Maximum Likelihood FRET Imaging

描述(申请人提供)：荧光共振能量转移(FRET)成像是一种使用染料对的显微镜技术。它检测蛋白质的结合，进而被用来推断细胞周期中的许多活动，如染色体动力学。当一种染料(供体)放弃一种能量状态时，它不是将能量转化为光子，而是将能量转移到附近的受体染料分子。然后，受体分子发射一个光子，被相机探测到。当这种情况发生时，供体和受体必须在彼此的纳米范围内，并且很可能结合到相同的蛋白质上。通过这种方式，可以监控蛋白质的结合情况。这项研究的长期目标是提供一种有利可图的商业软件产品，使FRET成像简单、廉价和准确。FRET需要一种图像处理算法来查看生物样本中发生FRET活动的区域，因此也就是化学键。今天广泛使用的算法的数学近似往往不准确，因此会引起可靠性问题。这些数学近似背后的主要物理原因是荧光染料对在其激发和发射光谱中存在严重的串扰和渗漏。矛盾的是，这种串扰和流血是FREAT发生所必需的，因此无法避免。从数学上讲，未知数比联立方程多，因此存在一个根本无法解决的数学问题。本文提出的软件克服了这些数学问题。它基于最大似然估计，消除了对联立方程的需要，而是提供了基于最大似然准则的解。本研究的具体目的是：(1)开发一个原型软件算法。(2)展示提供准确图像的可行性。这一目标将通过处理生物数据集和来自已知FRET指数的样本的测试数据来实现。

DESCRIPTION (provided by applicant): Fluorescence resonance energy transfer (FRET) imaging is a microscope technique that uses a dye pair. It detects bonding of proteins, and in turn, is used to deduce many activities such as chromosome dynamics through the cell cycle. When one dye (the donor), gives up an energy state, instead of converting the energy into a photon, it transfers the energy to a nearby acceptor dye molecule. The acceptor molecule then emits a photon which is detected by the camera. When this happens, the donor and acceptor must be within nanometers of one another and are probably bound to the same protein. This way, protein bondings are monitored. The long-term objective of this research is to provide a profitable commercial software product that makes FRET imaging straightforward, inexpensive and accurate. FRET requires an image processing algorithm to see the regions in a biological sample where the FRET activity, and thus the chemical bonds, are happening. The algorithms that are used widely today have mathematical approximations that are often not accurate and therefore cause reliability questions. The main physical cause that underlies these mathematical approximations is that the fluorescent dye pairs have severe cross-talk and bleed-through among their excitation and emission spectra. Paradoxically, this cross-talk and bleed-through are necessary for FRET to occur, so it is impossible to avoid them. Mathematically, there are more unknowns than simultaneous equations, and therefore there is a fundamentally unsolvable mathematical problem. The software proposed herein overcomes these mathematical problems. It is based upon Maximum Likelihood Estimation, which eliminates the need for simultaneous equations, and instead provides a solution based upon the maximum-likelihood criterion. The specific aims of this research are: (1) Develop a prototype software algorithm. (2) Show feasibility of providing accurate images. This aim will be accomplished by processing biological data sets and test data from a specimen of known FRET indices.