Immersive analytics for the suggestion of opinions in the analysis of a medical report from a collaborative social network

Abstract

Doctors are mostly in need of several opinions in order to make a good decision. Indeed, medical social networks have become at place of exchange of experiences between doctors. In addition, supervised learning of medical record will help physicians quickly find the right concise explanation of their medical images. To do this, we present in this article a model of opinion suggestion based on a collaborative social radiological network. Indeed, the opinions shared on a medical image in a medico-social network are represented in the form of a textual description which in most cases requires a cleaning using a medical dictionary. In addition, we describe the textual description of medical image with TF-IDF weight vector using a " bag of words" approach. We use latent semantic analysis to establish relationships between textual terms and visual terms in shared opinions about the medical image. Multimodal modeling looks for medical information through multimodal queries. Our model is evaluated against the ImageCLEFmed baseline, which is the basic truth for our experiments. We have conducted many experiments with different descriptors and many combinations of modalities. The analysis of the results shows that the model based on two methods makes it possible to increase the performance of a search system based on a single modality, visual or textual.