Nawar Habboush: Solving the Inverse Problem for Localizing the Biomagnetic Activity in the Heart
Pdf-based submission (available freely via the MACAU system), 2024
This thesis aims to develop a comprehensive solution for both forward and inverse problems in modelling the human heart, with a specific focus on accurately analysing MCG and ECG datasets. The research methodology involves a meticulous process of data recording, MRI processing, and constructing a multi-regional model that segments different tissues based on their relevant characteristics, ultimately providing a solution for the forward problem. The proposed approach employs the use of Kalman filter and state-space models, followed by the GARCH model to solve the inverse problem, resulting in improved accuracy of data analysis and source localization.
This thesis marks the first attempt to apply the Kalman filtering methodology to analyse MCG data, drawing from extensive experience gained in brain research, particularly for EEG and MEG datasets. The proposed approach has been tested and validated using both simulated and real MCG and ECG datasets from individuals, demonstrating its efficacy in analysing heart activity data and its immense potential for clinical applications. The significance of this research lies in its potential implications for the diagnosis and treatment of various heart conditions. The developed methodology can precisely localize sources of heart activity, aiding in diagnosis and intervention planning, such as ablation or pacemaker implantation. The non-invasive method of activity localization using MCG and ECG datasets, as compared to the invasive method of using a catheter, opens up new avenues for the diagnosis and treatment of heart conditions.
While simpler inverse problem methods that do not require high computational power can be used to find the source activity of both MCG SQUID and ECG electrodes datasets, this thesis also aims to provide data analysis for sensors with a lower signal to noise ratio like the Magnetoelectric Sensors being developed in Kiel. The usage of such sensors is cheaper in terms of initial device costs and running costs.
This interdisciplinary research presents a novel methodology for analysing MCG and ECG datasets that has the potential to revolutionize the field of medical science, specifically the diagnosis of heart conditions. The immense potential of this research highlights the significant contributions that interdisciplinary research with engineering can make towards advancing medical science.
Christin Bald: Echtzeitlokalisierung magnetoelektrischer Sensoren
Pdf-based submission (available freely via the MACAU system), 2024
The measurement of magnetic fields is becoming increasingly important in medicine. The signals measured by magnetic field sensors outside the body can be used to infer the processes inside the body. This is done by solving a so-called inverse problem, which needs the positions and orientations of the measuring sensors besides the signals. The positions and orientations should be determined continiously during measurement, since they are not necessarily fixed during a magnetic measurement. This is done by a magnetic localization. This work presents a signal processing chain for magnetic localization in real time. This includes preprocessing steps for extraction of important information from the sensor data, the estimation of the position and orientation of the sensor(array)s and postprocessing of the estimated data. The proposed signal processing chain will be first described theoretically and afterwards evaluated by means of simulations and measurements. The signal processing chain is suitable for all kinds of magnetic sensors in principle. This work focusses on magnetoelectric sensors. Commercially available fluxgate magnetometers have been used for comparison purposes. Different parameters and distortions are investigated, that could have an influence on the localization accuracy. For example, the accuracy at different signal-to-noise ratios will be examined as well as the robustness in the presence of different errors in the forward model, like deviations from the sensor modell or the coil positions and orientations. The measurements are executed with a single sensor in 2D as well as a single sensor and a 3D sensor in 3D. In total, a higher localization accuracy can be achieved with the fluxgate magnetometers in comparison to the magnetoelectric sensors. Nevertheless it can be shown, that the magnetoelectric sensors are suitable for magnetic localization. Besides handing over the position and orientation information for the solution of different kinds of inverse problems, the localization can be used as a stand-alone application. This will be investigated in this work by localizing an ultrasound head. The localization accuracy is not impaired by the ultrasonic device and thus by the potential source of interference.
Arthur Wolf: Entwurf und Evaluierung von Algorithmen für ein Innenraumkommunikationssystem
Shaker-Verlag, 2023
The communication inside a car is often difficult due to the unfavorable seating position and the high background noise level. An in-car communication system (ICC) can improve the communication between passengers while driving. For this purpose, the voice is recorded with a close to the speaker mouth placed microphone and, after appropriate signal processing, it is played back via close to the listener positioned loudspeakers. The ICC system operates in a closed electro-acoustic loop. Therefore, the maximum system amplification is limited by the feedback of the playback signal. Because the system gain required for sufficient support is often in the range of the maximum gain, additional measures must be taken to suppress feedback. In addition to the speech and the feedback signals, the microphones also pick up the driving noises within the vehicle. If the noisy microphone signal is only amplified, the recorded interference will have a negative effect on the speech quality. These signal components should be reduced by suitable noise suppression to improve the ICC output signal.
For the passengers, the perceived system quality depends not only on the amplification level but also on the delay introduced by the ICC system between the direct sound and the system reproduction. This delay should be as low as possible and typically not exceed 15 ms. The acoustic localization of the speaker and the speech intelligibility of the playback is disturbed by a long delay. Due to psychoacoustic effects, the permissible delay depends heavily on the system gain and the arrangement of the loudspeakers in relation to the passengers.
In this work, the signal processing for an ICC system, which works robustly in a vehicle under real conditions, is presented. The focus is on runtime-optimized and computationally efficient algorithms for noise and feedback suppression. The delay at the listener’s ear is reduced to only 10 ms. With the feedback suppression measures described here, the ICC system can also be operated around maximum system gain without instabilities. The implementation of algorithms as real-time digital audio processing and the buildup of a real time demonstrator vehicle made it possible to evaluate the improvement in speech intelligibility achieved in the vehicle environment. In addition to the improvement in speech transmission recorded by measurements, the speech intelligibility and speech quality were confirmed in experiments by test subjects. The results of this evaluation show that with an active ICC system and the signal processing and measures presented here, the passengers on the rear seats (worst listening position) can hear and understand the driver just as well, as the front passenger (best listening position), even at high speeds.
Robbin Romijnders: Inertial Measurement Unit-Based Gait Event Detection in Healthy and Neurological Cohorts
Pdf-based submission (available freely via the MACAU system), 2023
Walking impairments are common in elderly people and its prevalence increases with age. Walking impairments have devastating consequences and are associated with a loss of mobility, increased institutionalization, increased fall risk and decreased quality of life. Numerous disorders of both the central and peripheral nervous system can cause an impaired walking pattern. The objective quantification of walking is therefore of high clinical interest for clinicians, researchers and neurological patients.
Walking is made up from repetitive gait cycles, that can be divided in a stance phase, during which the foot is in contact with the ground, and a swing phase, during which the same foot is swinging forward. These phases are demarcated by gait events that are referred to as initial and final contact. The robust and accurate detection of these gait events is critical for any clinical gait analysis. Recent advances in wearable inertial sensor technology potentially allow the clinical gait analysis to shift to long-term continuous monitoring in the habitual environment. However, to date, the algorithms to extract gait events from inertial measurement unit (IMU) data have limited ecological validity as they have been validated mainly in clinical research settings with straight-line walking trials.
In this thesis a deep learning (DL)-based network is developed to determine gait events from IMU data from a shank- or foot-worn device. The DL network takes as input the raw IMU data predicts for each time step the probability that it corresponds to an initial or final contact. The algorithm is validated for walking at different self-selected speeds across multiple neurological diseases and both in clinical research settings and the habitual environment. The algorithms shows a high detection rate for initial and contacts, and a small time error when compared to reference events obtained with an optical motion capture system or pressure insoles.
Based on the excellent performance, it is concluded that the DL algorithm is well suited for continuous long-term monitoring of gait in the habitual environment.
Michael Brodersen: Signalverarbeitung für Kommunikationssysteme von Atemschutzvollmasken
Pdf-based submission (available freely via the MACAU system), 2021
Full-face masks are essential for fire fighters to ensure respiratory protection in smoke diving incidents. While such masks are absolutely necessary for protection purposes on the one hand, they impair the voice communication of fire fighters drastically on the other hand. For this reason mask integrated communication systems can be used to amplify the speech, therefore, to improve the communication intelligibility and quality. The communication system picks up the speech signal by a microphone in the mask, enhances it by a digital signalprocesser, and plays back the amplified signal by loudspeakers located on the outside of such masks, transmits the signal via a local wireless network to other communication systems and routes the signal to an attached tactical radio. The enhancement via microphone and loudspeakers is only possible to a limited extend, due to the disturbing breathing and ambient noise, and the resulting coupling feedback of the loudspeaker to the microphone.
To increase the speech intelligibility and solve the problems shown before, this work examines different algorithms to improve communication for masks based on digital signal processing. Since breathing noise is picked up by the microphone, it is detected and suppressed by a voice activity detection. This algorithm ensures that only speech components are played back. In addition the ambient noise is estimated and suppressed. Due to the fact that the microphone is located close to the loudspeaker, feedback is occurring and this is reduced by feedback cancelation. To enhance the functionality of the canceler a decorrelation stage is applied to the signal. After the microphone enhancement the signals are mixed to the dedicated output signals. The post processing is possible for each output signal and includes an exciter, an equalizer, a dynamic range control, and a hard limiter. The exciter regenerates lost signal components due to the attenuation through non-linear characteristics. Equalization filters are applied to improve the stability of the system on the one hand and to enhance the perceived quality of the output signals on the other hand.
All described processing steps are implemented on a 16-bit fixed point digital signal processor and optimized for efficiency. Finally possible evaluation scenarios for masks communication system are presented.
Page 1 of 4
We are delighted to announce that Marten Finck has been awarded the GaS prize for his Master's thesis on "Evaluation of Speech Quality Using Machine Learning".
In his work, he investigated possible side effects of magnetic resonance-guided focussed ultrasound treatment (MRgFUS) on speech quality in patients with neurological diseases such as Parkinson's disease and essential tremor. This innovative and non-invasive treatment method is used to reduce tremor, i.e. the shaking of limbs.
To evaluate the effects on speech, pre- and post-operative speech recordings of 56 patients were analysed: Speech quality was assessed in five categories (monotony, clarity, fluency, auditory effort and vocal tremor) through a survey with 21 participants and used to create a convolutional neural network (CNN). This generates an objective assessment of speech quality using mel frequency cepstral coefficients (MFCCs), fundamental frequency and loudness. A mean absolute error (MAE) of 0.81 on a scale of 0 to 10 was achieved, compared to a standard deviation of 0.67 for the survey participants compared to a reference assessment. The evaluation of the survey revealed minor positive and negative changes in speech quality, which were balanced overall and tended to indicate natural speech variations in connection with the patients' underlying illnesses.
Congratulations on this well-deserved honour!