4-D Reconstruction for Dynamic Fluorescence Diffuse Optical Tomography

Xin Liu,Bin Zhang,Jianwen Luo,Jing Bai

DOI: https://doi.org/10.1109/tmi.2012.2213828

IF: 10.6

2012-01-01

IEEE Transactions on Medical Imaging

Abstract:Dynamic fluorescence diffuse optical tomography (FDOT) is important for the research of drug delivery, medical diagnosis and treatment. Conventionally, dynamic tomographic images are reconstructed frame by frame, independently. This approach fails to account for the temporal correlations in measurement data. Ideally, the entire image sequence should be considered as a whole and a four-dimensional (4-D) reconstruction should be performed. However, the fully 4-D reconstruction is computationally intensive. In this paper, we propose a new 4-D reconstruction approach for dynamic FDOT, which is achieved by applying a temporal Karhunen-Loève (KL) transformation to the imaging equation. By taking advantage of the decorrelation and compression properties of the KL transformation, the complex 4-D optical reconstruction problem is greatly simplified. To evaluate the performance of the method, simulation, phantom, and in vivo experiments (N=7) are performed on a hybrid FDOT/x-ray computed tomography imaging system. The experimental results indicate that the reconstruction images obtained by the KL method provide good reconstruction quality. Additionally, by discarding high-order KL components, the computation time involved with fully 4-D reconstruction can be greatly reduced in contrast to the conventional frame-by-frame reconstruction.