Distributed Intelligent Learning Environment for Mammographic Screening

Breast cancer is one of the major causes of death in the modern world. In the UK there is a national screening programme which women between ages of 50 and 70 can attend. Breast cancer screening involves taking breast X-Rays (called mammograms) and examining them for signs of cancer. The idea is that if cancers are detected and treated early (before there are noticeable symptoms) then treatments can be more effective. Examining mammograms for cancer is a highly skilled job carried out by trained radiologists who have to detect what are often very subtle abnormalities occurring only in a small proportion of the cases they examine. Our research will explore how computers can be effectively used to train radiologists to undertake the demanding task of breast screening. To do this we will develop and test an Intelligent Tutoring and e-Learning Environment (ITeLE) to provide instruction, support, practice and feedback for trainee radiologists intending to specialize in mammography.Although computer-based training tools have been developed for branches of radiology other than mammography, few have progressed to widespread and routine use. This is because the development of successful computer-based training systems presents us with a number of problems of different kinds, requiring different approaches and methods for their solution. To develop the ITeLE, we propose to use an interdisciplinary approach that draws upon and brings together insights from psychology, sociology and computer science, in the following ways:Cognitive psychology is concerned with how humans process information, and tells us how radiologists with different levels of skill approach the problem of interpreting medical images. Previous work has shown how, for example, novice radiologists are more likely to interpret images by applying rules that describe the difference between normal and abnormal features in an image. As they gain in experience, however, they come to rely more on matching features in the image in front of them with their memory of the many examples of similar features seen during their career. Psychology, then, gives us clues as to the sorts of training might be most appropriate as trainees' experience increases, which can be incorporated into an intelligent tutoring tool: initially tutorials to teach the rules of interpretation, followed by exercises giving trainees practice at distinguishing normal and abnormal presentations, progressing finally to simulating screening conditions where trainees would have to spot a variety of abnormal cases 'hidden' amongst normal ones.e-Learning environments need to be designed with an understanding of the work practices and expertise that they aim to support. We will draw upon the methods of sociology to understand the practical details of radiology training. By observing the work of training and the circumstances in which it takes place, and by involving trainees and mentors closely in the design and development of the ITeLE, we aim to produce an e-Learning environment that closely matches their needs, and which they find easy to understand and use. In this way, we can ensure that the tools we develop on the basis of our understandings of psychology are both useful and usable in practice.e-Learning environments make it possible to build a record of trainee decisions, including cases or features in the image they have struggled to identify correctly. Using methods from artificial intelligence (a branch of computer science), we intend to explore how this information can be used to automatically produce feedback (for example, indicating where a trainee's strengths and weaknesses lie) and advice (for example, concerning what tasks it would be appropriate for a trainee to tackle next). In the final stages of the project, we will undertake an evaluation of the ITeLE to demonstrate the effectiveness of the intelligent tutoring tool and of the different training strategies.