98 IEEE JOURNAL OF SELECTED TOPICS INQUANTUM ELECTRONICS, VOL. 14, NO. 1, JANUARY/FEBRUARY 2008 An Alternative Approach to Analyze Fluorescence Lifetime Images as a Base for a Tumor Early Diagnosis System Tomer Eruv, Moshe Ben-David, and Israel Gannot Abstract—Fluorescence lifetime imaging is a very promising imaging method for early detection of malignant tumors. It offers many advantages over conventional fluorescence methods, espe- cially because the acquired signal does not rely on the fluorophore concentration in the tissue. As in all imaging method, the goal is to determine the exact location of a malignant tumor. However, since we are dealing with optical imaging, the inverse problem, i.e., extracting the tumor location coordinates is not an easy task to fulfill. In this paper, we describe an alternative method of inter- preting the fluorescence lifetime image. The method extracts four features from each decay curve. We show that from these features one can extract the location of the tumor. The theoretical model is compared to the experimental results obtained from tissue-like phantoms. Index Terms—Fluorescence lifetime imaging, light propagation in tissue, lifetime based imaging, 3D reconstruction, Moshe. I. INTRODUCTION O PTICAL imaging methods have many advantages: they apply nonionizing radiation; the systems have relatively simple configuration; and the signals can be guided by optical fibers (for endoscopic procedures). They are relatively cheap and usually minimally invasive. Despite the aforementioned advantages, implementing it into a functional system is a great challenge due to the optical char- acteristics of tissue, i.e., absorption and scattering. The first pre- vents light from penetrating deep into the tissue, thus limiting the spectral region one uses to the near-IR where the absorption is relatively low. The latter causes light to deviate from its path to other directions. Both phenomena cause less light to reach the detector. Furthermore, it is difficult to trace the photon to its place of origin. Steady-state fluorescence and fluorescence lifetime imaging (FLI) are optical imaging methods, which use absorbed and emitted light at different wavelengths. Steady-state fluorescence measures the intensity of the emitted light, and consequently, Manuscript received October 5, 2007; revised November 5, 2007. T. Eruv and M. Ben-David are with the Department of Biomedical Engi- neering, Faculty of Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel (e-mail: erovtome@post.tau.ac.il; moshebd@walla.com). I. Gannot is with the Department of Biomedical Engineering, Faculty of Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel. He is also with the Program of Biomedical Engineering, Department of Electrical and Computer Engineering, School of Engineering and Applied Sciences, George Washington University, Washington, DC 20052 USA (e-mail: gannot@eng.tau.ac.il). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JSTQE.2007.913978 it is strongly dependent on concentration and excitation in- tensity of the fluorescent dye. Lifetime fluorescence measures the average time a molecule spends in the excited state (prior to emitting a photon). This is the molecule’s intrinsic char- acteristic and does not depend on the light intensity or dye concentration. Labeling a specific site is done by conjugating fluorescent dyes with antibodies that are specific to tissue surface markers (antigens), and then, rely on antigen–antibody binding process. The dyes might be regular fluorescent dyes or dyes that vary their intrinsic properties as a result of different environmental conditions (functional changes). FLI provides functional infor- mation such as pH, temperature changes (compared to normal), and other bioenergetic changes [1]. Even in the early stage of tumor progression, pH can drop to 5.6, in contrast to neutral pH in normal tissue [2]. The main challenge after labeling the inspected site (tumor) is the need to reconstruct the location of the tumor from the acquired signal. 2-D imaging is straightforward. However, de- termining the tumor depth is complicated due to tissue optical properties. Experimental and theoretical 3-D localization methods have been described by several researchers [3]–[6]. Most of the stud- ies are based on steady-state fluorescence, which depends on the fluorescence intensity that does not provide accurate results. In order to overcome the problems faced by steady-state fluo- rescence some of the researchers use optical tomography and base their calculation on the diffusion approximation of radi- ation transfer equation [7]. This approach requires a complex experimental setup, which comprises a source and a detector array. The image reconstruction requires very complicated in- verse algorithms, which is difficult to apply to every geometry. An alternative approach, which is based on random walk theory, was described by Gannot et al. [8]. The study was later extended to the time domain [9]. In this paper, we present a new approach to interpret the time- resolved fluorescense curve. While the traditional approach extracts only the lifetime parameter, we suggest looking at four parameters: first photon, first moment, maximum inten- sity, and lifetime. We will show that information about the 3-D localization of the site can be extracted from these parame- ters. The parameters were studied theoretically using a refined Monte Carlo (MC) simulation and experimentally on control- lable tissue-like phantoms. The new simulation deals with time- resolved fluorescence and simulate tissue with abnormalities (tumors). 1077-260X/$25.00 © 2008 IEEE