----------------------------------------------------------------------
1.  Submit to: MI04; Sonka, and Fitzpatrick

2.  Title of Conference: Image Processing

3.  Paper Title: Alignment of Multimodality, 2D and 3D Breast Images

4.  Author(s)
George J. Grevera, Jayaram K. Udupa
Medical Image Processing Group (MIPG)
Department of Radiology
423 Guardian Drive
University of Pennsylvania Health System
Philadelphia, PA  19104
voice: (215) 662-6780
fax:   (215) 898-9145
email: grevera@mipg.upenn.edu, jay@mipg.upenn.edu

5.  Presentation: Oral

6.  Five Paragraph Summary

6.1.  Purpose
In a larger effort, we are studying methods to improve the specificity 
of diagnosis of breast cancer by combining the complementary information 
from multiple imaging modalities.  Merging information is important
for a number of reasons.  For example, contrast uptake curves are an 
indication of malignancy so the reduction of the influence of motion 
artifacts is important.  Also, the determination of anatomical locations
in corresponding images from various modalities is necessary to determine
the extent of regions of tissue.  To facilitate this fusion of multimodal 
information, registration becomes necessary.  We describe in this paper 
a framework in which 2D and 3D breast images from MRI, PET, Ultrasound, 
and Digital Mammography can be registered to facilitate this goal.

6.2.  Methods
Prior to image acquisition by any modality, an alignment grid
is drawn on the breast skin using a marking pen.  In general, prior to image
acquisition, modality specific markers (vitamin E capsules for MRI, FDG for PET,
and standard lead markers for Digital Mammography) are placed at the
indicated grid points.  Images are then acquired with the external markers
appearing on the images.  In the case of PET, diagnostic images are first
acquired in the absence of markers, then the cameras are turned off,
next the markers are put in place, and finally the cameras are turned back
on.  Ultrasound does not use any markers but uses a position tracking
device which is attached to the probe and is calibrated to the grid.

6.3.  Results
Given the similarity between pre- and post-contrast breast MRI,
it is not surprising that their alignment is most accurate.  Although 2D, by
tracking the position of the Ultrasound probe with respect to the external
markers during image acquisition, Ultrasound images may be aligned as well.
Using the external marker system, PET images may be aligned as well but the
lack of internal anatomical information inherent in PET images reduces the
accuracy of the subsequent registration.  Differences in patient positioning
during image acquisition (supine for PET and prone for MRI) pose problems.
In the future we will use a dedicated breast PET scanner which
will dramatically increase the accuracy of PET-MRI registration (because 
patient positioning will be similar to that of MRI).  Digital Mammography
to MRI poses the greatest challenge because: 1) Digital Mammography is a 
2D projection image of the 3D breast, and 2) the breast is subjected to
compression during the imaging procedure.

6.4.  New Work to be Presented
This is the first study that we are aware of that has undertaken the
difficult task of registering 2D and 3D images of such a highly deformable 
structure (the breast) across such a wide variety of modalities.  This 
represents some very preliminary results of a large, ambitious project.

6.5.  Conclusions
With varying degrees of accuracy, one may register images from
a variety of modalities using an external marker system.

7.  Keywords
registration, MRI, PET, mammography, breast imaging

8.  Brief Biography
George Grevera received a Ph.D. in Computer Science from the University
of Pennsylvania, an M.S. in Electrical Engineering from Drexel
University, and a B.S. in Computer Science and Biology Research
(double major, Cum Laude) from the University of Scranton.  He is currently
a Research Assistant Professor in the Medical Image Processing Group at
the University of Pennsylvania.  His current research interests are in
medical visualization, image processing, and image analysis.
----------------------------------------------------------------------