UCL Discovery

Abstract

A time-resolved optical tomography system has been used to generate cross-sectional images of the human breast. Images are reconstructed using an iterative, nonlinear algorithm and measurements of mean photon flight time relative to those acquired on a homogeneous reference phantom. Thirty-eight studies have been performed on three healthy volunteers and 21 patients with a variety of breast lesions including cancer. We have successfully detected 17 out of 19 lesions, and shown that optical images of the healthy breast of the same volunteer display a heterogeneity which is repeatable over a period of months. However, results also indicate that the lack of accurate quantitation of optical parameters and limited morphological information limits the ability to characterize different types of lesions and distinguish benign from malignant tissues. Drawbacks of our current methodology and plans for overcoming them are discussed.