Advancing Non-invasive Human Motion Characterization in the Clinical Domain: Methods and Applications Call for Papers

1st Workshop on Advancing Non-invasive Human Motion Characterization in the Clinical Domain: Methods and Applications (ANIMA)

Workshop at BMVC, Glasgow, UK. 27th-28th November (exact date TBA)

https://anima2024.sites.uu.nl/
 

Motivation and topics

In the healthcare domain, understanding and characterizing human
motion is essential for tasks, including diagnostics, monitoring and
rehabilitation. Traditionally, the gold standard to accurately
characterize and study human motion relies on motion capture systems
and physical markers placed on the skin. These techniques are
intrusive, expensive and they may limit natural
movements. Furthermore, they limit the natural environment in which
the analysis can take place. Recently, video analysis has become an
increasingly viable alternative to marker-based systems to perform
human motion analysis. This is due to the increasing progress – in
terms of accuracy and computational resources needed – of deep
learning algorithms in solving computer vision problems. In
particular, recent advancements in deep learning-based Human Pose
Estimation (HPE) algorithms enable the automated quantitative analysis
of human motion from video data.

 
The application of computer vision in healthcare has the potential to
revolutionize how we analyze human behavior. This workshop is
positioned at the intersection of computer vision and medical
applications, emphasizing the importance of extracting meaningful
insights from video data. Our primary interest lies in the behavioral
analysis of human motion. This focus is particularly crucial in
healthcare, where precise understanding of an individual's movements
can aid in early detection of neuromotor disorders, personalized care
plans and effective rehabilitation strategies.

 

The medical domain poses unique challenges in ensuring robustness and
high accuracy. Moreover, clinical applications require tailoring to
specific demographics such as infants, elderly, or people with
physical impairments. Consequently, dealing with data scarcity for
training and benchmarking is another challenge. Our workshop aims to
contribute to the broader computer vision community by focusing on
those challenges that are inherent, but not unique, to the medical
domain. We believe that tackling these topics in behavioral motion
analysis within the medical domain will not only advance healthcare
technology but also push the boundaries of computer vision research.
 

Topics of the workshop

    Motion quantification: measurement of human pose and motion in 2D
    or 3D.

    Motion classification: detection of specific human motions,
    training classifiers with limited data.

    Clinical datasets: dealing with data scarcity, privacy, federated
    learning, synthetic data, and benchmarking.

    Motion recording: use, calibration and combination of various
    sensors.

    Applications: in the domain of infant analysis, diagnostics and
    rehabilitation.

    Real-time analysis: algorithms to perform human motion analysis in
    real-time, enabling applications such as continuous monitoring in
    clinical settings.

    Ethical considerations: studies that address ethical implications
    of using computer vision in healthcare, including issues related
    to privacy, consent and bias in algorithmic decision-making.

 

Invited speakers

Dr. Dimitris Tzionas is an assistant professor at the University of
Amsterdam. He conducts research on the intersection of Computer
Vision, Computer Graphics and Machine Learning. His motivation is to
understand and model how people look, move and interact with the
physical world and with each other to perform tasks. This involves:
(1) accurately “capturing” real people and their whole-body
interactions with scenes and objects, (2) modeling their shape, pose
and interaction relationships, (3) applying these models to
reconstruct real-life actions in 3D/4D and (4) using these models to
generate realistic interacting avatars in 3D/4D. Potential
applications include Ambient Intelligence, Virtual Assistants,
Human-Computer/Robot Interaction and Mixed Reality. The long-term goal
is to develop human-centered AI that perceives humans, understands
their behavior and helps them to achieve their goals.

 

Dr. Logan Wade is a Research Fellow at the University of Bath, United
Kingdom. As a clinical biomechanist, his research harnesses computer
vision and machine learning to identify how patients move, with the
goal of integrating biomechanical measures into clinical
practice. Recent advances in Artificial Intelligence has seen the rise
of motion capture methods that are fast and minimally invasive,
allowing collection of data in clinics that was previously restricted
to high-end biomechanical laboratories. However, while the accuracy of
these systems has drastically improved over the past decade,
determining if their accuracy is sufficient for use on an individual
patient level is still to be determined. His long-term goal is to
develop computer vision tools that are clinically relevant, employing
mediums such as markerless video capture to identify movements of the
body and 3D ultrasound to examine patient-specific spinal postures.

 

Dr. Sara Moccia is an Associate Professor in bioegineering at
Universit`a degli Studi G. d’Annunzio (Chieti, Italy). She works on
designing AI algorithms for clinical data analysis, with a specific
focus on preterm infants’ care. She is the author of more than 50
papers. She is PI for three research projects for a total budget of
around 2 mln euro. She serves as Associate editor for two
international journal and currently as program chair for IPCAI

 

Dr. Simona Tiribelli is the director for AI Ethics of the Institute
for Technology & Global Health at the MIT-funded spin-off PathCheck
Foundation (Boston, US), assistant professor at the University of
Macerata (Italy), where she teaches Ethics of Artificial Intelligence
and Global Justice and Technology, 2023 visiting scholar in AI ethics
at the New York University (NYU), and 2020 Fulbright awarded and
fellow at the MIT Media Lab, Massachusetts Institute of Technology,
US. She is also a founder of the spin-off GAIA (AI Ethics and
Governance) and AI Ethics advisor for companies in Europe and US. She
authored two books and a number of articles in leading scientific
international journals on ethics of artificial intelligence and
digital technology, and delivered on invite more than 50 talks in
academic institutions such as Harvard University, Tufts University,
Toronto University, and many more, in Europe, Canada, and USA.

 

Important dates

Paper submission: August 25th, 2024

Notification of acceptance: September 8th, 2024

Camera-ready submission: September 16th, 2024

 

Submission

Workshop papers should adhere to the paper guidelines of the main
conference: https://bmvc2024.org/authors/author-guidelines/ Accepted
papers will be included in the BMVC workshop proceedings published and
DOI-indexed by BMVA. Submissions can be made through the submission
system: TODO

 

Organizers

Lucia Migliorelli: Department of Information Engineering, Marche
Polytechnic University, Italy, l.migliorelli@staff.univpm.it

Matteo Moro: Department of Informatics, Bioengineering, Robotics and
Systems Engineering (DIBRIS), University of Genova & Machine Learning
Genoa (MaLGa) Center, Genova, Italy, matteo.moro@unige.it

Ronald Poppe: Department of Information and Computing Sciences,
Utrecht University, Utrecht, The Netherlands, r.w.poppe@uu.nl