Article ID Journal Published Year Pages File Type
558946 Biomedical Signal Processing and Control 2008 7 Pages PDF
Abstract

Malignant tumors will cause localized temperature increases (LTI) on breast surface which show as spot or vascular patterns in breast infrared thermograms. Thermographic detection of breast cancer now primarily depends on the visual analysis of these patterns by physicians, which is hard to provide objective and quantitative results. In this paper, we propose a new criterion of breast cancer detection and the method of its realization as follows: (1) surface temperature distribution of a healthy breast usually exhibits a gentle variation which is background. (2) Localized surface temperature of a carcinomatous breast will increase on the basis of the above background which is LTI. (3) The carcinomatous possibility is proportional to the LTI maximum (amplitude) of the suspicious focus region. (4) The LTI amplitude can be measured through morphological signal processing. According to the above conception, we obtained LTI amplitudes of 117 breast disease patients including 70 benign cases and 47 malignant cases. The optimal LTI amplitude threshold for breast cancer detection is determined as 1 °C from the ROC (receiver operating characteristic) curve based on the rule of Youden's index maximization. 44 cases were screened out from 47 breast cancer patients under this threshold. High sensitivity 93.6% and high NPV (negative predictive value) 91.2% demonstrate the value of this method on preliminary screening of breast cancer. High FPR (false-positive rate) 55.7% suggests that it is better to apply this method together with others.

Related Topics
Physical Sciences and Engineering Computer Science Signal Processing
Authors
, , , ,