Image Representation using Multi-dimensional Biomedical Functional and Anatomical Features
Use of image-derived features to represent the complex images volumes for data management and image understanding.
Professor David Feng, Associate Professor Jinman Kim.
It is well recognized that healthcare must capitalize on Information technology innovations in order to improve diagnosis and patient care. With the ever-growing image databases in the hospital and other clinical environments, the ability to represent the images based on its image-derived “features” has the potential to significant impact on how the image data are currently managed and used in routine clinical practices. Some of potential benefits are:
- Efficient image visualization, i.e., automated focus-of-attention;
- Improved image understanding e.g., pre-calculated volumes-of-interests for quantitative diagnosis; as well as
- Searching and retrieval capabilities, i.e., content-based image retrieval.
This project’s objective is to improve our understanding of image features that can be used to represent the massive and complex biomedical imaging data such as the dual-modal functional/anatomical images of the human body (PET/CT imaging). Our research will innovate in exploiting the physician’s mental model of human structural understanding, where diagnostic interpretation relies on knowing the location of body structures and its relationship to its neighboring structures, in order to construct an image feature-based representation of dual-modal biomedical images that incorporates both the functional and anatomical features.
Want to find out more?
The opportunity ID for this research opportunity is 316
Other opportunities with
Professor David Feng