loading page

Mark3D -A semi-automated open-source toolbox for head-surface reconstruction and electrode position registration using a smartphone camera video
  • +3
  • Suranjita Ganguly,
  • Malaaika Mihir Chhaya,
  • Ankita Jain,
  • Aditya Koppula,
  • Mohan Raghavan,
  • Kousik Sarathy Sridharan
Suranjita Ganguly
Dept. of Biomedical Engineering, Indian Institute of Technology

Corresponding Author:[email protected]

Author Profile
Malaaika Mihir Chhaya
Dept. of Biomedical Engineering, Indian Institute of Technology
Ankita Jain
Dept. of Heritage Science and Technology, Indian Institute of Technology
Aditya Koppula
Dept. of Biomedical Engineering, Indian Institute of Technology, Dept. of Physiology, Apollo Institute of Medical Sciences and Research
Mohan Raghavan
Dept. of Biomedical Engineering, Indian Institute of Technology
Kousik Sarathy Sridharan
Dept. of Biomedical Engineering, Indian Institute of Technology

Abstract

Source localization in EEG necessitates aligning EEG sensor coordinates with the subject's MRI, typically done through electromagnetic tracking or 3D scanning. Both methods have drawbacks-electromagnetic tracking is slow and immobile, while 3D scanners are expensive. Photogrammetry offers a cost-effective alternative, but requires multiple photos to sample the head, with good spatial sampling. Post-reconstruction, existing tools for electrode position labelling on the 3D head-surface have limited visual feedback and an inability to accommodate customized montages, typical in multi-modal measurements. Mark3D is an open-source, integrated tool for 3D head reconstruction from phone camera videos. Its eliminates the need for precise spatial sampling during image capture. It includes blur detection algorithms, a user-friendly interface for electrode and tracking, and integrates with popular toolboxes like FieldTrip and MNE Python. The accuracy of the proposed method was benchmarked with the head-surface derived from a handheld 3D scanner ("ground truth"). The derived head shapes were used to reconstruct source information and the source estimates from the ground truth and video-based models were compared. The error in source reconstruction using the photogrammetry model was found to be 0.033 ± 0.016 mm and 0.037 ± 0.017 mm in the left and right cortical hemispheres respectively.
14 May 2024Submitted to TechRxiv
20 May 2024Published in TechRxiv