Machine-Learning on Liver Ultrasound to Stratify Multiple Diseases via Blood-Vessels and Perfusion Characteristics

Jules Bayet, Tim Hoogenboom, Rohini Sharma, Elsa D. Angelini

Abstract: Liver vessels can be visualized at sub-millimetre scale with contrast-enhanced ultrasound. In this work we exploit a cohort of 97 subjects (healthy volunteers and 4 liver disease types), exploiting multiple videos acquired at locations within the liver hand-picked by the sonographer to perform the diagnostic task. Annotation was performed at subject-level (disease subtype or healthy), along with scoring of image quality. We propose an original approach exploiting the abstraction capabilities of maximum intensity projections (MIPs) to feed a deep-learning classifier. Two architectures were tested for which we compared performance with different scenarios regarding the exploitation of transfer learning and the number of input MIPs per subjects. Our results show over 88% accuracy for a 2-class task (healthy versus disease), and 70% for a 3-class task (healthy versus 2 disease sub-types). We demonstrate, for the first time, that deep learning with minimal supervision and no pre-filtering can accurately classify liver diseases based on vascular ultrasound imaging acquired in a clinical setting. We also report findings on specific misclassication patterns which will guide further studies, augmentation of the cohort and subject annotation.