Localizing pulmonary nodules for surgical resection using ultrasound multiple scattering

Using conventional ultrasound to image pulmonary nodules is elusive due to multiple scattering in highly heterogeneous lung tissue. It is possible to leverage multiple scattering as a source of contrast between nodules and healthy lung parenchyma, because lung nodules do not contain air-filled alveoli. We developed a method based on the separation of multiple and single scattering using singular value decomposition. When combined with a depression detection algorithm, this allows us to render a map of the regions exhibiting less multiple scattering, associated with the presence of nodules. These techniques allowed for localization of 55 out of 59 nodules placed in ten lungs, as small as 5 mm diameter. In this study, we are demonstrating the feasibility of the method to locate nodules with respect to a surgical stapler by merging the rendered nodule map with a B-mode image of the surgical device. Using Verasonics Vantage scanner with array transducer, inter-element responses matrices were acquired from ex vivo pig/dog lungs, in which artificial nodules were implanted. We show that the relative position of nodule to the stapler can be measured. By measuring the distance of the nodule to the surgical stapler, it is possible to resect the nodule with safe margins.