Markerless Motion Analysis for Early Detection of Infantile Movement Disorders

The analysis of spontaneous movements provides valuable information for diagnosing infantile movement disorders. However, analysis is time-consuming and interpretation requires well-trained experts. We present an automated system that captures 3D joint po

  • PDF / 887,198 Bytes
  • 4 Pages / 595.276 x 790.866 pts Page_size
  • 92 Downloads / 238 Views

DOWNLOAD

REPORT


1 Fraunhofer Institute of Optronics, System Technologies and Image Exploitation IOSB, Gutleuthausstr. 1, 76275 Ettlingen, Germany Department of Paediatric Neurology and Developmental Medicine, Dr. von Hauner Children’s Hospital, Ludwig-Maximilians-Universit¨at (LMU), Munich, Germany 3 Section for Neuroelectronic Systems, Neurosurgery, Medical Center - University of Freiburg, Germany Faculty of Medicine, University of Freiburg, Germany Freiburg Institute for Advanced Studies (FRIAS), University of Freiburg, Germany

Abstract— The analysis of spontaneous movements provides valuable information for diagnosing infantile movement disorders. However, analysis is time-consuming and interpretation requires well-trained experts. We present an automated system that captures 3D joint positions and head rotation of infants without attached markers or sensors. We introduce motion parameters of head, trunk, upper and lower limbs of both body sides that are related to range, variability, and symmetry of motions and offer objective diagnostic information for assessment of motor behavior. We analyze 6 recordings of 5 infants who are at high-risk of impaired motor development, and show how the system highlights movement characteristics that hint at disorders.

Fig. 1: System setup. A depth camera captures an infant lying on an examination table. The system running on a connected laptop estimates 3D joint positions and head rotation and provides diagnostic information for the assessment of motor development.

Keywords— Motion analysis, infants, diagnostics

I.

I NTRODUCTION

Recent studies show that delays and perturbations of motor development affect up to 12% of infants, depending on the time of examination [1]. While a multitude of causes and symptoms exists, early intervention can dampen the effect of impairments [2]. Clearly, the step preceding intervention has to be detection. However, detailed neurological examination requires time and expertise, and is therefore only performed if justified reasons for its necessity have been found. An automated system can help to discover infants who would benefit from early diagnostic and therapeutic action. For wide-spread use, it is required to be cheap, easy to use, unintrusive to the infants and it has to provide objective measures. We propose a system that only depends on a commodity depth sensor (e.g. Microsoft Kinect) and a connected laptop (see Fig. 1). It requires no calibration or attachment of markers or sensors to the children. We build upon a recently introduced system for body pose estimation in depth images using random ferns [3]. We capture 3D body joint positions and head rotation over time, to calculate a variety of motion © Springer Nature Singapore Pte Ltd. 2018 H. Eskola et al. (eds.), EMBEC & NBC 2017, IFMBE Proceedings 65, DOI: 10.1007/978-981-10-5122-7_50

parameters related to range, variability and symmetry of movements.

II.

R ELATED WORK

Several systems have been proposed for the task of detecting infantile movement disorders, especially cerebral palsy, based on