Reconstruction of Organ Boundaries With Deep Learning in the D-Bar Method for Electrical Impedance Tomography.

OBJECTIVE: Medical electrical impedance tomography is a non-ionizing imaging modality in which low-amplitude, low-frequency currents are applied on electrodes on the body, the resulting voltages are measured, and an inverse problem is solved to determine the conductivity distribution in the region of interest. Due the ill-posedness of the inverse problem, the boundaries of internal organs are typically blurred in the reconstructed image. METHODS: A deep learning approach is introduced in the D-bar method for reconstructing a 2-D slice of the thorax to recover the boundaries of organs. This is accomplished by training a deep neural network on labeled pairs of scattering transforms and the boundaries of the organs in the data from which the transforms were computed. This allows the network to "learn" the nonlinear mapping between them by minimizing the error between the output of the network and known actual boundaries. Further, a "sparse" reconstruction is computed by fusing the results of the standard D-bar reconstruction with reconstructed organ boundaries from the neural network. RESULTS: Results are shown on simulated and experimental data collected on a saline-filled tank with agar targets simulating the conductivity of the heart and lungs. CONCLUSIONS AND SIGNIFICANCE: The results demonstrate that deep neural networks can successfully learn the mapping between scattering transforms and the internal boundaries of structures.

Estimating regions of air trapping from electrical impedance tomography data.

OBJECTIVE: Electrical impedance tomography (EIT) has been shown to be a viable non-invasive, bedside imaging modality to monitor lung function. This paper introduces a method for identifying regions of air trapping from EIT data collected during tidal breathing and breath-holding maneuvers. APPROACH: Ventilation-perfusion index maps are computed from dynamic EIT images. These maps are then used to identify regions of air trapping in the area of the lung as regions that are poorly ventilated but well perfused throughout the breathing and cardiac cycles. These EIT-identified regions are then compared with independently identified regions of low attenuation, or air trapping, on chest CT. Results of this method are demonstrated in two children with cystic fibrosis and on a healthy control subject. MAIN RESULTS: In both CF children, the EIT-identified regions of air trapping matched the regions indicated from the chest CT. The EIT-based method is only validated with CT scans within 4 cm of the chest cross-section defined by the electrode plane. SIGNIFICANCE: The results indicate the potential use of EIT-derived ventilation-perfusion index maps as a non-invasive method for identifying regions of air trapping.

Evaluation of surrogate measures of pulmonary function derived from electrical impedance tomography data in children with cystic fibrosis.

OBJECTIVE: Lung function monitoring by spirometry plays a critical role in the clinical care of pediatric cystic fibrosis (CF) patients, but many young children are unable to perform spirometry, and the outputs are often normal even in the presence of lung disease. Measures derived from electrical impedance tomography (EIT) images were studied for their utility as potential surrogates for spirometry in CF patients and to assess response to intravenous antibiotic treatment for acute pulmonary exacerbations (PEx) in a subset of patients. APPROACH: EIT data were collected on 35 subjects (21 with CF, 14 healthy controls, 8 CF patients pre- and post-treatment for an acute PEx) ages 2 to 20 years during tidal breathing and also concurrently with spirometry on subjects over age 8. EIT-derived measures of FEV1, FVC, and FEV1/FVC were computed globally and regionally from dynamic EIT images. MAIN RESULTS: Global EIT-derived FEV1/FVC showed good correlation with spirometry FEV1/FVC values (r = 0.54, p = 0.01), and were able to distinguish between the groups (p = 0.01). Lung heterogeneity was assessed through the spatial coefficient of variation (CV) of EIT difference images between key time points, and the CVs for EIT-derived FEV1 and FVC showed significant correlation with the CV for tidal breathing (r = 0.47, p = 0.01 and r = 0.50, p = 0.01, respectively). Global EIT-derived FEV1/FVC was better able to distinguish between groups than spirometry FEV1 (F-values 776.5 and 146.3, respectively, p < 0.01.) The same held true for the CVs for EIT-derived FEV1, FVC, and tidal breathing (F-values 215.93, 193.89, 204.57, respectively, p < 0.01). SIGNIFICANCE: The strong correlation between the CVs for tidal breathing, FEV1, and FVC, and the statistically significant ability of CV for tidal breathing to distinguish between healthy subjects and CF patients, and between the studied CF disease states suggests that the CV may be useful for measuring the extent and severity of structural lung disease.