Dispersed two-phase flows occur in a wide range of technical and scientific applications, such as compressors, fuel cells, clouds and droplet research. Such flows comprise bubbly flow, mist flow and sprays. The study of such flows is often carried out using optical measurements so as not to disturb the flow. The optical measurement of two phases in three dimensional space presents many challenges, such as distinguishing between the phases and locating the particles in three dimensional space. Further difficulties arise when optical access to the measurement volume is limited and the flow has to be reconstructed from a single camera. This project aims to improve the detection and three-dimensional localisation of particles, better discriminate between phases and improve the sizing of particles such as bubbles and droplets. Modern data science methods, including machine learning and optimisation methods for an inverse scattering problem, are used to develop improved measurement algorithms for existing measurement hardware. The method aims to improve measurement accuracy and allow for higher particle densities in the flows being studied. Further single optical access capability makes the approach more broadly applicable to different measurement scenarios.