0278-0062 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMI.2016.2646518, IEEE Transactions on Medical Imaging 1 Copyright (c) 2010 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending a request to pubs-permissions@ieee.org. Abstract—Diffuse Optical Tomography commonly neglects or assumes as insignificant the presence of optically clear regions in biological tissues, estimating their contribution as a small pertur- bation to light transport. The inaccuracy introduced by this prac- tice is examined in detail in the context of a complete, based on realistic geometry, virtual fluorescence Diffuse Optical Tomogra- phy experiment where a mouse head is imaged in the presence of cerebral spinal fluid. Despite the small thickness of such layer, we point out that an error is introduced when neglecting it from the model with possibly reduction in the accuracy of the reconstruc- tion and localization of the fluorescence distribution within the brain. The results obtained in the extensive study presented here suggest that fluorescence diffuse neuroimaging studies can be im- proved in terms of quantitative and qualitative reconstruction by accurately taking into account optically transparent regions espe- cially in the cases where the reconstruction is aided by the prior knowledge of the structural geometry of the specimen. Thus, this has only recently become an affordable choice, thanks to novel computation paradigms that allow to run Monte Carlo photon propagation on a simple graphic card, hence speeding up the pro- cess a thousand folds compared to CPU-based solutions. Index Terms— Biomedical imaging, Cerebral spinal fluid, Clear tissues, Diffuse optics tomography, Diffusion equation, Forward modelling, Fluorescence, Monte Carlo methods, Neuroimaging. I. INTRODUCTION ccounting for the effect of light scattering through biolog- ical tissue is the major challenge of non-invasive biomed- ical imaging at optical wavelengths [1,2]. Despite the fact that light scatters mainly in the forward direction, after a few transport mean free paths, i.e. a few millimeters of propagation, light becomes completely diffusive losing any information on the initial directionality. Many approaches are being currently developed to enable imaging in different scattering regimes de- fined as Microscopy, Mesoscopy and Macroscopy [2] depend- ing on the specimen or the functionality considered in the study. The general rule of thumb is that deeper imaging corresponds to lower resolution ability. Enhancement on this side has been feasible in part thanks to the increasing contribution of complex computational methods that deal with data acquisition and post processing, registration, light diffusion simulation and inverse This work was supported by the grants “Skin-DOCTor” and “Neureka!” im- plemented under the "ARISTEIA" and "Supporting Postdoctoral Researchers" Actions respectively, of the "OPERATIONAL PROGRAMME EDUCATION AND LIFELONG LEARNING", co-funded by the European Social Fund (ESF) and National Resources and from the EU Marie Curie Initial Training Network “OILTEBIA” PITN-GA-2012-317526. J. R. acknowledges support from the EC FP7 CIG grant HIGH-THROUGHPUT TOMO, and MINECO grant FIS2013-41802-R MESO-IMAGING. problem based reconstruction. In particular, optical tomo- graphic methods such as Optical Projection Tomography (OPT) [3], Optical Coherence Tomography (OCT) [4] as well as fluo- rescence-based methods like Selective Plane Illumination Mi- croscopy (SPIM) [5] and Fluorescence Molecular Tomography (FMT) [6] or in the more general form fluorescence diffuse optical tomography (fDOT) used in small animal studies [7] of- fer the capabilities to reconstruct quantitatively three dimen- sional models with resolutions down to the sub-cellular level. In this context, the development of novel computational tech- niques, such as automatic image segmentation and registration [8,9], image reconstruction [10,11], phase retrieval [12,13], computational light diffusion models [14,15], and the creation of accurate virtual biological phantoms with specific optical properties are playing a crucial role in terms of quantitative and accurate reconstruction of the measured specimen. The devel- opment of the biomedical imaging field is therefore tightly bound to the increasing complexity of the computer architec- tures, which nowadays enables fast parallel computation, ap- proaching quasi real time data processing for most of the appli- cations. One of the most highly time consuming methods, the Monte Carlo Photon Propagation (MC-PP), can be trivially [16,17] parallelized increasing the speed up to thousand fold by implementation in modern GPU programming paradigms, such as CUDA and OpenCL. This allows simulations of light diffu- sion in a feasible time scale even in low cost desktop personal computer solutions. Nowadays, the most common simulation involving models with complex realistic geometries is per- formed by solving the Diffusive Equation (DE) to obtain the photon flux distribution within the tissue, which represents an approximate solution of the Radiative Transfer Equation (RTE) [18]. The DE however, is accurate only in the diffusive regime, when scattering predominates absorption. Consequently, opti- cally clear layers embedded in scattering tissues constitute a challenge for modelling the photon propagation and therefore they are either commonly neglected [19], treated using radiosity theory [20-22], included with boundary conditions [23] or used in mixed DE-MC interfaces [24] and coupled RTE-DE [25] ap- proaches, introducing further approximations or limiting their application to planar slab or simple geometries. D. A. is with the Material Science Department of the University of Crete, Heraklion, Greece (daniele@iesl.forth.gr). J. R. is with the Department of Bio- engineering and Aerospace Engineering, Universidad Carlos III de Madrid, Spain and with the Experimental Medicine and Surgery Unit, Instituto de In- vestigación Sanitaria del Hospital Gregorio Marañón, Madrid, Spain. (jorge.ripoll@uc3m.es). D.A., A. Z. and G. Z. are with the Institute of Elec- tronic Structure & Laser, Foundation for Research and Technology - Hellas, Heraklion, Crete, Greece (azacharo@iesl.forth.gr, zahari@iesl.forth.gr). Fluorescence Diffusion in the presence of Optically Clear Tissues in a Mouse Head model Daniele Ancora, Athanasios Zacharopoulos, Jorge Ripoll, and Giannis Zacharakis A