Tweet
Dear Biomch-L Readers,
I am using metallic (iron) markers in CT-Scan to build some
reference systems.
I used iron balls instead of aluminium (aluminum for our US
colleagues) because the latter was no available in small diameter
(0.1mm). I knew iron will give artefacts, but hoped that the artefacts
will be homogeneous and that the centroid of the landmarks wouldn't be
affected.
Sub-pixel centroid location was then used and the result are
satisfactory. But looking at the 3D models of the markers (generated
from the CT-Scan) I'm a bit puzzled!
Interested readers can find two pictures (z1.gif and z8.gif) showing my
problem at:
http://isb.ri.ccf.org/data/vsj/im001.html
They show the geometry model of one marker along with the 3D
reference system of the CT-Scan. The marker model has been generated from a
dataset with the following helicoidal parameters:
-slice thickness: 1.1mm
-slice increment: 0.2mm.
-pitch: 0.7D.
-voltage: 120 KV
-amperage: 293 mA.
The original marker size is 1mm of diameter.
FILE z8.gif:
The dimensions of the model in the X-Y directions (which
is the plane of each slice) are quite similar (one tick on the axes
is 1mm). The models have +/- a 2.5 mm diameter (instead of the original 1mm),
but averaging the X-Y coordinates with the pixel grey values finds the
right centroid.
FILE z1.gif:
Now, along the Z dimension the dimension is close to 3mm and the
marker has a "zeppelin-like" shape. Here again averaging the grey
value gives satisfactory sub-pixel centroid location, but the results
would be probably better without these artefacts (I guess!?).
My questions are:
1- Does somebody know the relationships between
-the pitch of the helix used by the CT-Scan during the data
acquisition and
- the signal from a metallic structure located in the X-Ray field ?
I have the feeling (but no numbers!) that the "zeppelin-like" shape is
related to this pitch ...
2- Is there a way to avoid this artefact when using spiral CT and iron markers?
3- On FILE z1.gif: the central part seems to show a spiral groove
(which main direction seems to go to Z+). Do you think it is related
to the spiral path of the CT-Scan?
4- I am using spiral sequences to allow very thin slice increment which
should in turn allows more accurate sub-pixel centroid location. Is this
statement correct ?
5- Any (easy) alternative ?
6- I am aware of (I think) most of the publications using markers as
landmarks in a CT field,
but found nothing about that kind of artefacts ? Any pointers ?
Thanks a lot in advance! I'm looking forwards for your answers!!
Oops, I forgot to tell you that all markers inside a same dataset gives
similar signals (same amplitude, same direction, same artefacts, ...).
S. VAN SINT JAN
--------------------------------------------------------------------------
Serge VAN SINT JAN, Ph.D. | In sabbatical from:
School of Computing Sciences | Department of Human Anatomy and Embryology
De Montfort University | University of Brussels, Belgium
Hammerwood Gate |
Milton Keynes MK7 6HP - UK
phone: + 44 1908 695511 (ext. 4146)
+ 44 1908 834933
fax: + 44 1908 834948
email: sintjans@dmu.ac.uk
WWW: http://homepages.ulb.ac.be/~sintjans
--------------------------------------------------------------------------
---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------
I am using metallic (iron) markers in CT-Scan to build some
reference systems.
I used iron balls instead of aluminium (aluminum for our US
colleagues) because the latter was no available in small diameter
(0.1mm). I knew iron will give artefacts, but hoped that the artefacts
will be homogeneous and that the centroid of the landmarks wouldn't be
affected.
Sub-pixel centroid location was then used and the result are
satisfactory. But looking at the 3D models of the markers (generated
from the CT-Scan) I'm a bit puzzled!
Interested readers can find two pictures (z1.gif and z8.gif) showing my
problem at:
http://isb.ri.ccf.org/data/vsj/im001.html
They show the geometry model of one marker along with the 3D
reference system of the CT-Scan. The marker model has been generated from a
dataset with the following helicoidal parameters:
-slice thickness: 1.1mm
-slice increment: 0.2mm.
-pitch: 0.7D.
-voltage: 120 KV
-amperage: 293 mA.
The original marker size is 1mm of diameter.
FILE z8.gif:
The dimensions of the model in the X-Y directions (which
is the plane of each slice) are quite similar (one tick on the axes
is 1mm). The models have +/- a 2.5 mm diameter (instead of the original 1mm),
but averaging the X-Y coordinates with the pixel grey values finds the
right centroid.
FILE z1.gif:
Now, along the Z dimension the dimension is close to 3mm and the
marker has a "zeppelin-like" shape. Here again averaging the grey
value gives satisfactory sub-pixel centroid location, but the results
would be probably better without these artefacts (I guess!?).
My questions are:
1- Does somebody know the relationships between
-the pitch of the helix used by the CT-Scan during the data
acquisition and
- the signal from a metallic structure located in the X-Ray field ?
I have the feeling (but no numbers!) that the "zeppelin-like" shape is
related to this pitch ...
2- Is there a way to avoid this artefact when using spiral CT and iron markers?
3- On FILE z1.gif: the central part seems to show a spiral groove
(which main direction seems to go to Z+). Do you think it is related
to the spiral path of the CT-Scan?
4- I am using spiral sequences to allow very thin slice increment which
should in turn allows more accurate sub-pixel centroid location. Is this
statement correct ?
5- Any (easy) alternative ?
6- I am aware of (I think) most of the publications using markers as
landmarks in a CT field,
but found nothing about that kind of artefacts ? Any pointers ?
Thanks a lot in advance! I'm looking forwards for your answers!!
Oops, I forgot to tell you that all markers inside a same dataset gives
similar signals (same amplitude, same direction, same artefacts, ...).
S. VAN SINT JAN
--------------------------------------------------------------------------
Serge VAN SINT JAN, Ph.D. | In sabbatical from:
School of Computing Sciences | Department of Human Anatomy and Embryology
De Montfort University | University of Brussels, Belgium
Hammerwood Gate |
Milton Keynes MK7 6HP - UK
phone: + 44 1908 695511 (ext. 4146)
+ 44 1908 834933
fax: + 44 1908 834948
email: sintjans@dmu.ac.uk
WWW: http://homepages.ulb.ac.be/~sintjans
--------------------------------------------------------------------------
---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------