SEIZEµ: Automated Video-EEG Analytic System for Seizure Detection in Epilepsy Monitoring Unit.

<<Back to Recent Research Projects

<< Back to Past Research Areas & Projects

Long-term video-EEG monitoring (VEEG) is a key tool in the diagnosis and classification of epilepsy and to help determine which patients are candidates for epilepsy surgery. It is currently only available at tertiary and quartenary centres and involves prolonged monitoring with both EEG and video data collection for 2- 5 days. Prolonged training is required to attain competency in VEEG interpretation. In addition it is labour-intensive and requires manual evaluation of the entire EEG recording. Despite optimal clinical set-up, important video and EEG data is missed when the patient is out of camera view or facing away from the camera during the seizure onset or there is excessive movement or muscle artefact respectively. Fig. 1 shows our proposed system diagram.

Fig. 1 Our system diagram

Fig. 2 Cameras placed at KKH Hospital which monitor the suspected patients for seizure detection and analysis.

Fig. 3 Body parts labeling and tracking during seziure and normal random movements.

Despite skilled interpretation it is difficult at times to determine where the seizure begins due to movement and muscle artefacts or because the seizure start deep within the brain or spread rapidly. In addition out-of-hours interpretation is not available at many centres, raising clinical issues with seizure recognition and management. It requires high capacity data collection and storage equipment. Video EEG provides simultaneous recording of video and electrical activity in the brain. This enables physicians to relate any abnormal electrical patterns to any physical manifestations that a seizure might cause. This can help physicians identify the seizure focus, estimate seizure frequency, and differentiate seizures from non-epileptic events. In this project we are trying to develop a Video-EEG Analytic system for Seizure Analysis in Epilepsy Monitoring Unit. Figs. 2 and 3, show the patients in the Epilepsy Monitoring Unit (EMU) having seizures. The body parts are segmented and they are tracked in long sequence of videos. Fig. 4 shows another example at different times of the day.

Table 1 shows the performance of our proposed system.

Table 1

We have developed an unobtrusive video-based method for quantifying and characterizing limb movements in epilepsy monitoring. The proposed color-based video system uses a single ceiling camera with customized color pyjamas. After a simple user-initialization on the first frame, our system extracts the positions and angles of patient’s limbs automatically. We further perform (time-domain) displacement analysis as well as frequency analysis to characterize limb movements and detect motor seizures. We identify sustained displacement and strong oscillation as two useful features. In experimental studies on 15 sequences from five patients, the oscillation feature has achieved performance comparable to EEG-based features, while the displacement feature is inferior. On the whole, the proposed video-based system is a promising approach for home monitoring and a good addition to investigation on quantified motor semiology for other movement disorders or behavioral changes, such as sleep disorder analysis.

Details read here:

[J1] H. Lu, P. Yaozhang, B. Mandal, H. L. Eng, G. Cuntai and D.Chan, “Quantifying Limb Movements in Epileptic Seizuresthrough Color-based Video Analysis,” IEEE Transactions on Biomedical Engineering (TBME), vol. 60, no. 2, pp. 461-469, Feb 2013. (Impact factor: 2.525) [PDF]

[C5] B. Mandal, How-Lung Eng, Haiping Lu, D. W. S. Chan and Y.-L. Ng, “Non-intrusive Head Movement Analysis of Videotaped Seizures of Epileptic Origin,” in Proceedings of the 34th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC'12), San Diego, California, USA, pp. 6060-6063, 28 Aug - 1 Sept, 2012. (Oral) [PDF]

[C4] B. Mandal, D. Chan, How-Lung Eng, Haiping Lu and Yen-Ling Ng, “Optical flow information and video seizure recognition,” 12th International Child Neurology Congress (ICNC 2012), Developmental Medicine & Child Neurology, vol. 54, pp. 10, Supplement 4, Brisbane, Australia, 27th May to 1st June 2012. (Poster) [PDF]

[C3] D. Chan, Haiping Lu, B. Mandal, Yen-Ling Ng and How-Lung Eng, “Automated markerless video seizure detection,” 12th International Child Neurology Congress (ICNC 2012), Developmental Medicine & Child Neurology, vol. 54, Supplement 4, pp. 155, Brisbane, Australia, 27th May to 1st June 2012.(Oral/Platform) [PDF]

[C2] Haiping Lu, H.-L. Eng, B. Mandal, D. W. S. Chan and Y.-L. Ng, “Markerless Video Analysis for Movement Quantification in Pediatric Epilepsy Monitoring,” in Proceedings of the 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC'11), Boston, Massachusetts, USA, pp. 8275-8278, 30 August-3 September, 2011. (Oral) [PDF]

[C1] Chan D., Lu H., Eng H-L., Mandal B., Ng Y-L., “Computerised video analysis and quantification of limb movements in automation of seizure detection in children,” 29th International Epilepsy Congress (IEC 2011), Epilepsia, Vol. 52, No. s6, pp. 214, Rome, Italy, 28 August-1 September, 2011. (Oral) [PDF]