Tweet
Pattern recognition for gait analysis: Integration of clinical expert knowledge and machine learning techniques.
This project aims at developing a clinically relevant automatic classification system for pathological gait in children with cerebral palsy (CP) by improving the state of the art from a clinical and engineering point of view. Two parallel PhD scholarships are available (one with clinical focus and one with engineering focus).
The current announcement aims for both the engineering oriented and clinical oriented PhD scholarship.
From the clinical perspective, there is a continuous challenge to optimize the treatment planning of children with CP, by extracting clinically relevant information from a biomechanical analysis of the pathological gait. In this respect, there is a lack of effective and robust techniques to model the full complexity of gait data and a lack of automatic classification of gait patterns with widespread clinical acceptance. There is also a lack of techniques to model the uncertainty in gait classification, caused by gait inconsistency within and between patients and disagreement between the experts.
From the engineering perspective, there is still a need for algorithms for automatic analysis and classification of multi-dimensional waveforms that improve over data-mining approaches by incorporating expert knowledge and disagreement in a probabilistic approach.
Due the presence within the consortium of clinical expertise regarding gait analysis in CP and the presence of a large gait database (University hospital) and the available expertise on classification algorithms (engineering department), there is an ideal and rather unique opportunity for the development of automatic analysis and classification methods for the gait of children with CP that explicitly model and incorporate prior expert knowledge. This project seizes this opportunity and aims at improving the state of the art in the analysis and classification of continuous waveforms using a systematic model-based approach (engineering discipline), the extraction and mathematical modeling of expert knowledge (clinical discipline), and the incorporation of this expert knowledge in the analysis and classification (interdisciplinary). The common ground of the engineers and the clinicians is the use of an underlying kinematic and dynamic model (in this case the musculoskeletal model of the lower limbs) with different levels of detail.
The clinically oriented PhD will cover the clinical knowledge extraction and the integration of the clinical knowledge in the gait classification. This PhD project will be in close cooperation with the engineering PhD student who will focus on the analysis and classification of multi-dimensional continuous waveforms based on probabilistic Bayesian knowledge-based algorithms. From a clinical perspective the gait classifications in CP will benefit from (1) the incorporation of the expert clinical reasoning, (2) a refined and more generic definition of clinically relevant gait features and total gait patterns; (2) a focus on discrete as well as continuous gait features, (3) the quantification of uncertainty about these gait features and total gait patterns, (4) classification of movement patterns in all three anatomical planes of motion, (5) allowing a soft gait classification (a probability for belonging to each of the classes is calculated) instead of forcing CP gait to fit into one gait class, and (6) incorporating data from a large patient cohort, which includes the diverse nature and range of all possible gait patterns seen across all types of walking CP children.
The engineering oriented PhD will cover the analysis and classification of multi-dimensional waveforms based on probabilistic Bayesian knowledge-based algorithms. This PhD project will be in close cooperation with the clinical PhD student who will focus on the clinical knowledge extraction and the integration of the clinical knowledge in the gait classification.
From an engineering perspective the envisaged progress beyond the state of the art will be the development of a system for the automatic analysis and classification of waveforms that contains all the Leuven International Doctoral School Biomedical Sciences, following characteristics: (1) the underlying probabilistic approach using Bayesian networks, (2) utilizing both discrete waveform features and continuous waveform primitives, and (3) explicitly incorporating and exploiting expert knowledge.
The successful applicant will receive a PhD scholarship for 3 to 4 years.
Candidate requirements:
Candidates for the clinically oriented PhD must hold a master diploma (such as Physiotherapy, kinesiology, human movement sciences) that gives access to the KU Leuven Biomedical Doctoral School PhD program (MSc degree obtained "cum laude" or equal).
Candidates must be fluent in English.
Experience with Clinical motion Analysis, Gait analysis, Vicon Motion Analysis, Matlab, and/or OpenSim would be an advantage.
Candidates for the engineering oriented PhD must hold a master diploma (such as Mechanical, Electromechanical, or Electrical engineering, or Computer Science ) that gives access to the KU Leuven Arenberg Doctoral School PhD program (MSc degree obtained "cum laude" or equal).
Candidates must be fluent in English.
Experience with (bio-)mechanical modelling, signal processing, probability theory, Matlab, and/or C++ would be an advantage.
Key words: Bayesian probability theory, Bayesian networks, signal processing, gait analysis, biomechanics, cerebral palsy, clinical motion analysis
Latest application date: 2012-08-31
Latest decision data: 2012-09-03
Financing: available
Type of Position: scholarship
Source of Funding: OT (KU Leuven)
Duration of the Project : 4 years
Remarks: The academic supervisors of the project are Prof. Kaat Desloovere, Dr. Ir. Tinne De Laet, Prof. Dr. Guy Molenaers, and Dr. Ir. Erwin Aertbeliën, Prof. Dr. Ir. Herman Bruyninckx. The involved research centres (The Department of Rehabilitation Sciences and the Department of Mechanical Engineering of the KU Leuven and Clinical Motion Analysis Laboratory of the University Hospital Leuven) have the required scientific expertise to realize the application of machine learning for pattern recognition and classification of gait patterns. Their collaboration creates a unique opportunity as this interdisciplinary project combines their thorough research experience in the field of CP, gait analysis, and fundamental methodological research questions.
We can offer an enthusiastic, open minded, multi-disciplinary work environment with high focus on optimal integration of clinical and technical expertise, with intensive measurement, analysis and brainstorming sessions in the hospital's gait lab and at the engineering department.
To register your interest please send your CV, copies of you academic transcripts, two academic references and a cover letter that specifies your earliest possible commencement. While it is preferred to send your candidacy to both Kaat Desloovere (kaat.desloovere@uzleuven.be) and Tinne De Laet (tinne.delaet@mech.kuleuven.be), the primary contact person for the clinically and engineering oriented PhD are Kaat Desloovere and Tinne De Laet, respectively.
This project aims at developing a clinically relevant automatic classification system for pathological gait in children with cerebral palsy (CP) by improving the state of the art from a clinical and engineering point of view. Two parallel PhD scholarships are available (one with clinical focus and one with engineering focus).
The current announcement aims for both the engineering oriented and clinical oriented PhD scholarship.
From the clinical perspective, there is a continuous challenge to optimize the treatment planning of children with CP, by extracting clinically relevant information from a biomechanical analysis of the pathological gait. In this respect, there is a lack of effective and robust techniques to model the full complexity of gait data and a lack of automatic classification of gait patterns with widespread clinical acceptance. There is also a lack of techniques to model the uncertainty in gait classification, caused by gait inconsistency within and between patients and disagreement between the experts.
From the engineering perspective, there is still a need for algorithms for automatic analysis and classification of multi-dimensional waveforms that improve over data-mining approaches by incorporating expert knowledge and disagreement in a probabilistic approach.
Due the presence within the consortium of clinical expertise regarding gait analysis in CP and the presence of a large gait database (University hospital) and the available expertise on classification algorithms (engineering department), there is an ideal and rather unique opportunity for the development of automatic analysis and classification methods for the gait of children with CP that explicitly model and incorporate prior expert knowledge. This project seizes this opportunity and aims at improving the state of the art in the analysis and classification of continuous waveforms using a systematic model-based approach (engineering discipline), the extraction and mathematical modeling of expert knowledge (clinical discipline), and the incorporation of this expert knowledge in the analysis and classification (interdisciplinary). The common ground of the engineers and the clinicians is the use of an underlying kinematic and dynamic model (in this case the musculoskeletal model of the lower limbs) with different levels of detail.
The clinically oriented PhD will cover the clinical knowledge extraction and the integration of the clinical knowledge in the gait classification. This PhD project will be in close cooperation with the engineering PhD student who will focus on the analysis and classification of multi-dimensional continuous waveforms based on probabilistic Bayesian knowledge-based algorithms. From a clinical perspective the gait classifications in CP will benefit from (1) the incorporation of the expert clinical reasoning, (2) a refined and more generic definition of clinically relevant gait features and total gait patterns; (2) a focus on discrete as well as continuous gait features, (3) the quantification of uncertainty about these gait features and total gait patterns, (4) classification of movement patterns in all three anatomical planes of motion, (5) allowing a soft gait classification (a probability for belonging to each of the classes is calculated) instead of forcing CP gait to fit into one gait class, and (6) incorporating data from a large patient cohort, which includes the diverse nature and range of all possible gait patterns seen across all types of walking CP children.
The engineering oriented PhD will cover the analysis and classification of multi-dimensional waveforms based on probabilistic Bayesian knowledge-based algorithms. This PhD project will be in close cooperation with the clinical PhD student who will focus on the clinical knowledge extraction and the integration of the clinical knowledge in the gait classification.
From an engineering perspective the envisaged progress beyond the state of the art will be the development of a system for the automatic analysis and classification of waveforms that contains all the Leuven International Doctoral School Biomedical Sciences, following characteristics: (1) the underlying probabilistic approach using Bayesian networks, (2) utilizing both discrete waveform features and continuous waveform primitives, and (3) explicitly incorporating and exploiting expert knowledge.
The successful applicant will receive a PhD scholarship for 3 to 4 years.
Candidate requirements:
Candidates for the clinically oriented PhD must hold a master diploma (such as Physiotherapy, kinesiology, human movement sciences) that gives access to the KU Leuven Biomedical Doctoral School PhD program (MSc degree obtained "cum laude" or equal).
Candidates must be fluent in English.
Experience with Clinical motion Analysis, Gait analysis, Vicon Motion Analysis, Matlab, and/or OpenSim would be an advantage.
Candidates for the engineering oriented PhD must hold a master diploma (such as Mechanical, Electromechanical, or Electrical engineering, or Computer Science ) that gives access to the KU Leuven Arenberg Doctoral School PhD program (MSc degree obtained "cum laude" or equal).
Candidates must be fluent in English.
Experience with (bio-)mechanical modelling, signal processing, probability theory, Matlab, and/or C++ would be an advantage.
Key words: Bayesian probability theory, Bayesian networks, signal processing, gait analysis, biomechanics, cerebral palsy, clinical motion analysis
Latest application date: 2012-08-31
Latest decision data: 2012-09-03
Financing: available
Type of Position: scholarship
Source of Funding: OT (KU Leuven)
Duration of the Project : 4 years
Remarks: The academic supervisors of the project are Prof. Kaat Desloovere, Dr. Ir. Tinne De Laet, Prof. Dr. Guy Molenaers, and Dr. Ir. Erwin Aertbeliën, Prof. Dr. Ir. Herman Bruyninckx. The involved research centres (The Department of Rehabilitation Sciences and the Department of Mechanical Engineering of the KU Leuven and Clinical Motion Analysis Laboratory of the University Hospital Leuven) have the required scientific expertise to realize the application of machine learning for pattern recognition and classification of gait patterns. Their collaboration creates a unique opportunity as this interdisciplinary project combines their thorough research experience in the field of CP, gait analysis, and fundamental methodological research questions.
We can offer an enthusiastic, open minded, multi-disciplinary work environment with high focus on optimal integration of clinical and technical expertise, with intensive measurement, analysis and brainstorming sessions in the hospital's gait lab and at the engineering department.
To register your interest please send your CV, copies of you academic transcripts, two academic references and a cover letter that specifies your earliest possible commencement. While it is preferred to send your candidacy to both Kaat Desloovere (kaat.desloovere@uzleuven.be) and Tinne De Laet (tinne.delaet@mech.kuleuven.be), the primary contact person for the clinically and engineering oriented PhD are Kaat Desloovere and Tinne De Laet, respectively.