Standardized uptake value

The standardized uptake value (SUV) is a nuclear medicine term, used in positron emission tomography (PET) as well as in modern calibrated Single Photon Emission Tomography (SPECT) imaging for a (semi)quantitative analysis.[1] Its use is particularly common in the analysis of [¹⁸F]fluorodeoxyglucose ([¹⁸F]FDG) images of cancer patients. It can also be used with other PET agents especially when no arterial input function is available for more detailed pharmacokinetic modeling. Otherwise measures like the fractional uptake rate (FUR) or parameters from more advanced pharmacokinetic modeling may be preferable.

3-dimensional [¹⁸F]FDG-PET image with 3D ROI generated by a threshold based algorithm. The blue dot in the MIP image bottom right marks the maximum SUV within the ROI.

The SUV is the ratio of the image derived radioactivity concentration c_img and the whole body concentration of the injected radioactivity c_inj,

${\displaystyle SUV={\frac {\mathit {c_{img}}}{\mathit {c_{inj}}}}}$

While this equation looks simple, there are a number of details that need to be discussed, such as (1) the origin of c_img data, (2) the origin of c_inj data, (3) time, and (4) units.

The c_img data may be the pixel intensities of a calibrated PET image. Calculated SUV data can then be visualized as parametric SUV image. Alternatively, groups of such pixels may be selected e.g. by manually drawing or otherwise segmenting a region of interest (ROI) on the PET image. Then e.g. the average intensity of that ROI may be used as c_img input to calculate SUVs.

The c_inj value is calculated as ratio of two independent measurements: the injected radioactivity (injected dose, ID) and the body weight (BW) of the subject. The ID can be estimated e.g. as difference in the radioactivity of the syringe before and after injection, if deemed necessary with correction for physical decay between each of those measurements and the time of injection. Conventionally the time of injection is t=0. This reference concentration represents the hypothetical case of an even distribution of the injected radioactivity across the whole body. SUV values thus quantify the measured deviation from this even radioactivity distribution.

The injection of radioactivity is often followed by a waiting time interval and then a time span during which the PET image data are acquired. After image reconstruction, the image c_img (t) data need to be decay corrected to the injection time point t=0. The time point t may be the image acquisition start time, or in case of a long acquisition duration e.g. the midpoint of the PET image acquisition may be more appropriate. This decay correction needs to be done for each image in case of a series of images acquired after a single injection ("dynamic imaging").

The unit of c_img is MBq/mL or equivalent, based on (a) the pixel intensity calibrated with a radioactive source ("phantom") itself of known radioactivity and volume, and (b) the pixel volume or ROI volume. The unit of c_inj is MBq/g or equivalent, based on the measured radioactivity and the subject's body weight. This would give SUV in units of g/mL or equivalent. However, SUV is typically presented as a unitless parameter. The reason is that the ROI is usually defined over soft tissue which has a mass density of approximately 1 g/mL. Thus the image derived intensities are implicitly converted by dividing by 1 g/mL to yield c_img in the same units as c_inj. This results in a unitless SUV parameter.

In summary this gives the following equation to calculate SUV at time t post injection,

${\displaystyle SUV(t)={\frac {\mathit {c_{img}(t)}}{\mathit {ID/BW}}}}$

with (1) the radioactivity measured from an image acquired at (or around) the time t, decay corrected to t=0 and converted from a volume to a mass based unit via the factor 1/(1 g/mL), (2) the injected dose ID at t=0, and (3) the body weight BW (on the day of imaging).

In addition, some authors replace the body weight by the lean body weight[2] or the body surface area.[3]

Also for c(t) from a region of interest, different measures are found in the literature, e.g. the maximum intensity value within the ROI, the mean intensity value of the ROI,[4] or the mean intensity value of the ROI after applying an intensity threshold (thus excluding a number of pixels of the ROI).

The SUV can be significantly affected among other things by image noise, low image resolution and/or user biased ROI selection.[5] For the semiquantitative analysis of [¹⁸F]FDG uptake in tissue or tumor, several corrections have been recommended (see [6] and references therein).

The ratio of the SUV from two different regions within the same PET image (i.e. from a target and a reference region) is commonly abbreviated as SUVR. An example is the ratio of regional Pittsburgh compound B PET signal intensity to the average of a wider region.[7] For the SUVR, the injected activity, the body weight and the volume to mass conversion factor that are all part of the SUV calculation, cancel:

${\displaystyle {\mathit {SUVR}}={\frac {\mathit {SUV_{target}}}{\mathit {SUV_{reference}}}}={\frac {\mathit {c_{img,target}}}{\mathit {c_{img,reference}}}}}$

The SUV concept has been only begun to be tested in connection with other radiotracers such as fluorothymidine F-18 ([¹⁸F]FLT) and conclusions on its usefulness and robustness in these cases might be premature.[8]

In summary, SUVs are a convenient measure for the evaluation of [¹⁸F]FDG PET images within a subject to study e.g. therapy monitoring/therapy response, and also for comparison between subjects. However, care has to be taken with respect to its pitfalls and with respect to the interpretation of the results.