Brain SPECT: improved displays for enhanced results
As software capabilities develop it seems imperative, besides working on quantification techniques, not to lose sight of the individual patient’ images. An improved visualization will go a long way towards continuing to spread the notion of Brain SPECT as a key functional imaging procedure. To this effect the end result displays should be comprehensive, detailed, precise, consistent and.…user friendly. In other words displays should become a most efficient, easy to relate to, communication tool.
There are essentially two ways to look at brain SPECT images:
- In a static mode (A) using the native images and / or using normalized images e.g. obtained after co-registration to a standard atlas / template .
- In a dynamic / cinematic mode (B).
- The native imagebasic display oforthogonal slices ( fig 1 columns 1-3). While this helps in a number of cases there is general agreement that significant improvements can be easily obtained by :
- Using an additional set of slices along the temporallobes axis (fig 1 column 4)
b .Using thresholded volumetric displays. One automatic way available applies four standardized thresholds on six views of the brain volume (fig 2 –only 2/4 thresholds and one single view are shown-) .
A most significant contribution to the requirements outlined above is obtained when adding he second type of images .
2.The normalized e.g. template co-registered displays of brain surface and volume. Among their advantages are the ease of discerning features on brain projections and volumes in addition to the ease of intra and between patient comparisons (given the fact that size / shape will be the same across subjects).
- The original normalization has been to the Talairach atlas as for ex. as done in the Neurostat algorithm (1 )originally used for statistical analysis but which can also be used for each individual patient’s functional map on eighttwo dimensional surface projections(fig 3)
- A next step introduced recently (2 )uses these same Talairach normalized images but for a format of 6 views of volumetric surface display views (fig 4)
- More recently a more detailed brain template called the MNI brain(Montreal Neurological Institute) has become available and theoretically the brain SPECT results can be co-registered to it .
- Dynamic / cinematic mode: there are multiple combinations possible and the ones presently available (2 ) are the following:
- The cine display of two of the volumetric-surface projections in the X & Y axis (fig 4).
- The cine display of three levels of rotating thresholded volumes in the X & Y axis . This enables a pertinent self explanatory evaluation of location and size of cortical as well as subcortical abnormalities (fig 5).
Two more question need to be addressed :
- Which color code ? Brain SPECT being a functional imaging , requires color displays which enable semi-quantitative ways of evaluating differential function. Nonetheless the question “which color code” is often heard. We selected here 4 out of several color schemes in existence : UBIQ a code with an easy to remember sequence of discrete shades (where the U & I stand for University of Illinois), then the so called “ cool” and “hot iron” (continuously transitioning colors ) and, the “Picker 20” discrete code .
- Is such a sequence of intricate supplementary data visualizations applicable in a routine clinical environment ? The answer is “yes” , it can be easily done via automation (2 ) . Indeed, with currently available software(2) a technologist is only required to provide two simple inputs : (1) indicate the file location containing the single slice, reconstructed / attenuation corrected images and (2) select from a list of color codes the one desired. In less than one minute a 5 page PDF file is then available for distribution whether via screen viewing (including full cinematic displays) and /or via color prints.
Illustration should be placed here and occupy the whole width of the column
A final thought : better displays and automation will not only enhance the value of results obtained but, will also make it much easier to achieve Brain SPECT standardization.
- Minoshima, S., Koeppe, R.A., Frey, K.A., & Kuhl, D.E. (1994). Anatomical standardization: linear scaling and nonlinear warping of functional brain images. Journal of Nuclear Medicine, 35(9), 1528-1537 and: http://www.smartbrief.com/s/2009/03/access-neurostat-3d-ssp-brain-pet-or-spect-analysis
- Good Lion Imaging , Columbia MD
Dan G. Pavel M.D.
Director PathFinder Brain SPECT Imaging
at The Neuroscience Center, Deerfield IL 60015