A Comparison of Video-based Methods for Neonatal Body Motion Detection

Preterm infants in a neonatal intensive care unit (NICU) or a medium care unit (MCU) are continuously monitored for their vital signs, such as heart rate and oxygen saturation. Body motion patterns are documented intermittently by clinical observations. Changing motion patterns in preterm infants are associated with maturation and clinical events such as late-onset sepsis and seizures. However, continuous motion monitoring in the NICU setting is not yet performed. Video-based motion monitoring is a promising method due to its non-contact nature and therefore unobtrusiveness. This study aims to determine the feasibility of simple video-based methods for infant body motion detection. We investigated and compared four methods to detect the motion in videos of preterm infants, using two datasets acquired with different types of cameras. The RGB dataset was collected in incubators at NICU, containing 9 hours of annotated videos from 5 infants. The other thermal dataset contains 32 hours of annotated videos from 13 infants in open beds at MCU. The comparison methods include background substruction (BS), sparse optical flow (SOF), dense optical flow (DOF), and oriented FAST and rotated BRIEF (ORB). The detection performance and computation time were evaluated by the area under receiver operating curves (AUC) and run time using the same hardware. We conducted experiments to detect motion (from any body part) and gross motion (from torso) respectively. In the thermal dataset, the best performance of both experiments is achieved by BS with mean (standard deviation) AUCs of 0.86 (0.03) and 0.93 (0.03). In the RGB dataset, SOF outperforms other methods with an AUC of 0.82 (0.10) for motion detection. DOF stands out with an AUC of 0.91 (0.05) when detecting gross motion. Our experiments indicate that it is possible to detect motion automatically. All methods are efficient to be integrated into a camera system when using low-resolution thermal cameras. The thermal camera can be used in all light conditions, but due to its lower resolution, the BS method is more suitable to detect (gross) motion for preterm infants. Regarding the high-resolution RGB camera, the BS and SOF are recommended to be implemented. Both methods can be used for (gross) motion detection, but BS is more sensitive to sudden illumination changes.