Editing Scanco User Instructions (section)

===Understand Morphological Parameters===
*Refer to [[Media:Reference_Document_-_Parameters_from_upat_list.doc|'''Explanation of UPAT_LIST Output''']] to understand the result file.
*Refer to [[Media:Reference_Document_-_Parameters_from_uct_list.doc|'''Explanation of UCT_LIST Output''']] to understand the result file.


All values are determined on a slice-by-slice basis, then averaged over all the slices to calculate morphological results:

:*BV/TV -> Sum all raw data within the GOBJ and divide by 1200 mg/cc HA
:*Tb.N  -> As per standard method
:*Tb.Th and Tb.Sp -> Derived from BV/TV and Tb.N above (semi-derived method)
:*D100  -> Average density of entire cross section
:*DTrab -> Average density of trabecular region
:*Ct.Th -> Cortical area (from seg) divided by circumference (from GOBJ)
:*GOBJ circumference can be determined with /gobj_circum.

<pre>
BV/TV  -> VOX-BV/TV
BS/BV  -> TRI-BS/BV
Tb.Th  -> DT-Tb.Th
Tb.Sp  -> DT-Tb.Sp
Tb.N   -> DT-Tb.N
ConnD  -> Conn-Dens
SMI    -> TRI-SMI
DA     -> TRI-DA
</pre>

Usually I'd use the names on the left for the plots, but these correspond to the names on the right of the arrow in the Scanco output files.

<b>Note</b>, if you use the TRI-Tb.Th rather than the DT-Tb.Th*, then you are using the derived (plate-model) morphological parameters.  These are less accurate than the direct parameters.  They are calculated by triangulating the surface (marching cubes) to get BS, and then calculate thickness
Th = 2 * BV / BS, etc...

'''Mean Values'''
<pre>
mean 1: mean of bone AND marrow --> 'apparent density'
mean 2: mean of what was considered bone only (i.e. what was segmented) 
'bone tissue density'
</pre>

Note, there are five possible categories for the mean Mu.  They are assigned by setting <code>-region_number</code> in the <code>/voxgobj_scanco_param</code> function.  The region number is zero-based, so setting it to 0 corresponds to mean1, 1 corresponds to mean2, etc.  So, although mean1 and mean2 are typically apparent and tissue density, respectively, this can be adjusted in the IPL script by setting the region_number.  Mean3 might be useful for storing cortical bone density, for example.

<pre>
/voxgobj_scanco_param      Voxel or Mu Scanco Parameters
  -input                     Gray scale or Segmented AIM object
  -gobj_filename             GOBJ: input or Automatic
  -peel_iter                 Peel Iterations or Automatic
  -region_number             Region number for Mu only

! -region_number             0   ! mean1
! -region_number             1   ! mean2
! -region_number             2   ! mean3
! -region_number             3   ! mean4
! -region_number             4   ! mean5
</pre>

<b>What you should also know</b>: because of partial voluming effects, for Mean2, we are DISCARDING the 2 surface voxels of EVERY trabeculae - thus, to get meaningful Mean2, you have to do your scan with the chosen resolution high enough, such that the mean trabeculae is depicted by at the very least 5 pixels, better actually more than 6 pixels thickness. (You can do this check at the end in the XLS sheet: divide DT-Tb.Th* by el-size-mm.x, and you need to get numbers bigger than 6 or 7 or 10 or so.

For parameters such as ConnD where no GOBJ is taken as input, the TV is determined from the processing log stored in an AIM that has previously had a GOBJ applied (stored as the relative region of interest; ROI).  So, it is important to apply a GOBJ prior to ConnD measurements to ensure the TV is correctly determined (and not taken as the whole AIM).

'''Trabecular Number (TbN)'''
The epsilon values can be set to account for differences in scanner resolution, and hence 'surface roughness' for individual pixel suppression.

For microct, for good resolution: if there is a pixel 'sticking out' from a flat surface we do not assign that pixel with a thickness value of 1, but we let it 'get eaten' by the bigger sphere that sits in the center of the trabeculae. So this is nice for 'noise'  (0.9/1.5). 

For XT resolution, where also many 'good' features are just one or two pixels wide, I liked a lower epsilon value better (0.9/0.5).


[[Image:Redirect.jpg|15px]] [[#top|''Return to Contents'']] or [[Main_Page|''Return to Main Page'']]