Analysis of Pathological and Non-pathological Human

Motion with 3D Motion Capture Methods

Analysis of Pathological and Non-pathological Human

Motion with 3D Motion Capture Methods

The 138/2001 ICT R&D Project

General goals – Buildup of a new MOCAP laboratory– Evaluation of human motion in 3D space: whole body,

facial expressions, gestures– Support of diagnostics, physical therapy and clinical

rehabilitation, emotional computing and telepresence.

Participants– TU Budapest Department of Control Engineering and

Information Technology– Foundation of the „Bone and Joint Decade”– The Collegiate Department of the Semmelweis University

Detailed information about the project: lovanyi@iit.bme.hu

In Hungary

• Low back pain is first amongst diseases causing the highest number of days taken off from work

• Low back pain is the third biggest cause disabling workers

• In 1999, there were 55000 cases of treatment in hospitals because of low back pain. The total cost is estimated at more than 3,6 billion HUF (about 16.1 million USD)

Sickness Benefit Days Caused by Different Locomotor Diseases

Muscle damage (0.2%)

Joint inflammation(0.1%)Autoimmune soft tissuediseases (0.5%)Seropositive rheumatoidarthritis (0.8%)Knee-joint diseases(2.5%)Other joint diseases(2.6%)Crystal arthropaties(3.7%)Arthrosis (10.3%)

Spinal abnormalities(79.2%)

Goals - from a medical perspective Registration of the basic motion of healthy individuals

from different groups 2D and 3D registration and analysis of the basic

motion of patients suffering from acute low back pain 2D and 3D registration and analysis of the basic

motion of patients suffering from chronic low back pain Development of a standard motion therapy for acute

and chronic low back pain, and the effective evaluation of it

Consultation of patients suffering in acute low back pain and in sciatic neuralgia

Conventional diagnostics on spinal lumbago

Conventional diagnostics on spinal lumbago

• Anamnesis• Inspection• Palpation• Functional examinations: active/passive motion detection• etc.

• Anamnesis• Inspection• Palpation• Functional examinations: active/passive motion detection• etc.

Conventional diagnostics is a rather complex task! It is hard to interpret these steps from an engineering point of view aiming to substitute them by computer – aided solutions

3D MoCap devices

• Ultrasound devices are generally very robust, realtime - but they have limited functionality

• Optical solutions are expensive, they use several cameras generating complex illumination / background / calibration problems - but there is no need for cables.

We are trying Sensor fusion (ultrasound / optical)

• Markers: where? how many?• Inverse kinematic modelling: the number of markers

can be reduced

Hierarchy of algorithms for video-rate image

processingEven accurate real-time interpreting of a single image feature is difficult… A hierarchical set of features is needed for a sequence of images! Example:

Intensity /histogram

Color

Edges

Series of convolution

Geometrical properties

FFT / shape coefficients

Textures

Contours

Topological properties

From simple but less robust features

To complex but more robust features

Goals - from an ICT perspective• 2D-2,5D-3D-4D data acquisition• Measurement and modeling of human behavior in

daily life • Creation of human movement standards and

reference database for healthy and pathological cases• Projection of clinical models into adequate information

system representation• Development and implementation of numerically

stable mathematically founded image processing algorithms

• Application-specific, content-based indexing

Goals - from an ICT perspective

• Real-time 2D / 3 D image processing and visualization

• Improving efficiency of diagnosis and therapy by means of statistics and reference database

• Tele-consulting / tele-diagnostics methods for therapy and prevention

• Theraphy: teaching of optimal motion using real-time electromechanical, visual, acoustical feedback

• Web-based educational material for learning and assessment

• Buildup of a new medical MoCap system• Equipping of the MOCAP laboratory

Taxonomy of the „Looking at people” problem

– Full body motion detection– Facial expression evaluation– Gesture analysis

Our field of interest: medical applications of human movement analysis using 3D image processing tools

3D MoCap

Applications

Registration of typical symptomsTelemedicine

Disease standards

The continuous objective following up of the patients

3D Analysis

The early recognition of the disease

More effective treatment Cost-reduction

Clinical rehabilitation

Physical therapy

Non-pathological human motion: Sport

Non-pathological human motion:Training

Non-pathological human motion: National Cultural Heritage (eg. Folk dancing)

Conclusion

Due to experiences acquired by testing available medicalpurpose MoCap systems (e.g. Zebris) we decided to startour system development by carefully choosing anappropriate MoCap device as a „starting point”. Then ourhardware-software modifications will be added (e.g.

sensorfusion). Such way development cost and time mayconsiderably be reduced.