A new method is described for obtaining a 3-D reconstruction of a bioluminescent light source distribution inside a living animal subject, from multispectral images of the surface light emission acquired on charge-coupled device (CCD) camera. The method uses the 3-D surface topography of the animal, which is obtained from a structured light illumination technique. The forward model of photon transport is based on the diffusion approximation in homogeneous tissue with a local planar boundary approximation for each mesh element, allowing rapid calculation of the forward Green's function kernel. Absorption and scattering properties of tissue are measured a priori as input to the algorithm. By using multispectral images, 3-D reconstructions of luminescent sources can be derived from images acquired from only a single view. As a demonstration, the reconstruction technique is applied to determine the location and brightness of a source embedded in a homogeneous phantom subject in the shape of a mouse. The technique is then evaluated with real mouse models in which calibrated sources are implanted at known locations within living tissue. Finally, reconstructions are demonstrated in a PC3M-luc (prostate tumor line) metastatic tumor model in nude mice.
In vivo bioluminescence imaging depends on light emitted by luciferases in the body overcoming the effect of tissue attenuation. Understanding this relationship is essential for detection and quantification of signal. We have studied four codon optimized luciferases with different emission spectra, including enzymes from firefly (FLuc), click beetle (CBGr68, CBRed) and Renilla reniformins (hRLuc). At 25°C, the in vitro max of these reporters are 578, 543, 615, and 480 nm, respectively; at body temperature, 37°C, the brightness increases and the firefly enzyme demonstrates a 34-nm spectral red shift. Spectral shifts and attenuation due to tissue effects were evaluated using a series of 20-nm bandpass filters and a cooled charge-coupled device (CCD) camera. Attenuation increased and the spectra of emitted light was red shifted for signals originating from deeper within the body relative to superficial origins. The tissue attenuation of signals from CBGr68 and hRLuc was greater than from those of Fluc and CBRed. To further probe tissue effects, broad spectral emitters were created through gene fusions between CBGr68 and CBRed. These resulted in enzymes with broader emission spectra, featuring two peaks whose intensities are differentially affected by temperature and tissue depth. These spectral measurement data allow for improved understanding of how these reporters can be
A spectral imaging technique applied to in vivo bioluminescent imaging is presented that provides an estimate of the depth of bioluminescent reporters inside living animals. The model, based on the standard diffusion approximation of light propagating in a slab sample, is described in this paper. Validation experiments performed on phantom and tissue models, as well as preliminary <i>in vivo</i> mouse images, demonstrate the ability of spectral imaging to provide a correct estimate of depth based upon a single view imaging system.
Two different methods are presented and discussed for the evaluation of the point spread function (PSF) in multicore fibers (MCFs) used as an image guide in microendoscopy. In the first method, the coupling intensities between the fiber cores in a MCF are measured at three different wavelengths (632, 520, and 488 nm) by scanning the
input face of the MCF with an illuminated pinhole. In the second method, the intensity coupled into a fiber core from all the neighboring fiber cores is measured by obturating the fiber core and measuring the output light intensity. The nonnegligible contribution of leaky modes in the large intercore distance coupling is demonstrated. The two methods are fairly well correlated. The investigations have been carried out for two MCFs presenting well-differentiated characteristics. The major coupling mechanisms between fiber cores have been identified and their roles quantified as functions of the coupling distance. At short distance, crosstalk dominates, with a maximum coupling intensity of 15% at 632 nm for the
nearest neighbor fiber core. At larger distances (ranging from 3.8 to 5 intercore distances), leaky modes play a predominant role in the coupling, which remains nonnegligible.
Among the new possibilities offered by an endoholographic method, the vision in turbid media could facilitate the surgical interventions, in particular in vessels, where blood masks the vascular wall. The holographic process allows us to select the coherent light, used to form the image of an object embedded in a turbid medium. An in situ holographic technique based on the use of a flexible miniaturized endoscope (diameter less than 1 mm) coupled to a CCD camera, to record the hologram, was developed for medical applications. The hologram is formed, by reflection, on the tip of a multicore optical fiber (MCF), sampled, and then treated electronically. The image is reconstructed numerically, providing more flexibility to the holographic process. We present here the first experimental results obtained with this imaging system, tested in vitro with conditions matching the typical situations encountered in endoscopy. The possibility of extracting an image out of the ambient noise, produced by the diffusers present in the turbid medium, is described and analyzed.
An in-situ holographic technique, involving the use of a flexible miniaturized endoscope (diameter less than 1 mm) coupled to a CCD camera, to record the hologram, has been developed for medical applications and more particularly in-vivo biopsy. The hologram is formed, by reflection, on the tip of a multimode, multicore fiber (MCF), sampled, and then treated electronically. The image is reconstructed numerically, providing more flexibility to the holographic process. Reconstructed images show the capability of the microendoscopic system to restore 3D informations of the observed scene. Our predictions and experimental results have shown that the hope to achieve tissue observations at the cellular level is realistic. Furthermore, the different sources of noise of the experimental device were analyzed and their influence on the quality of the reconstructed image quantified. Images of simple cell models such as epithelial cells easily taken in the oral cavity, have been taken and analyzed. The possibility of using the microholographic technique for in-vivo biopsy is discussed both from theoretical considerations and experimental observations.
The quality of microendoscopic images ((phi) < 0.5 mm) results from numerous factors. The microlens, the optical interfaces, and the multicore fiber (MCF) play a role. The intercomparison of the performances of these various components requires a precise evaluation of their modulation transfer function (MTF). The MTFs of a representative sample of MCFs have been evaluated in detail and the results presented in previous communications. The sampling density, the cross-talk, and to a lesser extent the leaky modes and propagation in the cladding have been found to be the main factors worsening the MTF. The Selfoc Grin lens also limits, in a non-negligible proportion, the overall quality of the microendoscopic image. Comparative data have been obtained with a slightly modified evaluation procedure for the MTF. This synthetic analysis results in a global understanding of the factors limiting the performances of microendoscopes.
The question of the image quality evaluation in microendoscopy ((phi) < 0.5 mm) and of its improvements is addressed in this article. The MTF has been carefully evaluated by a precise and reproducible method and reveals significant differences between the various multicore fibers tested (twelve different fibers from three firms). The measuring conditions, such as the aperture of the incident light beam, the orientation of the sampling direction, and the position of the evaluation point on the bundle section, were also of concern in the evaluation of the image quality. The measurement of the overall contrast loss is related not only to the limit of sampling by the density of fiber cores, but also to the intercore fiber coupling.
An in situ holographic technique, involving the use of a flexible miniaturized endoscope (diameter less than 1 mm) coupled to a CCD camera, to record the hologram, has been developed for medical applications. The hologram is formed, by reflection, on the tip of a multimode fiber bundle (MMB), sampled, and then treated electronically. The image is reconstructed numerically, providing more flexibility to the holographic process. Reconstructed images show the capability of the microendoscopic system to restore 3D informations of the observed scene. The limitations of the holographic approach with the microendoscope have been evaluated and discussed in terms of the resolution limit. In particular, the low frequency sampling of the hologram through the MMB is not a limiting factor for the range of observation distance investigated (4 - 10 mm). A good accordance between the experimental results and the theoretical predictions was found by comparing the cut-off frequency obtained. Our results show that, for the considered observation distances, objects of a few micrometers can be clearly identified. The different sources of noise are analyzed and their influence on the quality of the reconstructed image have been quantified.
The quality of an image transmitted by a multicore, multimode fiber bundle has been investigated. The influence of the limiting factors such as the spatial sampling of the bundle, the cross-talk between each core and the radiative modes in the intercore space has been determined. Two different methods to determine the MTF are compared. The analysis of the data obtained experimentally for the coupling coefficients between fiber cores suggest that two different mechanisms must be invoked to explain the degradation of the image transmitted through the fiber bundle: the cross-talk and the leaky modes.
An `in situ' holographic technique, based on the use of a flexible miniaturized endoscope (diameter less than 1 mm) has been developed for medical applications. The holographic process provides quantitative information on the 3-D geometry of the observed scene, including deformations and movements. The holographic system presented here is based on the use of a multimode fiber bundle (MMB) as a recording medium for the endoscopic holograms. The hologram is obtained by the interference between the light reflected by the object and a reference beam travelling back along the axis of the MMB. The interferogram, sampled on the MMB tip, is then treated electronically. The image is reconstructed numerically.
A new holographic technique based on a numerical reconstruction method is presented and applied to endoscopic holograms. The hologram is obtained by the interference between the light reflected by the object and a reference beam taken as a plane wave travelling back along the axis of the multimode fiber bundle (MMB). The reconstruction of the image is computed numerically, using a fast algorithm to perform the Fresnel transform of the illuminated hologram. The quality of the reconstructed image has been evaluated by simulation and the limitations of the microendoscopic process enlightened: reduced aperture, sampling through the MMB. The resolution limit reachable by a miniaturized endoscope has been predicted by deriving the pseudo 3-D amplitude modulation transfer function (AMTF) of the system, and the noise originating from the twin image. Both AMTF (contrast included) and noise allow the calculation of the SNR (signal to noise ratio) of the reconstructed object. The size of the smallest object observable from the hologram can be established. This size is taken as an image quality index (IQI). Our simulations have shown that this IQI is good enough to identify small size objects (100 micrometers or less) from a small aperture (0.31 mm) endoscopic hologram. The sampling of the on-axis hologram on the tip of the MMB has been shown not to significantly degrade the image.