Impact of noise on image reconstruction for diffuse optical tomography

Abstract

Diffuse optical tomography (DOT) uses the transmission of near-infrared light through tissue to image absorption and scattering. Especially for mammography applications DOT might become of clinical use. In this work we present simulation results on the influence of noise on the detectability of lesions for the Philips mammoscope system. Noise can have a significant impact on the image quality. It can prevent the detection of even high contrast lesions. The influence of noise can be reduced by a proper treatment in the reconstruction algorithm. But it changes also the effective sampling pattern of the imaging system if noisy data are not used (or used with a lower weight). In the case of this optical tomography system this means that lesions are only detectable up to a certain depth. This depth depends on where the signal power gets close to the constant noise floor, and on the volume and contrast of the lesion. Our simulation results show that the detection of lesions with 10 mm diameter and 100% absorption contrast should be possible with the mammoscope system, even in the worst case where the lesion is located in the center of the breast.