Use of finite difference time domain simulations and Debye theory for modelling the terahertz reflection response of normal and tumour breast tissue.

Anthony J Fitzgerald,Emma Pickwell-Macpherson,Vincent P Wallace,Irene Georgakoudi

doi:10.1371/journal.pone.0099291

Anthony J Fitzgerald, Emma Pickwell-Macpherson + Show 2 more

Open Access

https://doi.org/10.1371/journal.pone.0099291

Copy DOI

Abstract

The aim of this work was to evaluate the capabilities of Debye theory combined with Finite Difference Time Domain (FDTD) methods to simulate the terahertz (THz) response of breast tissues. Being able to accurately model breast tissues in the THz regime would facilitate the understanding of image contrast parameters used in THz imaging of breast cancer. As a test case, the model was first validated using liquid water and simulated reflection pulses were compared to experimental measured pulses with very good agreement (p = 1.00). The responses of normal and cancerous breast tissues were simulated with Debye properties and the correlation with measured data was still high for tumour (p = 0.98) and less so for normal breast (p = 0.82). Sections of the time domain pulses showed clear differences that were also evident in the comparison of pulse parameter values. These deviations may arise from the presence of adipose and other inhomogeneities in the breast tissue that are not accounted for when using the Debye model. In conclusion, the study demonstrates the power of the model for simulating THz reflection imaging; however, for biological tissues extra Debye terms or a more detailed theory may be required to link THz image contrast to physiological composition and structural changes of breast tissue associated with differences between normal and tumour tissues.

Highlights

A number of papers have been published on the applications of terahertz (THz) technology for the detection and diagnosis of cancer [1,2,3] including skin cancer [4,5,6] cervical cancer [7], colon cancer [8,9] and breast cancer [10,11,12]
The inputs into the classification algorithms are based on a set of parameters that are either heuristic in nature, i.e. selected due to observed differences between pulses reflected from normal and diseased tissues, or through principal component analysis (PCA), i.e. statistically significant differences between the pulse profiles from different tissue pathologies
Liquid water has been well studied in the THz regime and the values from this work are compared with existing values in the literature

Summary

Introduction

Background A number of papers have been published on the applications of terahertz (THz) technology for the detection and diagnosis of cancer [1,2,3] including skin cancer [4,5,6] cervical cancer [7], colon cancer [8,9] and breast cancer [10,11,12]. The usefulness of the pulse profile characteristics in identifying tissue pathology is highlighted by the classification accuracy of 92% using PCA for breast cancer [15] and around 90% with colon cancer [16] The reason for this high discrimination ability is often attributed to differences in the tissue water content which provides contrast between normal tissue and tumours, this has not been fully explored. It has been suggested by others for example, Png et al [17] and Sy et al [18] that water is not the only source of contrast and other tissue pathology features may contribute, it is not clear how at this stage

Objectives

Methods

Results

Discussion

Conclusion