Recent advancements in the technology of medical imaging and the content-based
analysis of multimedia have resulted in new opportunities in the field
of the medical image management. These days people are talking about
the concept of "Future Hospitals", in which all medical
documents including images are in digitial format. It's extremely important
in such hospitals with enromous amount of medical image content to have
content management tools with which one can interact with the data efficiently.
One important and very common medical imaging modality is the Echocardiogram
video, which captures the function and the structure of the heart
from different viewing angles. Echocardiograms are highly structured,
rendering rich spatio-temporal patterns in the video sequence. We
exploit this structure to create statistical models of their content.
These models are then used to parse the echocardiogram videos into
their constituent units of structure. By recovering the structural
elements of the echocardiogram video we can summarize its content
for efficient browsing and access. See image for a snapshot of the
browsing interface where the user could navigate the summary of the
echocardiogram and only watch the video segments of interest. The
elements of the content of the echocardiogram could be linked to a
synchronized 3D graphic model for a more intuitive interface, and
to the contextual text information usually accompanying these videos.
The challenges of this research are:
- Automatic annotation of the cardiac objects
- Automatic recognition of the different view angles of the heart
- Modelling the activity of the heart operation by multi-part spatio-temporal
models
This research is part of the "DLI-Phase 2: A Patient
Care Digital Library: Personalized Retrieval Summarization of Multimedia
Information (PERSIVAL)" project funded by the National
Science Foundation (NSF Award 9817434).
(3D Heart Model courtesy of New York
University School of Medicine)
A brief introduction to echo video is included here.
S. Ebadollahi, S.-F. Chang,
H. Wu, "Echocardiogram Video: Indexing and Summarization",
Presented in Watson Workshop on Multimedia (WWM), June 2003, IBM.
T.J. Watson Research Center, Yorktown, NY, 2003 (Awarded)
S. Ebadollahi, S.-F. Chang,
H. Wu, Echocardiogram
Videos: Summarization, Temporal Segmentation, and Browsing, Proceedings
of International Conference on Image Processing 2002, September 22-25,
2002, Rochester, NY, 2002.
(PS.GZ/PDF)
S. Ebadollahi, S.-F. Chang, H. Wu, "DEVL: Digital Echocardiogram
Video Library", Proceedings of the European Conference on Picture
Archiving and Communication System (EuroPACS'02), September 5-7, 2002,
Oulu, Finland, 2002. (PS.GZ/PDF)
S. Ebadollahi, S.-F. Chang,
"View Segmentation and Static/Dynamic Summary Generation for
Echocardiogram Videos", Proceedings of First Joint IEEE/ACM Conference
on Digital Libraries (JCDL'01), June 24-28, 2001, Raonake, VA, 2001
(PDF)
S. Ebadollahi, S.-F. Chang,
H. Wu, S. Takoma, "Indexing and Summarization of Echocardiogram
Videos", Proceedings of Scientific Session of American College
of Cardiology (ACC'01), March 2001, Orlando, FL, 2001 (PDF)
S. Ebadollahi, S.-F. Chang,
H. Wu,"Echocardiogram Video Summarization", Proceedings
of SPIE Medical Imaging 2001, San Diego, CA, 2001 (PDF)