| 查看: 1740 | 回复: 0 | ||
[交流]
法国勃艮第大学CNRS Le2I实验室招一个图像处理方向的CSC博士生
|
|
法国勃艮第大学CNRS Le2I实验室招一个图像处理方向的CSC博士生 研究课题：Design and development of fast multispectral image processing for non-contact measurement of oxygen saturation 课题简介与工作任务： Description More and more diagnostic methods for health are based on non-conventional optical imaging. In the case of tissue study, they are usually based on the analysis of the light reflected by the tissue. This re-emitted light carries important information about the physical and optical tissue parameters. Such information can be acquired with multispectral or hyperspectral imaging-based systems. Multi/hyper-spectral images are composed of several spectral bands (up to hundreds) at different wavelength channels for the same tissue area. This technology combines advantages of both spectrophotometer (spectral resolution) and digital camera (spatial resolution). It can provide information that is not detectable by the human eye and can lead to objective quantification of some parameters. In this work, we will explore the possibility to measure oxygen saturation with multispectral cameras. Peripheral blood oxygen saturation (SpO2) is a measure of the relative concentration of oxygenated haemoglobin molecules in the arterial blood with respect to the total amount of haemoglobin. SpO2 is widely used in clinical practice, and is usually measured with pulse oximeters. This technique is based on photoplethysmography, and takes advantage of the fact that oxyhaemoglobin (HbO2) and deoxy haemoglobin (Hb) absorb light differently at different wavelengths and that arterial blood is mostly pulsatile in nature. This method is very efficient, widely-used and low cost but SpO2 probes still need to be worn – generally on fingers or clipped to earlobes – leading to poor compliance and risks associated with infection and skin irritation. We will investigate two strategies to measure SpO2 with multispectral cameras, i.e. light-tissue interaction model inversion and extension of contact regular pulse oximetry techniques to non-contact environment. * First, we will use light-tissue interaction models and the measured reflected light from the skin to estimate SpO2. The characteristics of the reflected light can be used with some models (such as Monte-Carlo simulations, Kubelka-Munk or the modified Beer-Lambert law) to predict some skin parameters including peripheral blood oxygen saturation. These methods are computationally very expensive and a rigorous numerical analysis of the problem will be critical. * Second, the extension of contact regular pulse oximetry techniques to non-contact environment is also non-trivial. The limitations of cameras will bring some very interesting challenges. We will need to manage uncontrolled environments with broadband light-sources with unknown intensity; low frame rates (～ 30 Hz); motion artefacts, surface reflections and shadowing. The PhD will be based at the Université de Bourgogne – Dijon (France) in the LE2I lab. The selected student will have the opportunity to be tightly integrated in a dynamic research team specialized in the development of innovative image processing algorithms and the implementation of hardware and software systems with high temporal constraints. 要求：英语流利，具备图像处理方向研究经验，复核CSC2016年公派留博士研究生申请标准 申请方式：将求职信、个人英文简历、硕士期间成绩单、文章著作（从事过的项目简介或者硕士论文手稿亦可）发送至 lichao8601@hotmail.com 或者 fanyang@u-bourgogne.fr |
回复此楼
