[Insight-users] Tumor Measurements..

[Miller, James V (Research)]

Volume criteria (change in cubic mm) would be most 
accurate measure of tumor volume.  This is easy to 
do from any label map (segmentation) using the 
RelabelComponentImageFilter. If the segmentation results
in a mesh, there are established techniques for calculating
volumes from polyhedron (based on Green's thereom I believe).

Note that volume changes very rapidly with a change
in diameter.  An almost imperceivable change in diameter
can result in a 25% change in volume. So changes in 
discretization (changes in pixel size or pixel alignment
with respect to the tumor) can impact tumor size greatly.

The RECIST measure was developed because it involved
measurements that a radiologist could perform reliably.
So while is may not be most accurate measurement (being 
based soley on 2D measurements), it is "robust" to the 
operator.

Another consideration is that radiologists "picks" a 
window/level setting when measuring tumors. 
Window/level settings can affect the "perceived" 
location of the boundary of a tumor.  Different 
clinicians will use different w/l and obtain different 
measurements.

Note, clinicians tend to measure a tumor as being
larger than what a computerized algorithm will report.
If you are using an edge-based algorithm, the tumor 
boundary is usually placed at the half power point
of the edge response. This tends to be half way 
between where the intensities transition from 
background to foreground.  Clinicians, on the other 
hand, tend to indicate the boundary of the tumor
as where the foreground starts to pull above the 
background. So boundaries defined by a clinician
tend to be "outside" the boundaries defined by a 
computerized algorithm.

Finally, the point spread function of the scanner
coupled with the discretization process and any
smoothing operations performed by an algorithm will
"move" the boundaries of a tumor. In other words, 
the imaged boundary will be displaced by any
smoothing operations.  The amount and direction 
of the displacement is a function of curvature
and scale. Since the boundaries are displaced, 
the volume measurements are biased.

Jim



-----Original Message-----
From: yasser salman [mailto:yass71 at yahoo.com]
Sent: Monday, November 01, 2004 3:47 PM
To: insight-users at itk.org
Subject: [Insight-users] Tumor Measurements..


Hi ALL..,
To any body interested in the field of tumor
segmentation  and measurements ,while i am doing my
survey i found that the methods of tumor measurements
based on the tumor's diameter . WHO(World Health
Organization) criteria defined a shrinkage of a tumor
as a decrease in the product of the largest
perpendicular diameters in the largest slice of the
tumor. RECIST criteria defines a tumor's shrinkage as
the decrease in the length of it's largest diameters,
which i think will be more accurate if we calculate
the no. of pixels in the tumor's area (we also can
convert it to physical coordinates in mm^2 cm^2)., 
and what is the problems for using Pixel based. 
can any body confirm me in this Issue or let me know
the right way to go on Or Back OFF!
Thanx  
regards..,
Yasser Salman...,




		
__________________________________ 
Do you Yahoo!? 
Check out the new Yahoo! Front Page. 
www.yahoo.com 
 

_______________________________________________
Insight-users mailing list
Insight-users at itk.org
http://www.itk.org/mailman/listinfo/insight-users