Ultrasound Guided Fluorescence Tomography
ABSTRACT
In this study, a hybrid-model imaging system combining fluorescence and ultrasound (US) was investigated with the motivation of providing structural priors towards improvement of fluorescence reconstruction. A single element transducer was scanned over the sample for anatomy. In the fluorescence part, a laser source was scanned over the sample with the emission received by an EMCCD camera. Synchronization was achieved by a pair of motorized linear stages. Structural information was derived from the US images and a profilometry and used to constrain reconstruction. In the reconstruction, we employed a GPU-based Monte Carlo simulation for forward modeling and a pattern-based method to take advantage of the huge dataset for the inverse problem. Performance of this system was validated with two phantoms with fluorophore inclusions. The results indicated that the fluorophore distribution could be accurately reconstructed. And the system has a potential for the future in-vivo study.
1. INTRODUCTION
Hybrid-model imaging methods have been widely studied with the goal of exploiting molecular information. So far, investigations have been conducted to incorporate different anatomical imaging modalities with fluorescence imaging. Several studies combined X-ray CT with fluorescence to better quantify and localize fluorophore distribution. For example, Fang et al combined X-ray imaging with diffuse optical tomography (DOT) for diagnosis of human breast cancer [1]. In a recent study, A. Ale et al combined X-ray CT with FMT and conducted experimental studies with different mice models, which demonstrated that the dual-modality system would be a potent tool for small animal imaging studies [2]. MRI has also been investigated to guide fluorescence functional imaging in both human and small animal scenarios [3–5]. Finally, ultrasound imaging, with the advantages of low cost and non-invasiveness, has been employed as a complement to fluorescence imaging. For example, C. Snyder et al employed US imaging to assess tumor size in mice to provide guidance for fluorescence imaging [6]. Also, Zhu et al used 2D US structural prior for a better fluorescence reconstruction in terms of localization and quantification [7].
In a previous study, we combined 3D US imaging with PMT-based fluorescence tomography to explore both anatomical and functional images, and compared results with previously reported US-Fluorescence systems [8]. Herein, we describe an upgraded system with respect to sampling precision and reconstruction technique. We evaluated this system using two phantoms with different geometries of inserted inclusions in the form of fluorescent tubes. Each tube was filled with Cy5.5 fluorophore at different concentrations. In the fluorescence imaging subsystem, an EMCCD camera was used to image sample from its top side with raster scanned illumination from the opposite side controlled by a pair of motorized linear stages. Acoustic imaging scanning was achieved using the same stages with micrometer step-size to recover accurate structural images. For fluorescence reconstruction, the forward model was simulated by the GPU-based Monte Carlo algorithm [9], and the fluorescence was reconstructed by the Levenberg-Marquardt (LM) minimization method with a regularization term encoding the structural information obtained by US and constraining the inverse problem [10]. Comparing to our previous system, the system described in this paper showed improvements in both optical recording and acoustic sampling. When associated with the GPU-based Monte Carlo reconstruction, this imaging system is expected to result in an improved quantification and localization of fluorophore distribution in-vivo supported by phantom data. Therefore, this multimodal imaging has the promise to contribute in animal study to explore both anatomical and functional information.
Для продолжения чтения вы можете скачать полную версию материала по ссылке ниже
