Didier Zugaj

Multi-spectral image analysis for skin pigmentation classification

In this paper, we compare two different approaches for semiautomatic detection of skin hyper-pigmentation on multi-spectral images. These two methods are support vector machine (SVM) and blind source separation. To apply SVM, a dimension reduction method adapted to data classification is proposed. It allows to improve the quality of SVM classification as well as to have reasonable computation time. For the blind source separation approach we show that, using independent component analysis, it is possible to extract a relevant cartography of skin pigmentation.

Spectral analysis and unsupervised SVM classification for skin hyper-pigmentation classification

Data reduction procedures and classification via support vector machines (SVMs) are often associated with multi or hy-perspectral image analysis. In this paper, we propose an automatic method with these two schemes in order to perform a classification of skin hyper-pigmentation on multi-spectral images. We propose a spectral analysis method to partition the spectrum as a tool for data reduction, implemented by projection pursuit. Once the data is reduced, an SVM is used to differentiate the pathological from the healthy areas. As SVM is a supervised classification method, we propose a spatial criterion for spectral analysis in order to perform automatic learning.

Estimation of an optimal spectral band combination to evaluate skin disease treatment efficacy using multi-spectral images

Clinical evaluation of skin treatments consists of two steps. First, the degree of the disease is measured clinically on a group of patients by dermatologists. Then, a statistical test is used on obtained set of measures to determine the treatment efficacy. In this paper, a method is proposed to automatically measure the severity of skin hyperpigmentation. After a classification step, an objective function is designed in order to obtain an optimal linear combination of bands defining the severity criterion. Then a hypothesis test is deployed on this combination to quantify treatment efficacy.

Multi-scale analysis of skin hyper-pigmentation evolution

In this paper, we use statistical inference and muti-spectral images to quantify the evolution of skin hyper-pigmentation lesions under treatment. We show that statistical inference allows getting change maps of the disease which can be useful for dermatologists to analyze the disease evolution. Indeed, a local change map is obtained by computing the deviation between two multi-spectral images in a region of interest (ROI). Then, we normalize the obtained map and develop a statistical inference framework to quantify the changes. Finally, we propose a criterion that integrates change maps in order to quantify the treatment efficacy on a patient.

Evaluation of unmixing methods for the separation of Quantum Dot sources

Quantum Dots (QDs) are semiconductor crystals with nanometer dimensions, which have fluorescence properties that can be adjusted through controlling their diameter. Under ultraviolet light excitation, these nanocrystals re-emit photons in the visible spectrum, with a wavelength ranging from red to blue as their size diminishes. We created an experiment to evaluate unmixing methods for hyperspectral images. The wells of a matrix [3×3] were filled with individual or up to three of five QDs. The matrix was imaged by a hyperspectral system (Photon Etc., Montréal, QC, CA) and a data “cube” of 512 rows × 512 columns × 63 wavelengths was generated. For unmixing, we tested three approaches: Independent Component Analysis (ICA), Bayesian Positive Source Separation (BPSS) and our new Consensus Non-negative Matrix Factorization (CNFM) method. For each of these methods, we assessed the ability to separate the different sources from both spectral and spatial localization points of view. In this situation, we showed that BPSS and CNMF model estimates were very close to the original design of our experiment and were better than the ICA results. However, the time needed for the BPSS model to converge is substantially higher than CNMF. In addition, we show how the CNMF coefficients can be used to provide reasonable bounds for the number of sources, a key issue for unmixing methods, and allow for an effective segmentation of the spatial signal.

Brimonidine displays anti-inflammatory properties in the skin through the modulation of the vascular barrier function

Rosacea is a chronic inflammatory skin disease. Characteristic vascular changes in rosacea skin include enlarged, dilated vessels of the upper dermis and blood flow increase.