Classification of breast tissue density by optical transillumination spectroscopy: optical and physiological effects governing predictive value

Abstract

Preventive oncology is in need of a risk assessment technique that can identify individuals at high risk for breast cancer and that has the ability to monitor the efficacy of a risk reducing intervention. Optical transillumination spectroscopy (OTS) was shown to give information about breast tissue composition and tissue density. OTS is non-invasive, and in contrast to mammography, uses non-ionizing radiation. It is safe and can be used frequently on younger women, potentially permitting early risk detection and thus increasing the time available for risk reduction interventions to assert their influence. Before OTS can be used as a risk assessment and/or monitoring technique, its predictive ability needs to be demonstrated and maximized through the construction of mathematical models relating OTS and risk. To establish a correlation between OTS and mammographic density, Principle Components Analysis (PCA), using risk classification, is employed. PCA scores are presented in 3D cluster plots and a plane of differentiation that separates high and low tissue density is used to calculate the predictive value. Stratificaiton of PCA for measurement position on the breast in cranial-caudal projection is introduced. Analysis of PCA scores as a function of the volunteer's age and body mass index (BMI) is examined. A small but significant correlation between the component scores and age or BMI is noted but the correlation is dependnet on the tissue density category examined. Correction of the component scores for age and BMI is not recommended, since a priori knowledge of a women's tissue density is required. Stratification for the center and distal measurement positions provides a predictive value for OTS above 96%.