Video and audio processing in paediatrics: a review.

Video and sound acquisition and processing technologies have known great improvements in the last decades, with many applications in the biomedical area. The motivation of this paper is to provide an overall state of the art of the advances within these topics in paediatrics, to evaluate their potential application for monitoring in Neonatal Intensive Care Unit (NICU). For this purpose, more than 150 papers dealing with video and audio processing were reviewed. For both topics, clinical applications are described according to the considered cohorts, either full-term newborns, infants and toddlers, or preterm newborns. Then, processing methods are presented, in terms of data acquisition, feature extraction and characterization. The paper is firstly focused on the exploitation of video recordings, that began to be automatically processed in the 2000s, and we show that it has been mainly used to characterize infant motion. Others applications, including respiration and heart rate estimation and face analysis, are also presented. Audio processing is then reviewed with a focus on cry analyses. We describe that this topic arose earlier, from first studies focused on induced-pain cries to newest ones dealing with spontaneous cries, and that they are mainly based on frequency features. Then, some papers dealing with non-cry signals are also discussed. Finally, we show that, even if recent improvements in digital video and signal processing allow for an increasing automation of processing, the context of NICU makes still difficult a fully-automated analysis of long recordings. Propositions to overcome some limitations are given.