Elsevier

Early Human Development

Volume 85, Issue 9, September 2009, Pages 541-547
Early Human Development

Using computer-based video analysis in the study of fidgety movements

https://doi.org/10.1016/j.earlhumdev.2009.05.003Get rights and content

Abstract

Objective

Absence of fidgety movements (FM) in high-risk infants is a strong marker for later cerebral palsy (CP). FMs can be classified by the General Movement Assessment (GMA), based on Gestalt perception of the infant's movement pattern. More objective movement analysis may be provided by computer-based technology. The aim of this study was to explore the feasibility of a computer-based video analysis of infants' spontaneous movements in classifying non-fidgety versus fidgety movements.

Method

GMA was performed from video material of the fidgety period in 82 term and preterm infants at low and high risks of developing CP. The same videos were analysed using the developed software called General Movement Toolbox (GMT) with visualisation of the infant's movements for qualitative analyses. Variables derived from the calculation of displacement of pixels from one video frame to the next were used for quantitative analyses.

Results

Visual representations from GMT showed easily recognisable patterns of FMs. Of the eight quantitative variables derived, the variability in displacement of a spatial centre of active pixels in the image had the highest sensitivity (81.5) and specificity (70.0) in classifying FMs. By setting triage thresholds at 90% sensitivity and specificity for FM, the need for further referral was reduced by 70%.

Conclusion

Video recordings can be used for qualitative and quantitative analyses of FMs provided by GMT. GMT is easy to implement in clinical practice, and may provide assistance in detecting infants without FMs.

Introduction

Preterm infants are at increased risk for adverse neurodevelopmental outcomes [1]. Up to 18% of surviving infants who are born extremely preterm develop cerebral palsy (CP) [2], and the total rate of neurological impairments is up to 45% [3], [4], [5]. Neuroimaging and clinical neurological examination during the neonatal period are used to assess the risk of later disabilities. Follow-up programs after discharge are implemented in most tertiary care centres caring for these infants in order to provide specific intervention programs and accurate information to parents about their infant's capabilities and prognosis.

A new approach to functionally assess the young nervous system has been presented by Prechtl [6], [7], [8]. Assessment of general movements (GMs), a part of the spontaneous movement repertoire, is a reliable and sensitive tool for the assessment of infant motor development [9], [10]. In particular, the absence of the so-called fidgety movements (FMs) in infants at 9–20 weeks post-term age has been shown to be a marker for later disability and cerebral palsy in particular [7], [11], [12], [13].

FMs are small movements of moderate speed with variable acceleration of neck, trunk, and limbs in all directions [7], [14]. The quality of GMs is observed from video recordings and evaluated by trained observers, and the assessment of general movements is based on a global visual Gestalt perception described by Konrad Lorenz [15]. Lorenz described the mechanism of Gestalt perception as analogous to “subconscious conclusions”, or as three classical steps of inductive natural science; accumulation of observations, systematic ordering of these observations and abstraction of a governing principle. Lorenz highlighted the danger of attending details and loosing the Gestalt perception that is sought [15]. It is, therefore, crucial that the general movement assessment (GMA) observer masters the principle of not focusing on any details in the infant movements during the assessment.

Due to the experience that is needed and the qualitative nature of GMA, the implementation, generalizability and overall utility of the method have been questioned [11], [16], [17]. There are indications that GMA is limited in use in ordinary clinical practice [18]. The Gestalt perception technique requires experience, and clinicians working alone will be at risk of drifting away from the GMA standards over time. Verification of a GMA result needs a second opinion from another experienced GMA observer. Computer-based analysis of GMs, and the incorporation of its results in clinical follow-up programs may offer a supplement to existing clinical methods.

New motion capture technologies have made it possible to perform quantitative analyses of movement and, thereby, discrimination of normal versus pathological movement based on objective criteria. However, such methods are often restricted to laboratories because of the need for comprehensive instrumentation and advanced analyses [19], [20], [21], [22]. To be the first choice in clinical practice, computer-based analysis should be quick to set up, easy to use, and noninvasive for the subjects being studied. Recently, by the use of 2D video recordings, Jensenius et al. [23] developed the Musical Gesture Toolbox (MGT), a software collection for performing video analysis of music-related movements in musicians and dancers. In addition to extracting quantitative measures from the movement in the video recording, the MGT also visualises the qualities of movement. One visualisation method is the motiongram, a 2D representation of movement over time [24]. For this study we have developed the General Movement Toolbox (GMT) as a software solution for studying general movements in young infants.

The aim of this study is to 1) describe the usability of motiongrams in the study of FMs, and 2) by using the GMT and quantitative parameters, to investigate the ability to detect non-fidgety versus FMs.

Section snippets

Subjects

The study group was recruited from St. Olav University Hospital, Trondheim, Norway. Most infants had participated in a previous study on GMA [11]. A convenience sample of preterm and term infants at low or high risk of neurological impairment was included during the period from 2002 to 2004. Infants born after 28 weeks of gestation without any pre- or postnatal complications were considered to be at low risk for neurodevelopmental disorders. Infants were considered to be at high risk of

Results

Eighty-two infants at high (n = 32) and low (n = 50) risks for later neurological impairments were included. The study group consisted of 37 boys and 45 girls. Forty-eight infants (58.5%) were born preterm. In the preterm group, the median gestational age was 29.5 weeks (range 23–36) and median birth weight was 1910 g (range 470–3350). A total of 137 video recordings were obtained from the 82 participating infants in the period 10–18 weeks post-term age with a median recording age of 13 weeks. The

Discussion

The custom-built GMT proved to be a feasible method to generate qualitative and quantitative data based on video recordings of general movements in young infants. Visual representations of the quantity of motion, centroid of motion and motiongrams in particular, can be used for visualisation and qualitative analysis of FMs. Furthermore, quantitative analysis of the variability of centroid of movement proved to be an objective measure to classify the absence or presence of FMs. By employing the

Conclusion

The present study demonstrates a novel, non-intrusive and easily applicable computer-based method to identify the presence of FMs in young infants. A motiongram based on a video recording displayed similar qualitative features as the clinical GMA. Quantitative features related to the quantity of motion and the variability of the centre of movement, were significantly associated with the presence of FMs. More studies are needed on well-defined high-risk populations. The accuracy of CP prediction

Acknowledgements

This work was supported by the Department of Clinical Services and Department of Pediatrics, Trondheim University Hospital, in Trondheim. We thank physiotherapist Toril Fjørtoft for invaluable discussions about GMA, Øyvind Stavdahl for technical assistance and all health professionals contributing to data acquisition in our study.

References (25)

  • J. Schmidhauser et al.

    Impaired motor performance and movement quality in very-low-birthweight children at 6 years of age

    Dev Med Child Neurol

    (2006)
  • C. Einspieler et al.

    Prechtl's assessment of general movements: a diagnostic tool for the functional assessment of the young nervous system

    Ment Retard Dev Disabil Res Rev

    (2005)
  • Cited by (0)

    View full text