Abstract
Early prediction is clinically considered one of the essential parts of cerebral palsy (CP) treatment. We propose to implement a low-cost and interpretable classification system for supporting CP prediction based on General Movement Assessment (GMA). We design a Pytorch-based attention-informed graph convolutional network to early identify infants at risk of CP from skeletal data extracted from RGB videos. We also design a frequency-binning module for learning the CP movements in the frequency domain while filtering noise. Our system only requires consumer-grade RGB videos for training to support interactive-time CP prediction by providing an interpretable CP classification result.
Introduction
Cerebral Palsy (CP) is a permanent movement disorder, and early intervention is the single most effective way to improve long-term outcomes for affected infants. Currently, the "gold standard" for early detection is the General Movement Assessment (GMA).
While effective, GMA has a major bottleneck: it requires highly trained clinical experts to manually observe an infant’s movements. This makes it time-consuming, expensive, and difficult to scale to the general population, especially in resource-limited areas. The researchers behind CP-AGCN aim to automate this process using "consumer-grade" technology—standard cameras and accessible AI—to provide a low-cost, non-intrusive screening tool.
Methodology
The CP-AGCN pipeline consists of three main stages designed to transform raw video into a clinical prediction:
1. Pose Extraction
The system uses OpenPose to extract 2D skeletal joint locations from standard RGB video. This turns a raw video file into a structured "graph" of human movement, focusing the model on kinematics rather than visual background noise.
2. Frequency-Binning Module
Instead of analyzing raw spatial coordinates alone, the system applies a Fast Fourier Transform (FFT) to analyze movement in the frequency domain. A custom binning method was designed to filter out high-frequency noise (such as camera jitter) and focus on the low-to-mid band frequencies where meaningful CP-related movement patterns, such as specific tremors or stiffness, occur.
3. Attention-Informed GCN
The core architecture is a Graph Convolutional Network based on the ST-GCN backbone. The authors integrated an attention mechanism that learns to weight different joints and frequency features. This not only improves accuracy but also provides interpretability, allowing clinicians to see which body parts or movement frequencies the AI is prioritizing in its decision.
Experimental Results
The model was validated on two key datasets: MINI-RGBD (a synthetic dataset) and RVI-38 (a real-world dataset). The system achieved state-of-the-art (SOTA) performance on both, proving its robustness across both simulated and real-world data.
Beyond accuracy, the system is designed for clinical efficiency. While training requires a GPU, the inference (running the prediction) takes only 45 seconds for a 33-second video on a standard Intel i7 CPU. This makes it viable for use in a typical hospital environment without specialized hardware.
Key Insights
- Frequency over Space: Analyzing movement frequencies is more effective than analyzing raw spatial coordinates for distinguishing between normal fidgety movements and CP indicators.
- Interpretability as a Requirement: By using attention weights, the model can explain why it flagged an infant as high-risk, highlighting specific abnormal leg or arm movements for the clinician.
- Low Cost: Because it only requires RGB video (no depth sensors or wearable markers), the system is significantly more accessible than previous automated solutions.
Limitations and Future Work
Despite the strong results, the system’s accuracy is heavily reliant on the quality of the initial pose extraction. If the infant is obscured or the lighting is poor, the "skeleton" data may be inaccurate. Future versions aim to improve compatibility with "dirty" video data and explore moving the model onto embedded systems for real-time, autonomous monitoring. Furthermore, the authors suggest this frequency-based GCN approach could be extended to other motor disorders, such as Parkinson’s disease.
Summary updated