Determining the type of kidney stones allows urologists to prescribe a treatment to avoid the recurrence of renal lithiasis. An automated in-vivo image-based classification method would be an important step towards an immediate identification of the kidney stone type required as a first phase of the diagnosis. In the literature, it was shown on ex-vivo data (i.e., in very controlled scene and image acquisition conditions) that an automated kidney stone classification is indeed feasible. This pilot study compares the kidney stone recognition performances of six shallow machine learning methods and three deep-learning architectures which were tested with in-vivo images of the four most frequent urinary calculus types acquired with an endoscope during standard ureteroscopies. This contribution details the construction of an in-vivo dataset of endoscopic images with four of the most recurrent classes in clinical practice. It also describes the design, implementation, and results of the classifiers (shallow machine learning and deep learning-based methods) of kidney stones. Even if the best results were obtained by the InceptionV3 architecture (weighted accuracy, precision, recall and F1-Score of $97\pm 03\%, 97\pm 03\%, 98\pm 04\%$ , and $98\pm 03\%$ , respectively), it is also shown that choosing an appropriate colour space and texture features allows a shallow machine learning method to approach closely the performances of the most promising deep-learning methods (the XGBoost classifier led to weighted accuracy, precision, recall and F1-Score values of $96\pm 14\%$ ).

In the context of cutaneous carcinoma diagnosis based on in vivo optical biopsy, Diffuse Reflectance (DR) spectra, acquired using a Spatially Resolved (SR) sensor configuration, can be analyzed to distinguish healthy from pathological tissues. The present contribution aims at studying the depth distribution of SR-DR-detected photons in skin from the perspective of analyzing how these photons contribute to acquired spectra carrying local physiological and morphological information. Simulations based on modified Cuda Monte Carlo Modeling of Light transport were performed on a five-layer human skin optical model with epidermal thickness, phototype and dermal blood content as variable parameters using (i) wavelength-resolved scattering and absorption properties and (ii) the geometrical configuration of a multi-optical fiber probe implemented on an SR-DR spectroscopic device currently used in clinics. Through histograms of the maximum probed depth and their exploitation, we provide numerical evidence linking the characteristic penetration depth of the detected photons to their wavelengths and four source–sensor distances, which made it possible to propose a decomposition of the DR signals related to skin layer contributions.

Video endoscopy is one of the standard clinical procedures for visually detecting lesions on the internal wall of human bladders. In order to facilitate the diagnosis, it is helpful to build panoramic maps by registering consecutive images from the video sequence. We show how to efficiently reduce the computation time of graph cut based image registration by an order of magnitude. The number of nodes in a graph is greatly reduced using spatial image properties in order to minimize the loss of information. The set of edges in this sparse graph is obtained by applying a watershed transform on the set of nodes. This graph reduction has negligible negative effects on image registration quality compared to a dense graph cut, so that visually coherent panoramic maps of bladder walls can be built. Results demonstrate that the method improves the registration accuracy and reduces the computation time of other endoscopic bladder image registration methods. This work is an important step towards real time map construction.