Clinical Instrumentation

- Number of projections: the angular increment should be in the order of sampling,  therefore the number of projections N should be N = Rp/d  where R = radius of rotation, and d = pixel size. The greater the number of steps, the lower the noise introduced by backprojection and better resolution. There is a compromise with respect to the length of study. Priority is given to the number of projections rather than the total number of counts per projection.              

- Total number of counts per projection: it must be as many as possible within the time interval that the patient can remain motionless (not appropriate to extend the study for more than 30 to 40 minutes). If the number of counts is low, the statistical error will be high and the image noise will be amplified with the reconstruction. In practice, it is advised that 25-50 x 103 counts per projection for a 64x64 matrix and approximately 50-100 x 103 counts per projection for a 128x128 matrix, be accumulated approximately. Adding slices can be an alternative to reduce the statistical error, but this operation is better to do post-reconstruction, as it is better to use slices of 1 pixel thick to calculate the coronal, sagittal or oblique.

- Acquisition time of each projection: it is determined from the counts / projection desired, according to the above, and the counts per second (CPS) coming from the patient. (Example: If you want 50 kctas / projection in a patient who gives 2k CPS, you should select 25 sec / projection). The time selected should not be too long, because the patient would move affecting the quality of the image.

- Total rotation angle: 360th (solid organs that have structures in all projections, such as liver, skull, bone) or 180th (for axial organs (heart, thyroid, spine, jaw, orthotopic kidney).

- Diameter of rotation: the shortest possible to minimize the source-detector distance and optimize the spatial resolution.

- Matrix size: 64x64 or 128x128, depending mainly on the spatial resolution of the detector (to assure not to loose resolution, pixel size (p) should be p = FWHM / 3 where FWHM, characterizes the spatial resolution FWHM) and the energy of the radionuclide employed (low energy, FWHM 12 mm. ==> 128x128, medium energy  FWHM 20mm ==> 64x64 ).

- Reconstruction algorithm: According to available software.

- Filter: is chosen according to the spatial resolution, the total number of counts in the study, the characteristics of the organ of interest. a) prefilter: generally refers to a smoothing filter of 9 points which is used when there is low statistics in the acquisition or when the 3D reconstruction  program (sagittal and coronal) does not have a filter in between slices. b) ramp filter: always apply. It is a filter that allows high frequencies to pass. (high pass) limiting the low frequencies. This will reduce the contribution of background in the image. Conversely, this filter increases the high frequency component (noise). c) To smooth and restore filters, reduce the noise component. For smoothing: Butterworth, Hanning (low pass). Do not modify the low frequencies, but limit the high (noise). Removing high frequencies reduces the noise, but smoothing the images also reduces the fine details. Recovering filters: Wiener, Metz, Gaussian filter (pass band). Produce increase of contrast and resolution by boosting the midrange frequencies, and noise reduction by limiting the high frequencies.

- Attenuation correction: a) the method of Chang (constant damping factor) and b) Variable attenuation factor measuring transmission data for each projection.

- Correction of scattered radiation: the method of 2, 3 or multiple windows depending on the equipment options.

Page 6    Home       Index      Clinical Nuclear Medicine       Therapeutic Methods      Previous Page   Next Page