A brief comparison of PET analysis methods
Single-bolus studies
| Method | Outcome | Requirements for | Applicable to parametric imaging | Vulnerabilities | |
|---|---|---|---|---|---|
| PET scan | Input | ||||
| Fit to
compartment model (gold standard) or reference tissue input model |
Transport and binding/metabolism rate constants, vascular volume, perfusion, VT, Ki | Dynamic study | arterial plasma | no, except K1, Ki, or VT images | None, if comprehensive model. All plasma input methods are vulnerable to systematic errors in plasma metabolite analysis. Possibly nonspecific binding in plasma (fP) |
| BPND, R1 | Dynamic study | reference tissue | yes/no (depending on model) | Nonspecific binding in tissue (free fraction fND) | |
| Transport and binding/metabolism rate constants, vascular volume, volumes of distribution (VT), BPF, BPP, BPND, R1 | Dynamic study | arterial plasma and reference tissue | yes/no (depending on model) | Possibly nonspecific binding in plasma and/or tissue (fP and fND) | |
| Spectral analysis | Ki or VT, number of identifiable compartments | Dynamic study | arterial plasma | yes | Nonspecific binding in plasma and/or tissue (fP and fND) |
| Ratio | BPND | Dynamic study | reference tissue | yes | Nonspecific binding in tissue (fND), vascular volume, bias dependent on BP |
| Ratio, approaches BPND | Single scan | reference tissue | yes | Time from injection, nonspecific binding in tissue (fND), vascular volume | |
| Multiple-time graphical analysis (MTGA): Patlak and Logan plots | Ki | Dynamic study | arterial plasma | yes | Errors in plasma metabolite analysis, nonspecific binding to plasma proteins (fP) |
| Kiref | Dynamic study | reference tissue | yes | Nonspecific binding in tissue (fND) | |
| VT | Dynamic study | arterial plasma | yes | Errors in plasma metabolite analysis, nonspecific binding in plasma and tissue (fP and fND) | |
| DVR (VT/VND) | Dynamic study | reference tissue | yes | Nonspecific binding in tissue (fND), reference k2 | |
| Dual time point Patlak plot | Ki | Two late scans | arterial plasma at time of scans | yes | Population average of Patlak y axis intercept |
| Kiref | Two late scans | reference tissue at time of scans | yes | Population average of reference tissue AUC | |
| Fractional uptake rate | FUR, approaches Ki | Single scan | arterial plasma | yes | Errors in plasma metabolite analysis, distribution volumes of free and nonspecifically bound radioligand, vascular volume |
| Standardized uptake value | SUV | Single scan | i.d. | yes | Perfusion, peripheral clearance, nonspecific binding in plasma and tissue (fP and fND), vascular volume, dose extravasation |
| Autoradiography (ARG) | Perfusion (f) | Single scan | arterial blood | yes | Partition coefficient (p), vascular volume, time delay |
The outcome of many of the methods is the (equilibrium) volume of distribution VT (or DV). If valid reference region exists, the regional distribution volume ratio can be calculated as DVR = DVROI/DVREFERENCE. This, in turn, relates to the binding potential BPND: BPND = DVR - 1. However, this measure is vulnerable to change of nonspecific binding in tissue.
Bolus + infusion studies
| Method | Outcome | Requirements for | Applicable to parametric imaging | Vulnerabilities | |
|---|---|---|---|---|---|
| PET scan | Input | ||||
| Ratio | BPND | Single scan | reference tissue | yes | nonspecific binding in tissue (fND), vascular volume |
| VT | Single scan | venous plasma | yes | nonspecific binding in tissue (fND), vascular volume | |
Computer-aided diagnosis (CAD)
Computer-aided diagnosis aims to help physicians in the interpretation of medical images, combining physics, mathematics, statistics, medicine, and artificial intelligence. CAD systems are organ- and disease-specific, and the applications are rapidly expanding (Suzuki & Chen, 2018).
Literature
Bertoldo A, Rizzo G, Veronese M. Deriving physiological information from PET images: from SUV to compartmental modelling. Clin Transl Imaging 2014; 2: 239-251. doi: 10.1007/s40336-014-0067-x.
Gjedde A, Wong DF. Mathematical modeling and the quantification of brain dynamics. Neuromethods 2012; 71: 23-39. doi: 10.1007/7657_2012_55.
Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN, Holden J, Houle S, Huang SC, Ichise M, Iida H, Ito H, Kimura Y, Koeppe RA, Knudsen GM, Knuuti J, Lammertsma AA, Laruelle M, Logan J, Maguire RP, Mintun MA, Morris ED, Parsey R, Price JC, Slifstein M, Sossi V, Suhara T, Votaw JR, Wong DF, Carson RE. Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 2007; 27(9): 1533-1539.
Logan J, Alexoff D, Kriplani A. Simplifications in analyzing positron emission tomography data: effects on outcome measures. Nucl Med Biol. 2007; 34: 743-756.
Slifstein M, Laruelle M. Models and methods for derivation of in vivo neuroreceptor parameters with PET and SPECT reversible radiotracers. Nucl Med Biol. 2001; 28: 595-608.
Suzuki K, Chen Y (eds.): Artificial Intelligence in Decision Support Systems for Diagnosis in Medical Imaging. Springer, 2018. doi: 10.1007/978-3-319-68843-5.
Wong DF, Gründer G, Brasic JR. Brain imaging research: Does the science serve clinical practice? Int Rev Psychiatry 2007; 19(5): 541-558.
Updated at: 2018-08-13
Created at: 2008-11-28
Written by: Vesa Oikonen