Terahertz biomedical imaging: From multivariate analysis and detection to material parameter extraction

Abstract

Terahertz imaging is an interesting route for biomedical analysis. In particular, cancer imaging is a subject of study for different teams [1,2]. A work is done in Bordeaux in partnership with a hospital to do terahertz analysis of breast tissue. This work is done in reflection with time domain imaging setup with fresh samples. The aim is to accurately assess tumor margins and which could in the future allow a quick validation of the precision of the surgical procedure and know if new surgery should be performed. We have presented in a previous paper [3] the use of automatic methods of image generation with different parameters [4] in order to explore the different contrasts that exist in the time and frequency domain data of a terahertz imaging system. These methods make it possible to locate and identify areas containing breast tissue, cancer or fat. In this communication, we propose to present new results and images with both multivariate approaches (like multivariate component analysis) and material parameter extraction to give both 2D localization and comprehensive parameter description.