Gamma camera scintigraphy is confronted by a number of inherent limitations that may impact on the quality or integrity of image data sets. Noise and spatial resolution have already been mentioned. The very nature of nuclear scintigraphy offers another significant limitation. The imaging principle relies on the external detection of radionuclide emissions from deep within a patient. The path of emitted gamma rays can be tortuous or terminal. The interaction of photons in matter are well described in many nuclear medicine physics textbooks; Compton scattering, the photoelectric effect, pair production, photon attenuation and, on occasion, coherent scattering. Imaging relies on physical discrimination (collimator) and energy discrimination (pulse height analyzer) to eliminate object/image photon misalignment. In the case of the former, photons not generally incident perpendicular to the collimator are eliminated. In the case of the latter, photons with significant loss in energy due to scattering which by chance are incident perpendicular to the detector, are excluded from the image.

Despite these measures, a number of anomalies familiar to Nuclear Medicine personnel and attributed to Compton scattering appear in scintigraphic images. It is a common observation, particularly in the patient with a full figure, on whole body bone scans and whole body gallium scans to note an outline of the body edge that is marginally more intense than adjacent soft tissues. The body halo may also manifest as a line of apparent accumulation in locations of appositional skin (e.g. where the inner thighs meet, the breast fold etc.). The dynamic phase of a bone scan of the hands, when using the cubital fossa as the injection site, often shows what looks like a blood pool image of the hand and forearm prior to actual injection (Fig 1). While anecdotally, these phenomena are attributed to ‘scatter', there is a paucity of discussion in the literature explaining the process whereby scattered photons avoid energy discrimination.

Figure 1

Figure 1: Phantom appearance of radiotracer in the hand and arm prior to actual dose administration.

During the course of investigating extraction of 99mTc pertechnetate by grapes, apparatus were conceived to simultaneously measure 99mTc pertechnetate accumulation and water absorption. The grapes were imaged by the vertical gamma camera while suspended above a beaker of 99mTc pertechnetate on a set of scales. The level of 99mTc activity was outside the field of view and shielded using lead. The scales had a purpose build glass housing to minimize the effects of the environment on the accuracy of measurements. For the purposes of these experiments, the lid remained open. Figure 2 shows the scintigraphic image of the apparatus prior to obvious grape accumulation. The glass housing and lid were noted to be prominent despite not containing any radioactive material. The position of the 99mTc below and outside the field of view would require significant scatter of photons to create the appearance; presumably sufficient to result in exclusion by energy discrimination.

Figure 2

Figure 2: Non radioactive glass box with open lid encasing a beaker of 99mTc on a set of scales.

A number of brief experiments were subsequently performed to better define the observed phenomena. A non radioactive glass sheet was positioned perpendicular to the gamma camera. A collimated 99mTc source was positioned in a variety of locations outside the field of view. The gamma camera recorded a vertical distribution of activity corresponding to the glass sheet when the 99mTc source was positioned 10cm or more forward of the level of the collimator (Fig 3).

Figure 6

Figure 4: Brain and thyroid phantoms with target and background ROIs positions used for all 24 phantom studies.

Not unlike the potential small translocation errors associated with the angle of variation from perpendicular defined by the collimator ‘hole' length and diameter, limitations of the energy discriminator might also result in a small degradation of resolution. A physical principle rarely discussed with respect to gamma camera energy discrimination is that the residual energy of a scattered photon is a function of the actual starting energy and the angle through which it was scattered. Consequently, higher energy photons will result in a greater percentage loss of energy after scatter through the same angle as a lower energy photon. As an example, a 100keV photon retains 90% of its energy after 60 degrees of scatter while a 5MeV photon will only retain 17% through the same angle (Bushberg 2002). The energy of the scattered photon can be determined by the following equation (Bushberg 2002; Cherry, Sorenson & Phelps 2004):

Figure 4

where

Esc is the energy of the scattered photon,

Eo is the energy of the original photon, and

θ is the angle through which the photon was scattered.

From this equation it is clear that even 90 degree scatter of a 99mTc photon leaves a scattered photon with an 114keV energy. More importantly in the context of this discussion, a 20% symmetrical energy window about the 140keV photopeak of 99mTc will include photons scattered by as much as 60 degrees (126keV). Reducing the window to 15% will only exclude photons scattered by 45 degrees or greater.

Standard symmetrical energy discrimination about the photopeak is not a useful tool in eliminating small angle scattered photons from subsequent images. Assuming a model where the scattered photon only alters direction once on its path through the body, line of response translocation can range with origin depth from small resolution degrading translocations through to large background contributing translocations.

In order to determine the impact of scattered photons on image resolution and potential benefits to be gained by alternate energy window strategies, a series of 99mTc line sources were acquired. Line sources were acquired on a Philips Prism 1000 with high resolution parallel hole collimation. A 512 matrix was employed with a 2.0 zoom. Multiple energy windows were employed to provide simultaneous acquisition of data for comparison (table 1). Each energy window combination was acquired with the line source on the collimator surface, against 4.0cm of water attenuation and against 8.0cm of water attenuation. The resulting 198 line spread functions each had the full width half maximum (FWHM) calculated.

Figure 5

Table 1: A summary of energy windows employed in data acquisition.

In order to determine the impact of scattered photons on image contrast, a series of 99mTc thyroid and brain phantoms were acquired. Both the planar thyroid and planar brain phantoms were acquired on a Philips Prism 1000 with high resolution parallel hole collimation. A 512 matrix was employed with a 3.2 zoom. Dual energy windows were positioned to capture a 10% window below 140keV and a 10% window above 140keV. The thyroid and the brain phantoms were each positioned on the collimator surface, against 4.0cm of water attenuation and against 8.0cm of water attenuation. The resulting 18 phantom studies each had regions of interest (ROI) drawn in target and background areas for calculation of image contrast (Fig 4).

Figure 7

Figure 5: Brain phantom with grey matter, white matter and CSF ROIs.

The brain phantom was subsequently imaged to produce three otherwise identical planar studies with a single energy window of; 20% centered on 140keV, 15% centered on 140keV, and 15% centered on 144keV. The resulting images had four ROIs drawn to include grey matter, white matter, cerebral spinal fluid (CSF) and background outside the physical limits of the phantom (Fig 5).

Figure 8

The statistical significance was calculated using Student's t test and the F test analysis of variances was used to determine statistically significant differences within grouped data. A P value less than 0.05 was considered significant. The differences between independent means and proportions was calculated with a 95% confidence interval (CI). Image contrast was determined as (Rzeszotarski 1999):

Figure 9

Table 2: Summary of mean FWHM values for each energy window stratified against the attenuation thickness.

Where,

C is the object / background contrast

T is the object counts per pixel

B is the background counts per pixel