Reconstruction 3D de la forme d'aiguilles chirurgicales en utilisant la réflectométrie fréquentielle dans des fibres optiques

摘要

The main objective of the research project is to track the shape of minimally invasive surgical tools (mainly needles) by inserting optical fibers into them. By measuring the strain along the fibers, we can easily relate it to the curvature of the fibers. Using three fibers glued together in a triangular geometry, the difference in the measured curvature of each fiber allows one to orientate the curvature in a 3D frame. Since the approach for shape tracking is strictly based on the insertion of optical fibers inside the restricted space available in minimally invasive surgical tools, it can be used with many types of surgical tools such as catheter needles, colonoscopes, or any other remotely controlled instrument. The knowledge of the position of the device inside the human body is of paramount importance to maximise the success of the intervention.;Up to now, the most studied approach for shape tracking using optical fibers is based on fiber Bragg gratings (FBGs), which are useful devices to measure the strain in fibers. To the best of our knowledge, the best precision reached in the literature based on FBGs is ∼0.28mm, corresponding to the accuracy in the predicted needle tip position. To reach this precision, two sensors were used, each one containing a set of three fibres with 3 FBGs (one in each fiber) for a total of 6 FBGs. More studies have been made using similar devices, with more or less number of FBG sensors separated by different distances. Most of these studies achieve an accuracy in the order of few millimeters. However, this approach to measure strain along the fibers is completely discrete since the strain is only known at the positions where the FBGs are located. Approximations are thus necessary to extrapolate the strains to recover the whole shape of the needle.;This project suggests a truly distributed approach, different to the discrete FBGs technique, which has received little attention up to now for this type of applications. Our first hypothesis is that the precision of the shape tracking can be enhanced by using truly distributed strain sensing (instead of discrete sensing) since approximations are not needed to obtain the shape of the entire needle.;This approach is based on optical frequency domain reflectometry (OFDR), which is an interferometric method frequently used to measure the attenuation along fibers. Indeed, OFDR is based on Rayleigh scattering, which is caused by a random distribution of refractive index on a microscopic scale in the fiber core. A frequency swept signal is thus sent inside a fiber and by measuring the frequency of an interference signal, the complex amplitude of the backscattered signal as a function of the position along the fiber can be measured. By doing a Fourier transform of this signal, the spectrum of the reflected signal can be measured for small sections of the fiber (Deltax), which correspond to the spatial resolution of the strain sensing approach based on OFDR. When the fibers undergo strain, this spectrum shift to shorter or longer wavelengths, depending on whether the fiber is stressed or compressed. By measuring the spectral shift, the strain along the fiber can be quantified since these values are directly proportional, which leads to a truly distributed strain sensor.;Using this approach, shape tracking of surgical needles has been performed strictly in vitro. A geometric model based on the work of another research group has been used to transform the sensed strain to shape sensing. Using single mode fibers (SMF), a maximum accuracy of 0.9+/-0.3mm over the entire needle shape for a tip deflection of 6.35mm has been obtained, which is clearly enough for clinical applications. However, we note that in the present work, this accuracy decreases as the curvature of the needle increases.;Our second hypothesis is that using optical fibers with higher Rayleigh scattering coefficient can allow one to reach higher accuracy. To verify this hypothesis, the accuracy of such devices using three different types of fibers has been compared. These fibers are SMF, Germanium Boron doped photosensitive fibers (Redfern) and UV exposed SMF. By UV exposing SMF, it has been shown that Rayleigh scattering can be considerably enhanced. The maximal accuracy obtained has thus passed from 0.9+/-0.3mm for SMF to 0.6+/-0.2mm for UV exposed SMF, which is the fiber that showed the higher amount of Rayleigh scatter. This enhancement is the accuracy of the device is clearly our most valuable contribution to this field of research.