1. Thursday Case of the Day
Physics
Brad Kemp, Ph.D.
Department of Radiology, Mayo Clinic, Rochester, MN
History: A 60 year old male underwent a PET/CT study to investigate squamous cell carcinoma of the tongue base
Scanning protocol:
15 mCi FDG, 65 min uptake
3D acquisition with 4 min/bed on a GE Discovery RX PET/CT system
Fully-3D iterative reconstruction (IR) with CT attenuation correction (CTAC), scatter and randoms from singles correction
The arrows indicate regions of apparent reduced uptake. What is the cause of these artifacts? a) attenuation correction b) scatter correction c) reconstruction filter d) randoms correction
Figure 1: Coronal IR CTAC image

2. Figure 2: Fused PET/CT image
Findings: The correct answer is b)
This artifact is caused by the scatter correction; there is an overcorrection of scatter caused by a misalignment of the patient’s arms in the PET and CT images.
Motion of the patients arms are shown in the fused PET/CT image in Figure 2 as the hands are misaligned. The model-based scatter correction uses the CT images to define the outline of the patient and, after randoms correction, assumes that only scattered events are detected outside this outline. The estimated scatter distribution is then scaled by the coincidence events detected in this external body region (usually referred to as ‘tail fitting’). Since the patient moved their arms away from their body for the PET acquisition there are true coincidence events that are outside the body outline and incorrectly assumed to be scatter. As a result the scatter content is overestimated which results in an overcorrection of scatter and regions of reduced uptake in the reconstructed images.
Figure 2: Fused PET/CT image

3. Discussion: This artifact is caused by the scatter correction; there is an overcorrection of scatter caused by a misalignment of the patient’s arms in the PET and CT images.
Figure 3: Original IR CTAC image with scatter correction
Figure 4: IR CTAC image without scatter correction
Figure 5: IR NAC image
Figure 4 shows a coronal image with attenuation correction but no scatter correction applied. Note that the artifact is no longer present. However quantification of uptake (ie standardized uptake value or SUV) is not possible in images with no scatter correction. Figure 5 shows a coronal image without attenuation and scatter correction (NAC). These images do not have the artifacts associated with the overcorrection of scatter. There is an artifact adjacent to the bladder caused by inhomogeneous attenuation due to the hip bones and inconsistencies in the sinogram data.

4. Discussion: Images from a patient in which there was no arm misalignment and no overcorrection of scatter.
Figure 6: IR CTAC image with scatter correction
Figure 7: IR CTAC image without scatter correction
Figure 8: IR NAC image
Coronal images of another patient with the same body habitus, acquired with the same acquisition protocol and administered FDG dose. In this study the patient’s arms are aligned in the PET and CT images and there is no overcorrection of scatter. In the two bed positions over the pelvis the average scatter fraction estimated by the scatter correction was In comparison, in the patient study with the artifacts the average scatter fraction was 0.75.

5. Bibliography:
Lodge MA, Mhlanga JC, Wahl RL. Effect of patient arm motion in whole-body PET/CT. J Nucl Med 2011;52:464P
Lodge MA, Muhammad AC, Udall DN, Wahl RL. Characterization of a perirectal artifact in 18F-FDG PET/CT. J Nucl Med 2010;51:
Wollenweber SD. Parameterization of a model-based 3D PET scatter correction. IEEE Trans Nucl Sci 2002;49: