This research program is designed to develop automated systems that can aid in the diagnostic interpretation of digital biological imagery. Specifically, we propose to design and implement software modules that analyze the normal and abnormal movement of human tissues imaged with real-time magnetic resonance imaging (MRI). This investigation will focus on the development of software that can automatically analyze complex biological structures in terms of deformable tissue models. As the domain for the initial Phase I study, we plan to analyze the example of the human tongue during the act of swallowing. This domain provides us with a deformable constant volume tissue with fixed attachment to readily identifiable anatomical structures. Because of the symmetries involved, this motion can be studied in two or three dimension and the cycle time for the motion is appropriate for ultra-fast MR scanning. This study is designed to define and implement a 2D model of the tongue that can be used to extract and track the motion of tongue tissue. This modeling technique will be validated against interactive human measurements of the shape of tongue tissue in test imagery. This model will be appropriate for extension to 3D and full motion tracking in Phase II.
The software techniques to be developed in this research program will be applicable to a wide range of tissue motion analyses. While all initial research will be performed on tongue data, the model-based paradigm is designed to support arbitrary deformable tissue motion studies.
