Biomedical engineering

1. Diagnosis of Parkinson's disease based on speech analysis using machine learning methods

Supervisor: prof. dr hab. inż. Ryszard Tadeusiewicz

Department of Biocybernetics and Biomedical Engineering, Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract:  The results of the research will positively affect the prevention and treatment of Parkinson's disease.
Parkinson’s disease is the second most common age-related neurodegenerative disorder after Alzheimer’s disease. An estimated seven to 10 million people worldwide have Parkinson’s disease. The prevalence of the disease ranges from 41 people per 100,000 in the fourth decade of life to more than 1,900 people per 100,000 among those who are 80 and older The disease affects patients’ quality of life, making social interaction more difficult and worsening their financial condition,due to the medical expenses associated with the disease. Research assumes the design of algorithms to be implemented in mobile devices to automatically detect early stage of Parkinson's disease based on speech analysis. The designed system will support medical diagnostics and positively influence the prevention of Parkinson's disease. New machine learning methods will additionally be designed and compared in the research.

Research facilities: A proper set of data and a computer with adequate computing power are necessary to complete the planned research. There are many publicly available datasets for the diagnosis of Parkinson's disease based on speech signals. Adequate data sets can be found on popular websites such as UCI Machine Learning Repository or Kaggle. AGH also has appropriate computers to run highly demanding training-related computations and genetic optimization of machine learning algorithms.

Number of places: 1

 

2. Creation, improvement and adaptation of new associative and cognitive methods of artificial intelligence for knowledge modeling, clustering and classification of data, and fast processing of large data sets to control robots in a virtual and real environments.

Supervisor: dr. hab. Adrian Horzyk, prof. AGH

Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The work aims to use associative methods of artificial intelligence and associative graph data structures (i.e. AGDS, MAGDS, DASNG, APNN, AANG), various biologically inspired models of neurons, grounding symbol, associative semantic and episodic memories, machine learning and motivated learning for fast data processing and knowledge modeling in robot systems. The scope of the work includes the creation of associative graph data structures enabling the aggregated and associative representation of data, quick access, and further processing, as well as conclusions based on them. It is also planned to use various association and artificial intelligence methods to cluster and classify data. The researched solutions could be applied in measurement systems, expert systems, and medical diagnostics. It is planned to implement association solutions in the cloud and create a system of association database solutions. The scope of work also includes the creation and implementation of algorithms related to reading data from various sensors of the robot, controlling its motors and arms, grounding symbol, recognition, clustering and classification of images, creating temporal-spatial association relationships between objects, semantic memories, machine learning and cognitive system creation.

Research facilities: The research unit and the supervisor have knowledge and experience in the creation and development of methods of artificial intelligence, computational intelligence and knowledge engineering, so they have the ability to carry out research within scientific projects, including the above mentioned research topic.

Number of places: 2

 

3. Application of new associative models and methods of artificial intelligence for automatic analysis and classification of medical images, ECG signals and examination of disorders.

Supervisor: dr. hab. Adrian Horzyk, prof. AGH

Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The research will focus on the development of new computational models based on associative graph data structures and neural networks for medical data analysis in the form of time series and images. The result of the analysis will be the classification of certain disorders as well as an exploration of knowledge contained in the analyzed data. These models, along with algorithms operating on them, and their practical implementations, in particular, will be designed to help healthcare professionals and patients in non-trivial diagnostic issues. The proposed models will use known solutions, including convolutional and recursive neural networks and machine learning algorithms. In addition, existing associative models and methods of computational intelligence will be used. The developed models will be compared with other available methods of analysis of medical data.

Research facilities: The scientific unit and promoter have knowledge and experience in developing methods of artificial intelligence, computational intelligence and knowledge engineering, so it has the possibility to carry out research within scientific projects, including the above-mentioned research topic.

In addition, thanks to the established cooperation with an eminent specialist in the field of electrocardiology, prof. Marek Jastrzębski we have the opportunity to obtain unique and reliable records of electrocardiograms and to adapt computational models to solve specific problems of this branch of medicine. Current research in this area is very promising.

Number of places: 1

 

4. The use and development of computational intelligence methods for targeted medicine, with the use of digital pills for drug additions and chronic depressions through novela’s ‘NEUREKA’ neurological platform

Supervisor: dr. hab. Adrian Horzyk, prof. AGH

Auxiliary supervisor: dr Piotr Mankowski, Salam Gabran, PhD

Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The purpose of the work is to use computational intelligence methods, machine learning, and microchips and Novel Neuro equipment for medical diagnosis and treatment of drug addictions, epilepsy, depression, bipolar disorder, epilepsy and others, as well as the creation of computational models of neural activity in the brain characteristic of various neurological disorders and asylums. These studies will use, among others, associative neural structures to model processes occurring in biological nervous structures and various models and methods of machine learning.

Research facilities: The research unit and the supervisor have knowledge and experience in the creation and development of methods of artificial intelligence, computational intelligence and knowledge engineering, so they have the ability to carry out research within scientific projects, including the above mentioned research topic. The American company Novela is committed to supporting this research topic with all IT equipment required to conduct activities listed in this application: computers, Novela hardware, probes, cables, BLE modules, NEUREKA cloud repository, all necessary accessories, software and hardware, local engineering team and the company CTO dr Piotr Mankowski will be present at the office monthly to assist with any set-up and research related activities.

Number of places: 2

 

5. Detection and parameterization of the focal areas of the glial transformation of the brain in magnetic resonance imaging.

Supervisor: dr hab. inż. Adam Piórkowski

Department of Biocybernetics and Biomedical Engineering, Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The aim of the work is to develop a segmentation approach for focal areas of glial transformation for MR imaging. The scientific problems that are provided here are the exact determination of the boundaries of changes, the determination of their morphometric characteristics, the characteristics of changes in terms of differentiation. It is planned to create an application that will support the work of radiologists.

Research facilities: For the needs of research, sets of tests are guaranteed in significant quantities, and medical doctors' consultations will be available

Number of places: 1

 

6. Analysis of X-ray images in assessing the effects of distant arthroplasty procedures

Supervisor: dr hab. inż. Adam Piórkowski

Department of Biocybernetics and Biomedical Engineering, Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The issue of this work is to create a methodology and / or application for the needs of orthopedic surgeons and radiologists, allowing the assessment of the hip or knee post-surgery condition after the surgery for potential revisions. The adhesion zones and / or implant placement geometry will be analyzed.

Research facilities: For the needs of research, sets of tests are guaranteed in significant quantities, and medical doctors' consultations will be available

Number of places: 1

 

7. Processing and analysis of Reflectance Confocal Microscopy images

Supervisor: dr hab. inż. Andrzej Skalski

Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: Skin tumors are the most frequent cancer type in the general population with epidemiological data showing a rising trend. In recent years, reflectance confocal microscopy (RCM) has shown to be the most promising diagnostic tool. RCM is non-invasive examination of the skin at cellular resolution enabling visualization of the relevant skin layers.

There are three main issues currently being in the area of research interest:

  • detection of imaging features that enables the differentiation of tumor types,

  • automatic segmentation of RCM images,

automatic classification of RCM images using Deep Learning methods,

Research facilities: Cooperation with: the Collegium Medicum UJ and University of Modena and Reggio Emilia, Department of Dermatology, Modena, Italy, providing access to RCM image data.
Potentially, there is also the possibility of applying for a scientific grant with the mentioned parties.

Number of places: 1

 

8. Detection of heart beats and diagnostic features from irregularly sampled or missing samples electrocardiograms

Supervisor: prof. dr hab. inż. Piotr Augustyniak

Department of Biocybernetics and Biomedical Engineering, Faculty of Electrical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: Aim of the research is to propose and evaluate an efficient method to detect the heart beats from missing samples electrocardiograms. The performance and the loss of performance due to missing samples percentage have to be evaluated accordingly to the international standards. For irregularly sampled ECGs the evaluation has to be done for various sampling modulation patterns to reveal the optimal reliability-to-bitrate ratio.

These studies may also be continued towards adaptation of the regular ECG storage structures and data transmission equipment to accommodate additional motion and environmental data streams. Proposal and numerical validation of the method with recorded data is expected before building and validating the hardware prototype. Evaluation with volunteers and true-to-life scenario-based experiment is necessary for validation of the solutions proposed.

Research facilities: ECG-development software and standard evaluation databases are available, as well as certified interpretive ECG recorder, patient simulator and digital signal replicator. Together with Matlab numerical experimenting platform they consist a complete and unique test bed for virtually any prototype methods of ECG signal processing. PhD students are encouraged to join currently running scientific projects, doctoral international exchange programs or to apply for their individual research grants (such as NCN Preludium).

Number of places: 1

 

9. Synchronous pulse and scanpath measurement from UHD video sequences of multiple faces

Supervisor: prof. dr hab. inż. Piotr Augustyniak

Department of Biocybernetics and Biomedical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: The research consists in proposing an efficient method of multiple face detection, videoplethysmographic pulse detection and scanpath recording. The proposed method should be validated with recorded scenes, then implemented and validated with real time images of computer users. Additional study of continuous VPG signal recording from a mobile human by a multicamera system will be welcome. The topic can be further developed to a visual perception and affect detection study set allowing for inducing emotional states with standard visual cues and detection of the emotion intensity with changes of scanpath features. One of prospective application is personal adapting the computer games.

Research facilities: Affective cues databases (auditory and visual) eyetrackers and other equipment for physiological measurements are available for research. Together with Matlab numerical experimenting platform they consist a complete and unique test bed for virtually any studies on scanpath, visual perception and affect-related cognitive performance. PhD students are encouraged to join currently running scientific projects, doctoral international exchange programs or to apply for their individual research grants (such as NCN Preludium).

Number of places: 1

 

10. Usage of intelligent electrical energy meter and sensorized household equipment to pursuit of behavior in assisted living

Supervisor: prof. dr hab. inż. Piotr Augustyniak

Department of Biocybernetics and Biomedical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: The research aims at building of an intelligent electrical energy meter, tuning it for accurate recognition of typical household equipment and exploring the possible use for detection and assessment of behavioral patterns. Application in volunteers’ homes and scenario-based research is necessary for validation of the methods proposed. The proposed research may be extended towards building and validating of various IoT sensors embedded in domestic appliances. The system may be used together with an inertial or optical motion capture system for investigating the applicability and limitation areas.

Research facilities: An intelligent energy meter development kit can be used to capture device-specific pattern of load. Additionally, numerous IoT samples and gadgets allows for building of virtually any wireless-enabled household device. A wide range of human motion and physiology measurement systems may be used in the research: the ECG/EMG recorders, foot pressure cells, optical marker-based motion capture system. The data collection scenarios include native formats, custom graph-based format or any available digital standards (e.g. DICOM). for effective storage. The data analysis may be performed with Matlab numerical experimenting platform or a wide choice of artificial intelligence environments. PhD students are encouraged to join currently running scientific projects, doctoral international exchange programs or to apply for their individual research grants (such as NCN Preludium).

Number of places: 1

 

11. Compressed sensed multiple-shell high angular resolution diffusion magnetic resonance imaging of the brain.

Supervisor: prof. dr hab. inż. Piotr Augustyniak

Auxiliary supervisor: dr inż. Tomasz Pięciak

Department of Biocybernetics and Biomedical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: The research topic is related to the compressed sensing principle in diffusion magnetic resonance imaging. In the human brain, the molecules diffuse mainly along the primary fiber orientation, and therefore the diffusion magnetic resonance imaging seems to be a flexible tool to provide valuable diagnostic information about the underlying processes for instance in neurodegenerative diseases. Besides, diffusion imaging enables to provide qualitative fiber tract representation of the central nervous system. However, diffusion magnetic resonance imaging requires densely-sampled data using a large number of diffusion encoding gradients. Thus, it leads to a significant increase in the examination time and decrease in the patient's discomfort. Nowadays, typical scanners achieve both physical and biological limits. Therefore, to reconstruct the data from reduced measurements, the compressed sensing methodology is the appropriate mathematical tool. The compressed sensing technique involves several requirements to be fulfilled, such as the proper sampling and numerical methods used to solve non-linear problems. The research aims to optimize sampling schemes in diffusion imaging of the brain using non-Cartesian acquisition trajectories and derive new mathematical models to reconstruct the data from a non-uniformly sampled data accurately.

Research facilities: The research center has extensive experience in the field of signal and image processing methods, especially those related to medical images. The supervisor is an expert in the field of biomedical signal processing. The co-supervisor is an experienced researcher in the magnetic resonance imaging area. The co-supervisor has already established a collaboration with a leading research center in this area, namely the Laboratorio de Procesado de Imagen at the University of Valladolid (Valladolid, Spain). He also has numerous worldwide achievements related to diffusion imaging, including papers presented at top-tier conferences in the field (MICCAI, ISBI, ISMRM, ICIP) and world-class journals (Medical Image Analysis, IEEE Transactions on Pattern Analysis and Machine Intelligence). He also serves as a reviewer for top-tier journals such as the Medical Image Analysis, IEEE Transactions on Image Processing, Magnetic Resonance in Medicine and conferences including ISBI, CVPR and ICCV.

Number of places: 1

 

12. The assessment of the usefulness of the biocybernetic model of bee colony to predict the effects of giving the mother bee stress associated with space flight.

Supervisor: prof. dr hab. inż. Ryszard Tadeusiewicz

Department of Biocybernetics and Biomedical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: In the scientific team currently creating the Chair of Biocybernetics and Biomedical Engineering in the 1980s, numerous works related to biocybernetic modeling of bee colonies were carried out. Original algorithms modeling different aspects of the functioning of the considered biological object were determined and numerous simulation tests were carried out, confronting the results of the simulation with measurements carried out on the real object (appropriately equipped with measurement devices hive). A lot of work was published regarding the construction of the model itself as well as selected applications (especially in the context of the study of bee competition in the same area). It has been found that model-based studies are particularly useful when planning a method of controlling a bee colony for non-standard purposes, for example for obtaining honey instead of honey (which is typical) of a venom or royal jelly for pharmaceutical purposes.

These works were discontinued because a member of the team who had an apiary treated as an experimental laboratory died.

There is now a reason to return to this research because a new non-standard task related to bee breeding has emerged. Well, as part of plans for the so-called terraforming of Mars (creating atmosphere, vegetation, conditions for human life), we will need to transfer insects pollinating plants to this planet - probably bees. An open problem is how bees can take a journey from Earth to Mars. The topic is worth taking as part of a doctoral thesis using your biocybernetic model.

Research facilities: Currently, experiments are carried out involving the sending of bee mothers with rockets or stratospheric balloons to the boundaries of the space - and the promoter has access to the original (not published!) results of this research. These results can lead to important scientific results when they are combined with biocybernetic modeling of bee functioning. This can shed new light on the problem of bees' life when traveling in space and on Mars.

Number of places: 1

 

13. Intelligent embedded systems

Supervisor: dr hab. inż. Piotr Szymczyk

Auxiliary supervisor: dr inż. Magdalena Szymczyk

Department of Biocybernetics and Biomedical Engineering, Automatics, Computer Science, and Biomedical Engineering

Abstract: Intelligent embedded systems are a milestone in building an intelligent human environment. They can be used as an element of such systems as intelligent home, intelligent cars, smart city, etc. Moreover, they are an important and dynamically developing field of medicine - for example intelligent implants that can interpret data from different types of sensors and recognize the patient's state of health, life-threatening condition and take preventive measures at the early stage. The essence of the research topic is the development of the idea of intelligent embedded systems through research on the possibility of implementing artificial intelligence in embedded systems based on microcontrollers and the use of these solutions in medicine as well as in many other fields.

Research facilities: Available research facilities:

Laboratory of Electronic Medical Apparatus

Laboratory of Medical Statistics / Laboratory of Voice Communication Systems with a Computer

Laboratory of Computer Programming / Laboratory of Telemedicine Fundamentals

Laboratory of Embedded Systems and the Internet of Things

Number of places: 2

 

14. Selected phenomena simulation using ray trace technique.

Supervisor: dr hab. inż. Adam Piórkowski

Department of Biocybernetics and Biomedical Engineering, Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering

Abstract: The aim of the work is to create solution for ultrasound wave propagation in three dimensional model space. Data from the CT scans will be used and prepared algorithms should provide realistic US simulated imaging in real time, with selected wave phenomena included. The algorithm will use ray tracing technique and multiprocessor environment.

Research facilities: For the needs of research, sets of tests are guaranteed in significant quantities, and medical doctors' consultations will be available.

Number of places: 1

 

15. The use of the compartment modeling in the assessment of the extracorporeal liver therapy efficiency.

Supervisor: dr hab.inż. Aleksandra Jung

Faculty of Physics and Applied Computer Science

Abstract: Compartmental modeling allows a quantitative description of substances removed during extracorporeal liver therapy. From the clinical point of view, one can get important information on the generation of the substance and the volume of distribution. Until now, models of bilirubin, bile acids and urea kinetics have been developed. The aim of this work would be to link all three models and to find a useful parameter from a clinical point of view. In addition, the influence of selected parameters on the stability of individual models would be examined. In case of obtaining additional clinical data, further verification of the described models would be possible.

Research facilities: At this stage, access to the computer and the use of Matlab software are required, which conditions are met at the faculty.

Number of places: 1