Towards intelligent mechanical ventilation guided by electrical impedance tomography
Item statusRestricted Access
Embargo end date16/06/2024
Mechanical ventilation (MV) occupies a critical position in the field of modern clinical medicine. MV is an effective mean to artificially replace, control, or change spontaneous breathing movement and ventilation function. Therefore, it has been widely used in respiratory failure caused by various reasons, breathing management of anesthesia during major surgery, supportive respiratory care and emergency resuscitation. However, most the traditional ventilators can only provide one-dimensional information such as pressure. Such comprehensive information cannot analyze the local lung injury. Electrical impedance tomography (EIT) is a very promising imaging technique. After the electric field is applied, the induced boundary voltages can be measured to image the spatial conductivity distribution of the sensing region bounded by electrodes. EIT has the advantages of high speed, high temporal resolution, low-cost, non-invasive, and radiation-free, etc., which makes it arouse wide attention and interest in the field of biomedical imaging. Well-studied biomedical clinical applications of EIT include breast and prostate cancer screening, functional brain imaging, thoracic imaging, and lung ventilation monitoring. Therefore, it shows excellent potential in organ characterization, where experimental samples have the same electrical properties as tissues of the human body. This PhD work mainly explores the possibility of more emerging biomedical applications of EIT in pulmonary ventilation monitoring. The combination of EIT and ventilator realized real-time pulmonary monitoring and control, expanded the dimension of useful information and clearly saw the volume change of the lung, which could help doctors better analyze the quality of lung function. This work aims to construct a novel and intelligent EIT-guided mechanical ventilation modality, to evaluate the possibility of this medical device qualitatively and quantitatively, and to improve its performance in real-time 3D monitoring. Around this topic, the contributions of this thesis can be summarized from the aspects of Towards Intelligent Mechanical Ventilation Guided by Electrical Impedance Tomography mechanical ventilation system combined with ultrasonic sensor design, realtime and effective EIT reconstruction algorithm for monitoring respiratory status, EIT-based automated ventilation system and deep learning-based EIT reconstruction algorithm. A control system study was first developed to complete a single-variablecontrolled mechanical ventilation device. Firstly, three-dimensional lung imaging was studied through single-layer electrodes. Compared to traditional algorithms, this 3D algorithm proposed in the work achieved a correlation coefficient of 0.8. In addition, the tidal volume information was extracted from the 2D and 3D reconstructed images as the feedback input of the control system to realize closed-loop control. The simulation has been implemented and the results show that the error is controlled to within 3%. In the work, an active disturbance rejection control method with excellent anti-interference ability was designed to improve the stability of the system. According to the simulation results, the rise time of the system can be improved by 12.8%. At the same time, the error of the experimental results can be reduced by 22.2%. The evaluation results show that the proposed EIT-guided MV system is a powerful tool for real-time 2D and 3D biomedical imaging. The quality of tomographic images is vital for qualitative or quantitative analysis in biomedical applications. Finally, in order to achieve high-quality conductivity imaging, two deep learning-based image reconstruction algorithms were proposed. The RP-UNet algorithm has a structural similarity of 96.99% and a correlation coefficient of 97.27%. The proposed algorithm exhibits excellent spatial resolution and better noise reduction performance in reconstructed images. In conclusion, the work presented in this thesis validates the feasibility of the proposed image reconstruction algorithm using the developed EIT-guided mechanical ventilation system. It also demonstrates the potential of EIT for biomedical research and medical devices.