Detection of Breast Abnormality Using Rotational Thermography

Abstract

Breast cancer is considered to be one of the major causes for high mortality rates in young womenWomenin the developing countries. Survival rate in breast cancer patientsPatientsmay be improved significantly by early detectionEarly Detection. In order to detect cancer in its initial stages breast screening is recommended for women over 40 years of age. Due to the limitations of existing breast cancer screeningScreeningtechniques alternative modalities such as thermography are being explored. An elevation in localLocalsurface temperatureTemperaturedue to an underlying pathology is considered as one of the earliest indications of an underlying cancer. Such regionsRegionsare represented as hotspots on a conventional thermogram. Detection of these hotspots from conventional breast thermogramsThermogramsis quite challenging, mainly due to incomplete image acquisition. A novel technique called rotational thermographyRotational thermographyhas been developed to address this issue. In this chapter, a frame work has been presented for developing a breast cancer screening system using thermograms acquired with this new imagingImagingmodality. Image featuresFeatureare extracted from rotational thermograms in spatial, bispectral, and multi-resolution domains. Optimal features are identified using genetic algorithmGenetic algorithmand automatic classification is performed using support vector machineSupport Vector Machine (SVM). In addition to screening, attempt has been made to characterize a detected abnormality as benign or malignantMalignant. As rotational thermographyThermographyacquires images of the breast in multiple views, study is carried out to locate the position of the tumor in correlation with ultrasound and biopsy findings. Thus the potential of the system for screening, characterization, and localization of breast abnormalities is explored.