Optical discrimination between malignant and benign breast lesions

Time domain multi-wavelength (635 to 1060 nm) optical mammography was performed on 82 subjects with breast lesions (45 malignant and 38 benign lesions). A perturbative approach based on the high-order calculation of the pathlength of photons inside the lesion was applied to estimate differences between lesion and average healthy breast tissue in terms of: i) absorption properties, and ii) concentration of the major tissue constituents (oxy- and deoxy-hemoglobin, water, lipid and collagen). The absorption difference a between lesion and healthy tissue is significantly different for malignant vs. benign lesions at all wavelengths. Logistic regression fitted to the absorption data identifies 975 nm as the key wavelength to discriminate malignant from benign lesions. When the difference in tissue composition between lesion and healthy tissue is considered, malignant lesions are characterized by significantly higher collagen content than benign lesions. Also the best model for the discrimination of malignant lesions obtained applying regression logistic to tissue composition is based only on collagen.