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

Abstract

Preterm infants in a neonatal intensive care unit (NICU) 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 infants, using two datasets acquired with different types of cameras. The thermal dataset contains 32 hours of annotated videos from 13 infants in open beds. The RGB dataset contains 9 hours of annotated videos from 5 infants in incubators. The compared 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. We conducted experiments to detect motion and gross motion 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 the other methods in both experiments with AUCs of 0.82 (0.10) and 0.91 (0.05). All methods are efficient to be integrated into a camera system when using low-resolution thermal cameras.

Continuous sensing and quantification of body motion in infants: A systematic review

“…The motion is most frequently quantified as a separate physiological signal [ 10 , 23 ], preparing for the further analysis of various motion-related pathological conditions in infants as shown in Fig. 2 .…”

“…The order is sorted by the aim of motion quantification, and year for each sensing technology. Author (Refer) Year Subject size Sensing technology Product model Placement Aim of motion quantification Method of motion quantification Ossmy et al [ 21 ] 2020 I = 1 RGB Camera Not mentioned Overhead camera view Motion/position monitoring: Generate the moving path of the infant Changes in feet location detected by convolutional pose machine Zhao et al [ 22 ] 2020 I = 1 RGB Camera OV5642 On the ceiling of the incubator Motion/position monitoring: Trigger alarm system to notify caregivers Background subtraction Peng et al [ 23 ] 2022 I = 18 RGB/Thermal Camera UI-3860LE-C-HQ/FLIR Lepton5.5 On top of the incubator or bed Motion/position monitoring: Motion detection Background subtraction, optical flow, ORB Mazzarella et al [ 24 ] 2020 I = 14 RGB Camera 10-camera Vicon motion capture system In front of infants Neuromotor pathology: Extract features from quantified motion signals to detect perinatal stroke and cerebral palsy Changes in 3D coordinates of markers Malik et al [ 25 ] 2020 I = 2 RGB Camera Not mentioned Bedside Neuromotor pathology: Magnify motion signal to detect tremors Convolutional encoder-manipulator-decoder network …”

“…Since the NICU environments are usually dark, and RGB-based technologies suffer from this low light condition, infrared (IR)/thermal camera-based technologies have drawn attention and have been investigated in the context of motion detection in infants. Peng et al [ 23 ] did a comparison study of video-based methods (both in RGB and IR systems) for motion detection in 18 infants and compared the performance to the manual annotation of fine and gross motion. This study illustrated the feasibility of 4 different motion quantification methods including background subtraction, sparse optical flow, dense optical flow, and oriented FAST and rotated BRIEF (ORB) [ 56 ].…”

Fault Tolerance Smart Egg Incubation System with Computer Vision

“…Fertility detection is again targeted in the results, although the field of use in this case is pharmacy. www.ijacsa.thesai.org A large number of papers [13,14,15,16] oriented to a human incubators describe machine-vision mechanisms for monitoring the state of babies. Different type of cameras can monitor movements, temperature and persistence of a newborn.…”

Continuous Movement Quantification in Preterm Infants Using Fiber Mat: Advancing Long-Term Monitoring in Hospital Settings

“…To address this need, different noncontact or unobtrusive approaches to quantify movement in infants have been explored. As summarized in our recent systematic review [2], video-based motion quantification approaches were most often investigated as a noncontact and less obtrusive monitoring technology in infants, showing promising results on movement and cerebral palsy detection [5], [9]. Specifically, as reported in [9], an average area under the receiver operating curve (AUC) of 0.86 was obtained for general movement detection in preterm infants by using the background substruction method in video recordings.…”