D

scanner's axis

Fully 3D

Fully 3D

Figure 8. Diagram (not to scale) of the axial section of a 16-ring cylindrical PET tomograph operated in 2D mode (top) and 3D mode (bottom). In the 2D acquisition mode, the rings are separated by annular shielding (septa) and only in-plane and cross-plane LORs are allowed. In the 3D acquisition mode, the septa are retracted and coincidences allowed between any 2 rings.

Figure 8. Diagram (not to scale) of the axial section of a 16-ring cylindrical PET tomograph operated in 2D mode (top) and 3D mode (bottom). In the 2D acquisition mode, the rings are separated by annular shielding (septa) and only in-plane and cross-plane LORs are allowed. In the 3D acquisition mode, the septa are retracted and coincidences allowed between any 2 rings.

images representing the tracer distribution in the planes between the detector rings (cross planes). Hence, a 3D PET scanner consisting of N detector rings gives (2N-1) images representing the tracer distribution in adjacent cross-sections of a patient. Transverse sections of the image volume obtained are then stacked together to allow the spatial distribution of the radiopharma-ceutical to be viewed in 3D.

However, the use of septa also compromises the advantage of electronic collimation which otherwise records coincidence events efficiently within a cone-beam geometry and not just from the fan-beam geometry of the 2D mode. Thus, the 3D mode increases coincidence efficiency by about a factor of around five in comparison to a 2D acquisition. This can be accomplished by removing the interplane septa, at the expense of increasing scattered radiation. The septa themselves subtend an appreciable solid angle in the camera FOV, and as a result they have a significant shadowing effect on the detectors which can be as high as 50%. The geometric efficiency of the 3D mode requires that the detector be designed with a high count rate performance, which must be considered carefully in the selection of detector material (e.g. BGO have a relatively slow decay time). In fully 3D PET, the reconstruction algorithm must account for both oblique LORs formed from coincidences detected between different detector rings and for the increase in number of LORs which in turn depends on the number of crystal rings and degree of rebinning in comparison to 2D acquisitions. Different methods are used by the various scanner manufacturers to account for these characteristics. In addition, rebinning in 2D mode causes a variation in sensitivity along the axial FOV. In 3D mode, there is a much stronger variation in sensitivity (Figure 9) which peaks in the centre of the axial FOV.58

Likewise, different trues and randoms rates and different levels of scatter in 2D and 3D modes make it difficult to assess the real advantages of 3D vs 2D imaging. In order to more realistically assess these advantages, the concept of noise equivalent counts (NEC) has been proposed by Strother et al.59 and was defined as:

T + 2fR w where T and R are the total (i.e. trues+scatter) and randoms rates, SF is the scatter fraction and f is the fraction of the maximum transaxial FOV subtended by a centred 24 cm diameter field. The factor of two in the denominator results from randoms subtraction using data collected in a delayed coincidence window. The NEC quantifies the useful counts being acquired after applying perfect correction techniques for the physical effects and can be related to the global signal-to-noise ratio. As a result, the NEC has been extensively used to assess the real improvement obtained from 3D vs. 2D data collection strategies. The NEC does not, however, take into account all design or operational parameters that affect image quality.

0 0

Post a comment