Related Experiment Video
Updated: Jul 8, 2025
Quantifying Learning in Young Infants: Tracking Leg Actions During a Discovery-learning Task
Published on: June 1, 2015
Deep-Learning Markerless Tracking of Infant General Movements using Standard Video Recordings
Insights
This study introduces a low-cost, markerless AI system for tracking infant General Movements (GM) using handheld devices. This technology aids early detection of neurological impairments like cerebral palsy, enabling timely interventions.
Area of Science:
- Developmental neuroscience
- Medical imaging
- Artificial intelligence in healthcare
Background:
- Spontaneous General Movements (GM) are crucial for assessing infant neurodevelopment.
- Abnormal GMs predict neurological impairments, including cerebral palsy (CP).
- Early intervention is key for optimal neuromuscular outcomes, but current assessment methods are time-intensive.
Purpose of the Study:
- To develop a simple, inexpensive, and broadly applicable markerless pose-estimation method for automatic infant movement tracking.
- To utilize conventional video recordings from handheld devices for infant GM analysis.
- To enable early identification of infants at risk for neurological disorders.
Main Methods:
- A markerless pose-estimation approach using deep learning (resnet152, mobile-net-v2-1) to identify 12 anatomical locations per frame.
- Tracking infant movements from conventional video recordings using handheld devices.
- A novel unsupervised approach fusing Kalman filter and deep-net outputs for validation.
Main Results:
- Deep learning models (resnet152, mobile-net-v2-1) achieved high accuracy (>99.65% aPCK) in identifying anatomical locations.
- The system successfully tracked infant movements in unseen video frames.
- Validation confirmed the models' capability in identifying key body parts.
Conclusions:
- The proposed technology is feasible for automatic infant GM tracking using low-cost devices.
- This approach facilitates early diagnosis of neurological conditions when interventions are most effective.
- It offers a publicly accessible tool for monitoring infant neurodevelopment and identifying at-risk infants.
Abstract:
Monitoring spontaneous General Movements (GM) of infants 6-20 weeks post-term age is a reliable tool to assess the quality of neurodevelopment in early infancy. Abnormal or absent GMs are reliable prognostic indicators of whether an infant is at risk of developing neurological impairments and disorders such as cerebral palsy (CP). Therapeutic interventions are most effective at improving neuromuscular outcomes if administered in early infancy. Current clinical protocols require trained assessors to rate videos of infant movements, a time-intensive task. This work proposes a simple, inexpensive, and broadly applicable markerless pose-estimation approach for automatic infant movement tracking using conventional video recordings from handheld devices (e.g., tablets and mobile phones). We leverage the enhanced capabilities of deep-learning technology in image processing to identify 12 anatomical locations (3 per limb) in each video frame, tracking a baby's natural movement throughout the recordings. We validate the capability of resnet152 and a mobile-net-v2-1 to identify body-parts in unseen frames from a full-term male infant, using a novel automatic unsupervised approach that fuses likelihood outputs of a Kalman filter and the deep-nets. Both deep-net models were found to perform very well in the identification of anatomical locations in the unseen data with high average Percentage of Correct Keypoints (aPCK) performances of >99.65% across all locations.Clinical relevance-Results of this research confirm the feasibility of a low-cost and publicly accessible technology to automatically track infants' GMs and diagnose those at higher risk of developing neurological conditions early, when clinical interventions are most effective.
