The purpose of the project is to try to determine how artificial intelligence and automated image processing can assist radiologists in interpreting breast ultrasound images. Currently, radiologists mainly use qualitative descriptors to help classify the image, and this is adequate for many cases. However, there are always going to be borderline cases for which an accurate classification is difficult. Machine learning and automated feature extraction from the images have the potential to guide radiologists when assigning a probability of malignancy. Preliminary evidence from a small data set in our analysis suggests that this is possible, and we intend to apply these same techniques to a larger data set as soon as it is available. Ultimately, we aim to develop an interactive system that takes the best elements of both the radiologist and AI that will hopefully outperform either one of them individually.