Purpose: PET/computed tomography (CT) has been shown to detect lesions in patients with pulmonary tuberculosis (PTB) and may be useful for assessing PTB disease in clinical research studies. However, radiation dose is of concern for clinical research in individuals with an underlying curable disease. This study aimed to determine whether PET/MR is equivalent to PET/CT in PTB.
Materials and methods: Ten patients with microbiologically confirmed PTB were recruited. Patients received 129.0±4.1 MBq of fluorine-18-fluorodeoxyglucose. Five of the 10 patients underwent a PET/MR scan, followed by PET/CT. The remaining five were first imaged on the PET/CT, followed by the PET/MRI. PET acquisition began at 66.7±14.4 min (mean±SD) after injection when performing PET/MR first (PET/CT: 117.2±5.6 min) and 92.4±7.6 min when patients were imaged on PET/MR second (PET/CT: 61.1±3.9 min). PET data were reconstructed iteratively with Ordinary-Poisson Ordered-Subset Expectation-Maximization and reconstruction parameters were matched across the two scanners. A visual lesion detection task and a standardized uptake value (SUV) analysis were carried out. The CT Hounsfield unit values of PTB lesions were also compared with MR-based attenuation correction mu-map tissue classes.
Results: A total of 108 PTB lesions were detected on PET/MR and 112 on PET/CT. SUV analysis was carried out on 50 of these lesions that were observed with both modalities. Mean standardized uptake value (SUVmean) and maximum standardized uptake value (SUVmax) were significantly lower on PET/MR (SUVmean: 2.6±1.4; SUVmax: 4.3±2.5) than PET/CT (SUVmean: 3.5±1.5; SUVmax: 5.3±2.4).
Conclusion: PET/MR visual performance was shown to be comparable to PET/CT in terms of the number of PTB lesions detected. SUVs were significantly lower on PET/MR. Dixon-based attenuation correction underestimates the linear attenuation coefficient of PTB lesions, resulting in lower SUVs compared with PET/CT. However, the use of PET/MR to measure the response of lung lesions to assess response to treatment in research studies is unlikely to be affected by these differences in quantification.