Digital Theses Archive


Tesi etd-03302020-233409

Type of thesis
E-mail address
Infant spontaneous activity as a window into early brain development: towards new kinematic and electrophysiological biomarkers
Scientific disciplinary sector
Istituto di Scienze della Vita
Istituto di Scienze della Vita - TRANSLATIONAL MEDICINE
Membro Dott.ssa PAVLIDIS, ELENA
  • computational EEG analysis
  • computerised motor analysis
  • early motor development
  • functional biomarker
  • infant kinematic
  • neonatal EEG
  • spontaneous motor activity
Exam session start date
Early recognition of infants that eventually develop lifelong neurodevelopmental disabilities remains an open challenge for the global healthcare. More than one infant in ten is considered to be at neurodevelopmental risk and addressing a timely participation in early therapeutic intervention is pivotal for an effective improvement of long term outcome. The challenge remains to efficiently target early intervention to those who would most benefit from it. The overarching aim of this project has been to explore novel frontiers in early detection of neurodevelopmental disabilities, by the use of kinematic and electrophysiological characterization of infant spontaneous activity.<br>We explored the feasibility of adapting novel approaches to routine recordings to extract measurable features of biological signals, focusing on the search of biomarkers of neurodevelopmental trajectories, in both typical development and in the follow up of patients at risk. <br><br>In this context the organization of spontaneous movements has been advisably considered the reference standard for early identification of infants at high risk for cerebral palsy. As part of a worldwide study about very early natural history of CP, we focused on early characteristic patterns of atypical trajectories and defined in the motor repertoire early target movements as specific markers for type and severity of CP prognosis. The attention towards the motor repertoire, such as antigravity movements and postural patterns, has brought new insights onto prediction of an infant’s later motor function and inspired also the characterization of typical trajectories of development, as possible reference charts of neurodevelopment. <br>Based on 3D motion tracking, we provided quantitative estimation of the typical progression of the early neuromotor development, that allowed automated estimation of age based on upper limbs movement patterns. To overcome the limits of classic kinematics, which need expensive and a fixed set up, we then developed a multi-sensor smart jumpsuit that allows mobile accelerometer and gyroscope data collection of independently moving infants, with high accuracy. By a machine learning-based classifier it was then possible to address automated classification of key postures and movement patterns, and to differentiate high and low performing infants by simply quantifying mere incidence of movement categories.<br>As a further step, we then tested automated image estimation algorithms on video recordings of infant spontaneous movements. OpenPose estimator allows to reliably track objects and even body parts with high accuracy from normal 2D-video, converting ordinary recordings into kinematic time-series. The system proved sensitive and accurate enough to detect and reproduce movement patterns for visual analysis but also offered the opportunity to translate movement trajectories into time-series, used as kinematic metrics. This system offers a revolutionary way to think at kinematic and it granted a promising way to develop easily accessible, non-invasive methods for screening infants with suspected compromised neurodevelopment. <br><br>Parallel to automated motor studies, we further used quantitative analysis of early brain electrical activity as a marker of suboptimal maturation after early environmental stressors, such as stress due to early nutritional supply in prematures and brain lesion at birth.<br>We also tested quantitative expressions of EEG signals, not only to address maturation as a global features but also to target atypical brain reorganization following early brain injuries, such as neonatal stroke. We developed an algorithm for automated detection of sleep oscillation that turned into a valuable biomarker for early prediction of cerebral palsy in infants with focal brain lesion. <br>We finally included in this work also contribution coming from new clinical implementations of classical neurophysiological techniques. We proved the importance of neonatal SEP assessment, in adjunction to EEG, in the early days after perinatal asphyxia for improving long term outcome prediction.<br><br>As a whole, our studies confirm that quantitative metrics of physiological infants maturation have now become accessible for the objective evaluation of infants milestones in typical but also in atypical development. Early quantitative functional biomarkers are highly likely to increase our early predictive power as to long-term outcome in those infants at risk, with enormous benefits for the children and their families.