Tomographie en cohérence optique dans le domaine de Fourier appliquée aux tissus humains

Fourier Domain Optical Coherence Tomography FDOCT is a recent imaging technique based on the principle of Low Coherence Interferometry. In FDOCT cross-sectional images of diffusing objects are synthesised from a measure of backscattered light. The main advantages of FDOCT are high acquisition speed, extremely high sensitivity to discontinuities of the index of refraction and high axial resolution. Due to those advantages FDOCT became a standard in 3D invivo imaging of biological objects, in particular in ophthalmology. Apart from structural imaging of tissues, FDOCT is able to deliver information about functional parameters, as blood flow speed. In fact the Doppler effect allows for optical segmentation of moving particles and measurement of very high flow speed. FDOCT shows up to be the best choice for the investigation of the morphology and the physiology of human tissues. In the first part of this thesis we developed image processing tools needed for the analysis of data generated during 3D acquisitions. Those tools allow segmentation of different tissue layers, as well as quantification of parameters as mean layer thickness and index of refraction. We present evaluations of measurements on skin and retina. The second part deals with retinal physiology. We used the Doppler effect to enhance the contrast of moving particles, whereas static particles are suppressed. We measured the axial component of blood flow in the region of the optic nerve head. The main contribution of that second part remains in the elaboration of an algorithm allowing recovering the complete vectorial blood flow. We analyzed the retinal blood flow pulsation with and without flicker stimulation. A last investigation reveals the influence of flicker stimulation on the reflectivity of photo-receptor layers. In the third part we present a simple method which allows overcoming one major limitation of FDOCT without the need of additional devices. In fact the tomograms obtained by Fourier Transform of the intensity of the interferometric signal are symmetric, because the measured intensity is real. This limitation reduces the available axial distance to half. Our method is based on the insertion of a linear phase shift and the complex extension of the intensity containing this phase shift by Hilbert Transform. This complex signal provides tomograms without complex ambiguity and doubles the available axial distance. We integrated this method in a handheld device, developed during this thesis. We validated the theory by experimental measurements on static objects and we present invivo measurements of human tissues.