Offre de thèse : Intelligent inference of MSK tissues from markerless Motion Capture (MoCap) and Statistical Shape-Intensity Models (S2IM)

Title: Intelligent inference of MSK tissues from markerless Motion Capture (MoCap) and Statistical Shape-Intensity Models (S2IM)
LaTIM, INSERM UMR 1101, Brest, France
Co-supervision: Valérie BURDIN, LaTIM (LaTIM, IMT Atlantique, Brest, France)
                            Jocelyne TROCCAZ, TIMC (TIMC, CNRS, Grenoble, France)
                           Guillaume DARDENNE, LaTIM (LaTIM, CHRU Brest, France)
Starting date: October 2021 for 36 months
Application deadline: 31 May 2021

Reference: PhD-S2IM


Interested candidates (Please use the subject PhD-S2IM in your message)
Send to before 31 May 2021

  • Detailed CV and cover letter
  • Main diplomas
  • All transcripts (Bachelor and MSc or DNM or Engineering School)
  • Two letters of recommendation

En savoir plus

Motivations: The determination of musculoskeletal information (MSK) is of great interest for several clinical specialties, e.g. in functional rehabilitation to detect possible deficiencies, or in orthopaedic surgery to diagnose pathology or determine relevant intraoperative clinical parameters. This clinical information is usually assessed today using optical markers that can be recognised by three-dimensional (3D) locators. These markers can either be attached to instruments or placed directly on the bone or skin depending on the clinical application. Although these systems have been shown to significantly improve clinical benefits in many settings, they are not widely adopted by the clinical community today mainly due to their price and complexity of use.

Objectives: The objective of this thesis is therefore to develop new algorithms for markerless localisation of MSK tissue from skin envelope analysis in order to provide physicians with clinically relevant parameters in a  noninvasive manner.

Scientific content: Several studies have analysed the impact of skin movement on the accuracy of locating underlying bony structures using skin markers alone. It has been shown that by increasing the number of skin markers, improving their positioning and incorporating a priori modelling of the bone joint, accuracy can be significantly improved. More recently, it has been shown, in the context of a research work emerging from the labex CAMI, that it would be possible to estimate with sufficient accuracy the mechanical axis of the femur, using sparse markers and a statistical shape model (SSM). Based on these preliminary results, we would like to investigate the use of RGB-D motion capture (MoCap) to markerlessly acquire the skin envelope and extract some skin patterns related to the underlying MSK tissue. Based on this information, we want to derive clinically relevant and interpretable parameters.

Publié le 28.04.2021

par Gwenaël BRUNET