. In the second method short emission contamination (EC) scans are performed after the singles transmission scans. These are singles transmission scans without a ¹³⁷Cs transmission source. These EC scans are then subtracted from the transmission scans to remove contamination of the 662 keV energy window by 511 keV emission gamma rays. From ten patients studied by this method, the EC subtracted transmission images have average soft tissue attenuation coefficients of 0.074/cm, compared to the expected value of 0.086/cm for 662 keV gamma rays. Simple scaling of these transmission images to 511 keV values (0.095/cm) results in quantitative attenuation correction. A comparison of the two methods is made for these ten patient studies. Both methods result in accurate attenuation correction. The segmentation method performs low noise attenuation correction but is more limited to torso studies with the limbs out of the FOV. The EC subtraction method is more versatile and is applicable to the entire body but results in some more noise in the fully corrected images