Let me describe a little more about my phantom studies for you as a reference.
1). Hoffman Brain Phantom: SPECT acquired on Triad camera. Manually draw ( not use a
mirror ROIs template ) symmetric RIOs on TV images. Then comparing average counts of
each pair of ROIs. For these SPECTs acquired with 120 degree CCW per head , there is
10 to 20 % asymmetric difference with a pattern that most higher reading of ROIs on
left side; For these SPECTs acquired with 360 degree CCW per head (same acquisition
time but off course less time per step), there is 5 % (a fewer 7 %) asymmetric
difference with higher left ROIs and higher right ROIs even distributed. I did not
use the cerebral hemisphere as the reference region.
2). Hoffman Brain Phantom and SPECT Phantom: SPECT acquired on Picker Prism 3000. 120
degree CCW per head, 5 scans v/s 360 degree CCW per head, 2 scans (same acquisition
time). I have same result as my study on Triad camera. For the SPECT phantom, I used
only those TVs in uniformity section of the phantom. Since it's uniformity section, I
just random drew muti-ROIs on one side and flipped them over. I have same result from
So both institutes now use the 360 degree protocol instead 120 degree one which it's
usually set by manufacture as default protocol.
"Chris Gottschalk, MD" wrote:
> Drs Glabus and Fang [and I eagerly await Isak's input as well; if he
> doesn't chime in, I will go find him...]
> I am fascinated by this interchange regarding assymtery in *phantom* data
> as opposed to *human beings* deemed "normal". the data from a phantom
> define, I think, what the MINIMUM range of assymetry you should accept for
> your scans, given that a phantom is uniform, by definition. However, whose
> data are we using to define what the normal range of assymetry is in
> [presumably] resting images of people in a gamma camera? Certainly, those
> values will vary according to which structures you are identifying with
> ROIs to define assymetry.
> Perhaps I envision the wrong thing: are you both referring to percent
> differences between [whole] hemispheric values?
> >> I have several questions and statement here related to your E-mail.
> >> 1). Our NM physicians use the 7% based on ' plenty of NM literature '
> >> that the range of asymmetry in SPEC scans of normal population is from
> >> 5% to 10%. They call 7-9 % in borderline and > 9 % as abnormal with
> >> traditional RIO method; I did some phantom studies using
> >> Hoffman Brain Phantom and found there is 5% asymmetry with optimum setting
> >> acquisition; CERETEC has a 5-10% asymmetry on normal scan according to
> >> Amersham (ECD?). It's almost impossible to get a actual
> >> normal pediatric subject scan and they all come in with some kind of
> >> So Iselected these less than 7 % asymmetry as a control group not a normal
> >> baseline.
> >On the basis of your observation with Ceretec, these thresholds sound
> >and certainly consistent with observations I've made and with phantom
> >using the JB003 brain phantom with Tc99m, but imaged with the Strichman
> >Neuro 900,
> >12-detector system.
> >Routinely clinicians (in Edinburgh) would look for asymmetry of around 20%
> >to be
> >deemed pathological, but opinions vary. *** ISAK PROHOVNIK ***, if you're
> >out there,
> >can you provide some input on this? (Apologies for shouting, but it's a
> >"heads up" call...copyright Darren Gitelman)
> >> Do you think that 'no major differences between a custom SPECT and
> >> supplied PET template for normalization results' can be extended to
> >> pediatric patient (assuming supplied PET made from adult)?
> >I'm not sure but it may be prudent to consider making a paediatric template
> >it's bound to be more unlike the adult (PET) template than an "elderly"
> >> 3). When I learnt the SPM from a local spmer, I have been instructed
> >> that I had to do at least Coregister and reslice and Spatial
> >> Normalization (Smooth as an option) on an object image to create a
> >> snr---.img file for Statistics. I have been suggested to use supplied
> >> PET as template then. But I have a impression from your
> >> E-mails that the Coregister step might not be necessary in the process
> >Routinely, one would not normally coregister and reslice a SPECT image
> >in a separate stage from normalisation against (whatever) template image, so
> >I think you may be referring to the issue of "piggy-back" spatial
> >where MRI derived parameters are applied to the SPECT image from the same
> >subject. In theory this should work, but my experience (and of others) has
> >it's better to treat these images separately for spatial normalisation. My
> >is that errors arise due to poor co-registration between SPECT and MRI
> >images - again
> >the reason is probably that our Neuro 900 have low spatial resolution in the
> >z-plane, with 12-14 slices of 8 - 10 mm slice thickness. If you have a gamma
> >which acquires a more complete data set, this may not be an issue.
> >"Assessing spatial transformation options in a SPECT and MRI study of
> >depression and dementia using SPM'96". Glabus, M & Ebmeier, K.P., 1999,
> >Neuroimage,9:6:Pt 2 of 2:S149.
> >Regards - Mike
> >Mike Glabus, PhD
> >Visiting Fellow in Functional Neuroimaging
> >Unit on Integrative Neuroimaging,
> >Clinical Brain Disorders Branch, NIMH, NIH
> >Building 10, Rm 4C101, 9000 Rockville Pike
> >Bethesda, MD, 20892-1365, USA
> >Tel: + 301 496 7864 FAX: + 301 496 7437
> >[log in to unmask]
> Christopher Gottschalk, MD *
> Assistant Professor of Neurology & Psychiatry *
> Yale School of Medicine *
> Mailing Adress: *
> VAMC [116-A] tel  932-5711 x4329*
> 950 Campbell Avenue FAX 937-4791*
> West Haven, CT 06516 *