Separation of lung and heart functions in electrical impedance tomography using semi-Siamese U-Net

Abstract

Electrical impedance tomography (EIT) is widely used for bedside monitoring of lung ventilation status Its goal is to reflect the internal conductivity changes and estimate the electrical properties of the tissues in the thorax However poor spatial resolution affects EIT image reconstruction to the extent that the heart and lung-related impedance images are barely distinguishable Several studies have attempted to tackle this problem and approaches based on decomposition of EIT images using linear transformations have been developed Recently U-Net has become a prominent architecture for semantic segmentation In this paper we propose a novel semi-Siamese U-Net specifically tailored for EIT application It is based on the state-of-the-art U-Net whose structure is modified and extended forming shared encoder with parallel decoders and has multi-task weighted losses added to adapt to the individual separation tasks The trained semi-Siamese U-Net model was evaluated with a test dataset and the results were compared with those of the classical U-Net in terms of Dice similarity coefficient and mean absolute error Results showed that compared with the classical U-Net semi-Siamese U-Net exhibited performance improvements of 11 37% and 3 2% in Dice similarity coefficient and 3 16% and 5 54% in mean absolute error in terms of heart and lung-impedance image separation respectively EIT is a non-invasive technique that constitutes a promising tool for the real-time imaging and long-term monitoring of the ventilation distribution at bedside However clinical monitoring and diagnostic evaluations depend on various methods to assess ventilation-dependent parameters useful for ventilation therapy This study developed an automatic robust and rapidly accessible method for lung segmentation that can be used to define appropriate regions-of-interest (ROIs) within EIT images Approach: To date available methods for patients with lung defects have the disadvantage of not being able to identify lung regions because of their poor ventilation responses Furthermore the challenges related to the identification of lung areas in EIT images are attributed to the low spatial resolution of EIT In this study a U-Net-based automatic lung segmentation model is used as a postprocessor to transform the original EIT image to a lung ROI image and refine the inherent conductivity distribution of the original EIT image It is different from conventional approaches in that the trained U-Net network can perform an automatic segmentation of conductivity changes in EIT images without requiring prior information Main results: The experimental design of this study was based on a finite element method phantom used to assess the feasibility and effectiveness of the proposed method and the evaluation of the trained models on the test dataset was performed using the Dice similarity coefficient and the mean absolute error Significance: The use of a deep-learning-based approach attained automatic and convenient segmentation of lung ROIs into distinguishable images which represents a direct benefit for regional lung ventilation-dependent parameter extraction and analysis However further investigations and validation are warranted in real human datasets with different physiology conditions with CT cross-section dataset to refine the suggested model

Date of Award	2021
Original language	English
Supervisor	Kuo-Sheng Cheng (Supervisor)