MarLe : Markerless estimation of head pose for navigated transcranial magnetic stimulation

Renan H. Matsuda*, Victor H. Souza, Petrus N. Kirsten, Risto J. Ilmoniemi, Oswaldo Baffa

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

2 Citations (Scopus)
39 Downloads (Pure)


Navigated transcranial magnetic stimulation (nTMS) is a valuable tool for non-invasive brain stimulation. Currently, nTMS requires fixing of markers on the patient’s head. Head marker displacements lead to changes in coil placement and brain stimulation inaccuracy. A markerless neuronavigation method is needed to increase the reliability of nTMS and simplify the nTMS protocol. In this study, we introduce and release MarLe, a Python markerless head tracker neuronavigation software for TMS. This novel software uses computer-vision techniques combined with low-cost cameras to estimate the head pose for neuronavigation. A coregistration algorithm, based on a closed-form solution, was designed to track the patient’s head and the TMS coil referenced to the individual’s brain image. We show that MarLe can estimate head pose based on real-time video processing. An intuitive pipeline was developed to connect the MarLe and nTMS neuronavigation software. MarLe achieved acceptable accuracy and stability in a mockup nTMS experiment. MarLe allows real-time tracking of the patient’s head without any markers. The combination of face detection and a coregistration algorithm can overcome nTMS head marker displacement concerns. MarLe can improve reliability, simplify, and reduce the protocol time of brain intervention techniques such as nTMS.

Original languageEnglish
Pages (from-to)887–896
JournalPhysical and Engineering Sciences in Medicine
Issue number2
Publication statusPublished - Jun 2023
MoE publication typeA1 Journal article-refereed


  • Coregistration
  • Face detection
  • Markerless tracker
  • Neuronavigation
  • Transcranial magnetic stimulation


Dive into the research topics of 'MarLe : Markerless estimation of head pose for navigated transcranial magnetic stimulation'. Together they form a unique fingerprint.

Cite this