This proposal concerns the development and evaluation of ~space-time~ image-processing methods for dynamic nuclear-medicine studies, the goal being to improve image quality and quantification by exploiting important temporal information in dynamic image sequences. This goal will be pursued by reconstructing and restoring dynamic imaging data using methods that make use of the time dimension in addition to the usual two, or three, spatial dimensions. Our preliminary results suggest that substantial improvements in quantification can be obtained if temporal correlations in the data are used to advantage, and that such improvements may be otherwise unachievable. The proposed research will focus on dynamic receptor-imaging studies based on positron emission tomography (PET), which have important implications for brain disorders, aging, breast and ovarian cancer, and radiopharmaceutical development. Our primary objective has two components: 1) the attainment of high quantitative accuracy and good spatial resolution in dynamic PET studies by application of space-time image-reconstruction methods; and 2) the development of techniques which, by virtue of their modest computational requirements, have immediate practical relevance. The larger objective of the research is to develop a set of tools for image-sequence processing, which we plan to apply outside the scope of the proposed project to other imaging applications in nuclear medicine and to other imaging modalities. We expect our results to have significance for gated cardiac studies and other clinical applications in which image sequences are obtained. In addition, the results will be relevant to multi-modality, multi-isotope, and multi-spectral imaging situations which, like dynamic imaging, involve multiple images or data channels for the same object.
The specific aims of the proposed project are: (1) To investigate and evaluation several fundamental algorithm- design issues pertaining to the reconstruction of dynamic image sequences; (2) To determine whether the very substantial savings in computation time offered by some of the proposed approaches can be obtained without significant sacrifices in achievable image quality; and (3) To apply the proposed methods to PET receptor-imaging studies and evaluate them quantitatively using computer simulations, actual PET phantom studies, and PET data obtined from other investigators in studies of human and animal subjects.
| Lukic, Ana S; Wernick, Miles N; Yang, Yongyi et al. (2007) Effect of spatial alignment transformations in PCA and ICA of functional neuroimages. IEEE Trans Med Imaging 26:1058-68 |
| Lukic, Ana S; Wernick, Miles N; Tzikas, Dimitris G et al. (2007) Bayesian kernel methods for analysis of functional neuroimages. IEEE Trans Med Imaging 26:1613-24 |
| Lukic, Ana S; Wernick, Miles N; Strother, Stephen C (2002) An evaluation of methods for detecting brain activations from functional neuroimages. Artif Intell Med 25:69-88 |
| Narayanan, M V; King, M A; Wernick, M N et al. (2000) Improved image quality and computation reduction in 4-D reconstruction of cardiac-gated SPECT images. IEEE Trans Med Imaging 19:423-33 |
| Wernick, M N; Infusino, E J; Milosevic, M (1999) Fast spatio-temporal image reconstruction for dynamic PET. IEEE Trans Med Imaging 18:185-95 |
| Kao, C M; Yap, J T; Mukherjee, J et al. (1997) Image reconstruction for dynamic PET based on low-order approximation and restoration of the sinogram. IEEE Trans Med Imaging 16:738-49 |
