Image processing and analysis in a dual-modality optoacoustic/ultrasonic system for breast cancer diagnosis

Coregistered optoacoustic (OA) and ultrasound (US) images obtained using a dual modality optoacoustic/ultrasonic breast imaging system are used together for enhanced diagnostic capabilities in comparison to each individual technology. Therefore, an operator-independent delineation of diagnostically relevant objects (in our case breast tumors) with subsequent automatic analysis of image features is required. We developed the following procedure: 1) Image filtering is implemented on a US image to minimize speckle noise and simultaneously preserve the sharpness of the boundaries of the extended objects; 2) Boundaries of the objects of interest are automatically delineated starting with an initial guess made by an operator; 3) Both US and OA images are analyzed using the detected boundaries (contrast, boundary sharpness, homogeneity of the objects and background, boundary morphology parameters are calculated). Calculated image characteristics can be used for statistically independent evaluation of structural information (US data) and vascularization (OA data) of the studied breast tissues. Operator-independent delineation of the objects of interest (e.g. tumors and blood vessels) is essential in clinical OA spectroscopy (using multiple laser wavelengths to quantify concentrations of particular tissue chromophores, such as oxy- and deoxy- hemoglobin, water, and lipids). Another potential application of the suggested image analysis algorithm could be in OA imaging system design, when system performance should be evaluated in terms of quality of the images reconstructed from the well-defined objects of interest. The discussed principles of image analysis are illustrated by using real clinical US and OA data.