Diffuse Reflectance Spectroscopy (DRS) is a useful tool to study tissues of the body. It is also biologically non-invasive and affordable. DRS in specific configurations helps in deriving information regarding the structural and biochemical information of the target tissue, which are subjected to change during the progression of specific diseases.
A typical DRS instrument consists of a light source, spectrometer, specially configured fibre optic probe for transmitting, and collecting the light. Although the DRS instrument is simple enough, extraction of optical properties for further estimation of tissue constituents from the collected light is challenging. Availability of a DRS dataset in the form of a look-up-table (LUT) can help extract these parameters.
Prof. Sujatha Narayanan Unni 
Mr. Vysakh Vasudevan
In this study, conducted by Prof. Sujatha Narayanan Unni and Mr. Vysakh Vasudevan, from the Department of Applied Mechanics, Indian Institute of Technology Madras, Chennai, India, an in-silico model of the human skin is made and spatially Resolved Diffused Reflectance (SRDR) is derived using finite element analysis as compared to typically used and computationally intensive Monte Carlo Simulations (MCS). Flexibility in modelling, reduced computational time and comparable accuracy made FEM more favourable as compared to Monte Carlo method, which is important in the creation of large spectral database to assist the extraction of tissue parameters. Diffusion approximation of the Radiative Transport Equation (RTE) is used in the light tissue interaction simulations.
This study predicts the parameters of skin tissue such as depth and melanin concentration of epidermis and blood and oxygen content in the dermis to exemplify an application of the above-mentioned approach. The replicated tissue model studies and results were on par with the original tissues. A look-up table was created using the finite element based light tissue interaction simulations with an optimized number of source-detector separations and was used to extract tissue parameters from a turbid tissue medium for the first time to the knowledge of the authors of this paper. The FEM-based models were a success and could be an alternative for the traditional Monte Carlo methods. Such FEM-based models coupled with effective machine learning algorithms could lead to automated disease diagnosis during the clinical implementation of non-invasive optical biopsy devices.
Prof. Asima Pradhan from the Department of Physics, Indian Institute of Technology Kanpur, Kanpur, India, stresses the importance of this work with the following comments: “Spatially resolved diffuse reflectance is a well-established technique for extraction of optical properties from human tissue, amenable for clinical diagnostics, especially for superficial lesions. The method is generally based on analytical solutions to diffusion approximated radiative transport equation, which unfortunately are prone to errors. Monte Carlo simulations are highly accurate yet computationally intensive for spatially resolved reflectance. FEM based techniques have been seen to be effective and accurate in modelling light propagation in tissue with a potential for yielding local features. In this study, the authors have developed a FEM based bilayered skin tissue-mimicking model to generate a look up table for extraction of optical parameters. This is a detailed research on skin tissue with in homegeneities included and displays a good agreement with expected optical parameters. The paper establishes that FEM is an effective tool for modelling light propagation in skin tissue and is the first step towards in vivo diagnostics with better accuracy (using FEM) and clinically amenable (with SRDRS) for detecting different skin ailments.”
Article by Akshay Anantharaman
Here is the original link to the paper:
https://onlinelibrary.wiley.com/doi/abs/10.1002/cnm.3546
