Abstract
Thermography may enable early detection of a cancer tumour within a mammary gland at an early, treatable stage of the illness, but thermogram analysis methods must be developed to achieve this goal. This study analyses the feasibility of applying the Hurst exponent readings algorithm for evaluation of the high dimensionality fractals to reveal any possible difference between normal thermograms (NT) and malignant thermograms (MT). Thermograms were obtained using electronic contact thermography. Significant differences in the Hurst exponent readings for the MT and the NT were observed when comparing the following:
1. The right NT (H = 0.40 +/- 0.13) (mean +/- standard deviation) and the right MT (H = 0.36 +/- 0.10) (p = 0.037) (p - significance level).
2. The left NT (H = 0.40 +/- 0.14) and the left MT (H = 0.37 +/- 0.11) (p = 0.035).
Given these differences, we conclude that the Hurst exponent computed with the aforementioned algorithm may be used in thermogram processing for the early diagnosis of mammary gland cancer. (c) 2012 Elsevier Ltd. All rights reserved.