Development of super resolution ultrasound techniques for feature extraction in prostate cancer detection

Abstract

This thesis investigates a super resolution ultrasound based approach to prostate cancer detection. It consists of two parts; the development of new super resolution ultrasound imaging (SRUI) methods, and their application to prostate cancer detection. SRUI is a new technique to map the dynamics and structure of the vascular bed below the wave diffraction limit by localising and tracking microbubbles in the bloodstream. In the first part of this thesis, two new SRUI methods are developed to address the current limitations associated with the particle tracking accuracy, and are validated using synthetic and in vivo ultrasound data. These high-quality results provide the basis for further analysis. In the second part of this thesis, two new methods have been applied for prostate cancer detection by identifying abnormal features in SRUI. The first is an imaging biomarker based scoring system, referred to as cancer detection by super resolution ultrasound, where each biomarker is scored, and cancer is identified via a higher score based on a group of biomarkers. The second is a machine learning classifier where biomarkers serve as inputs for automated cancer detection. Both methods have been applied to a cohort of patients with known cancer locations. The findings demonstrate the potential clinical viability of these methods.