Software Tools for Breast Cancer Detection in Positron Emission Mammography Images

Positron Emission Mammography (PEM) is a novel imaging technology utilizing a dedicated Positron Emission Breast scanner. Two detectors in a configuration that resembles a conventional mammography (MG) machine cover a 250x120x200 mm3 volume, yielding 3D images of the breast, which is fixated between the detectors, but not compressed. The procedure depends on the administration of a radioactive contrast agent, 18F FDG, as used in common whole body PET scans. Clinical studies showed the equivalent sensitivity and superior specificity of PEM regardless of hormonal status, breast density, and other factors hampering MG and MRI. Yet, due to the novelty of the imaging modality, no software support besides basic viewing and measurement capabilities exists, and common PET software is not tailored to the specific tasks in PEM image analysis. We propose algorithms and prototypical implementations for the tasks of automatic background estimation, lesion detection, lesion volumetry, and lesion measurements.