Objects and instrumentsExperimental animals

The experimental animals were 12 female anaesthetised domestic pigs. Their mean weight and mean chest circumference of the subjects were 20.75 ± 2.56 kg and 57.67 ± 3.26 cm, respectively, and they were 2.5-3.0 months old. The study was approved by the Animal Welfare and Ethics Committee of the Air Force Medical University.

EIT imaging instrumentation

In this study, we employed the EC-100 pro High-performance EIT system, which was developed by the research team in collaboration with UTRON Technology Co., Ltd., Hangzhou, China [13]. The system has a maximum acquisition speed of 100 fps and a data acquisition accuracy of better than 0.01‰ within the frequency range of 10 kHz to 250 kHz [13]. The data acquisition in this study was conducted via the opposing excitation and adjacent measurement operating mode, with each data acquisition parameter set to 16 electrodes, a frame rate of 40 fps, an excitation current amplitude of 1.25 mA, and a frequency of 50 kHz.

Animal models and experimental proceduresAnesthesia and instrumentation

The animals were fasted and deprived of water for 12 h, and their body temperature, respiration, and heart rate were confirmed to be within the normal range. The anaesthetic method was as follows: The anaesthetic was induced via intravenous injection of propofol (2 mg kg− 1), zoletil® 50 (2 mg kg− 1), and xylazine (1.5 mg kg− 1) via the ear margin. Tracheal intubation and connection to a respirator (Mindray WATO EX-65) were then performed, and the anaesthetic was maintained with isoflurane gas. The respiratory rate was set at 14 breaths per minute, the tidal volume was 250 ml, the ventilation rate was 3.5 L min− 1, and the airway pressure did not exceed 35 cmH2O. In the postoperative period, a central venous catheter was first introduced into the left internal jugular vein and secured in place with sutures. Subsequently, the introducer sheath (SafeSheath II introducer system, Pressure Products, Xi’an Kunlun Medical Technology Co., Ltd.) was placed into the left external jugular vein. Finally, a 7.5 F Swan-Ganz catheter (Edwards Lifesciences, Xi’an Kunlun Medical Technology Co., The catheter (Ltd.) was introduced into the atrium of the animal via the introducer sheath and allowed to float to a branch of the pulmonary artery, with the pressure waveform at the end of the catheter monitored to ensure the balloon remained in a contracted state. The experiment was conducted under the monitoring of an animal monitor (PHILIPS) and an animal non-invasive sphygmomanometer (MINDRAY) in order to ensure the safety and well-being of the subjects. The ventilator settings were modified throughout the surgical procedure in accordance with the physiological parameters, including the electrocardiogram, blood pressure, heart rate, and oxygen saturation. Once all the experiments had been completed, the animals were euthanised using an overdose of anaesthetic (20 mg kg− 1 propofol). (Please refer to the Supplementary Material 1 for details of the physiological information data.)

Experimental protocol

The animal’s chest and back hair were shaved and subsequently treated with alcohol and scrubs. Subsequently, the electrode tape, comprising 16 electrodes (EH-PET-16-CS, UTRON Technology Co., Ltd., Hangzhou, China), was positioned between the third and fourth ribs of the pig in an equally spaced manner, with the electrodes secured by self-adhesive bandages. Subsequently, the electrodes were then connected to the EIT device, which was used for data acquisition and image monitoring. (Please refer to the Supplementary Material 2, which contains an experimental diagram.)

The following assumptions are made in this study: (1) there is no difference in the position of the lung electrodes among all experimental animals; (2) there is no difference in the metabolic ability of 10% NaCl among all experimental animals; (3) there is no difference in the degree of pulmonary embolism among all experimental animals.

As illustrated in Fig. 1, the image guardianship experiment was divided into three stages:

(1) A controlled experiment was conducted to monitor lung ventilation and perfusion images in the normal state (pre-embolism).

The EIT data were collected continuously for approximately 30 min while all physiological indexes of the animals were maintained at a stable level. This was done in order to observe the impedance and image changes associated with the pulmonary ventilation process. Subsequently, the animals were placed in a state of end-expiratory pause through the administration of a minimal dose of the respiratory inhibitor propofol (2 mg kg− 1) and the regulation of the ventilator. The vascular beat EIT data were initially collected continuously for 10 to 20 cardiac cycles, and then a hypertonic saline contrast agent (5 ml of 10% NaCl solution) was injected via the central venous catheter. Next, the alterations in the electrical impedance of the lungs were continuously collected and observed for a period of 10 to 15 s. The entire procedure was conducted for approximately 40 to 45 s, after which mechanical ventilation was reinitiated.

(2) A controlled experiment was conducted to monitor lung ventilation and perfusion images in the state of acute pulmonary embolism (post-embolism).

Similarly, EIT data were collected continuously for approximately 30 min while all physiological parameters of the animal remained stable. Subsequently, the aerostat gasbag at the end of the catheter was inflated via the aerostat gasbag console of the Swan-Ganz floating catheter until the pulmonary artery pressure waveform on the monitor was no longer discernible. As in stage (1), 10 to 20 consecutive cardiac cycles of vasopulse EIT data were collected during end-expiratory pauses, followed by 10 to 15 s of hypertonic saline EIT data. Subsequently, the aerostat gasbag at the end of the catheter was released, thereby restoring the pulmonary artery pressure waveform to its normal state.

(3) Regional identification of acute pulmonary embolism (post-embolism).

The EIT data were collected continuously for approximately 10 min while all physiological parameters of the animals remained stable. Next, the aerostat gasbag at the end of the catheter was inflated until the pulmonary artery pressure waveform disappeared. Then, 5 ml of 10% NaCl contrast agent was injected through the distal port of the Swan-Ganz floatation catheter during the end-expiratory pause. The lung area where the contrast agent flowed into (i.e., corresponding to the area of the pulmonary artery embolism) was visualised by the EIT images.

Fig. 1

Overall experimental flow chart. (A) Healthy phase (before pulmonary embolism occurs); (B) Pulmonary embolism phase (after pulmonary embolism occurs); (C) Pulmonary embolism location verification phase (after pulmonary embolism occurs)

A new method for extracting pulmonary perfusion parameters based on the pulsatile method of EITData processing and parameter extractionData Processing

The data processing procedure is comprised of two principal stages. Initially, a low-pass filter with a cutoff frequency of 5 Hz is employed to filter the boundary voltage signal, thereby obtaining measurement data that reflects the information about the changes in the lung over time. Subsequently, the Grazt consensus reconstruction algorithm for EIT (GREIT) imaging algorithm is utilised for image reconstruction [14]. The Adler group initially proposed this algorithm in 2009, subsequently applying it to pulmonary impedance imaging [14,15,16].

Accordingly, in order to examine the impact of pre- and post-pulmonary embolism alterations on conductivity, we calculated the rate of change of the average reconstructed conductivity across the entire lung based on each EIT image frame. This value represents the overall change in lung conductivity [17]. The formula is calculated as follows and is represented by the following equation:

$$ARC = (\sum\limits_^N })/N$$

(1)

Where \(\:\varDelta\:_\) denotes the rate of change related to the conductivity of the ith pixel in each frame of the EIT image, N denotes the total number of pixels, and ARC is in arbitrary units (a.u).

Following a pulmonary embolism, blood flow is primarily directed to the lung tissue on the non-embolised side. Consequently, a frame of EIT image is divided into two regions of interest (ROI), as illustrated in Fig. 2: the embolised region of interest in the embolised group (PE_ROIPE) and the non-embolised region in the embolised group of interest (PE_ROINPE). Similarly, two regions of interest were defined for the EIT images in the healthy group: the embolic region of interest in the healthy group (Normal_ROIPE) and the non-embolic region of interest in the healthy group (Normal_ROINPE). Subsequently, the ARCs corresponding to the aforementioned four regions of interest were calculated using Eq. (1) for the purpose of extracting pertinent indexes, such as perfusion, at a later stage. The selection of all regions of interest is based on a threshold of 40% of the maximum pixel value.

Parameter extraction

In this study, all beat-to-beat perfusion metrics were extracted from the waveform of the ARC. The extracted perfusion metrics were as follows: amplitude, maximum slope of the ascending segment, maximum slope of the descending segment, and area under the curve. These are illustrated in Fig. 2A. Finally, the mean value of the perfusion metrics of the consecutive M (M ≥ 5) cardiac cycles was obtained for analysis. The relevant formulae for calculating the aforementioned indexes are provided below:

Amplitude is defined as the difference between the crest and trough of the ARC waveform and is used to assess the maximum dilation of the arterial vasculature at blood perfusion pressure. In this context, “crest” denotes the peak of the wave, while “trough” denotes the bottom of the wave. Amplitude of the mth cycle is therefore given by the difference between the ARC crest and ARC trough, and is represented by the symbol (ARCcrest-ARCtrough)m. Finally, M denotes the total number of cardiac cycles. The formula is calculated as follows:

$$Amplitude = ^M } - AR}})_m})/M$$

(2)

Forward Slope is defined as the maximum slope of the rising segment of the ARC waveform, i.e. the rate of change of impedance values over time during the second half of the cardiac cycle. It assesses the volume of blood passing through the artery per unit of time (from trough to crest), and f denotes the number of frames. The formula is calculated as follows:

$$Forward\;Slope = (\sum\limits_^M \matrix), \hfill \cr f \in (trough,crest + 1)) \hfill \cr} /M$$

(3)

Negative Slope is defined as the maximum slope of the descending segment of the ARC waveform, i.e., the rate of change of impedance values over time during the first half of the cardiac cycle. It assesses the volume of blood passing through the artery per unit of time (from crest to trough). The formula is calculated as follows:

$$Negative\;Slope = (\sum\limits_^M \matrix), \hfill \cr f \in (crest,trough)) \hfill \cr} /M$$

(4)

SARC is defined as the area under the curve of the ARC waveform, which is used to assess the magnitude of total blood perfusion. The formula is calculated as follows:

$$} = (\sum\limits_^M }^}} } df)})/M$$

(5)

V/Q indicators and the methods of extraction

Furthermore, ventilation/perfusion matching metrics were extracted through the analysis of lung ventilation regions, with the objective of assessing changes before and after pulmonary embolism.

Lung ventilation was evaluated by initially identifying the trough moments as background frames in the real-time acquisition of lung EIT data across all breaths, followed by EIT imaging for a single breath cycle. Subsequently, the EIT images of lung ventilation for 10 consecutive respiratory cycles were averaged. Ultimately, a threshold value of 25% of the maximum pixel value from the averaged lung ventilation EIT images was selected, with all pixel regions exceeding this value defined as lung ventilation regions, denoted as V [18].

Lung perfusion was evaluated by initially identifying the point at which the wave crest occurred as the background frame in the real-time acquisition of lung EIT data during all end-expiratory pauses, followed by one cardiac cycle of EIT imaging. Subsequently, the lung perfusion EIT images from 10 consecutive cardiac cycles were superimposed and averaged. Subsequently, the threshold value was set at 25% of the maximum pixel value from the averaged lung perfusion EIT images. All pixel regions exceeding this threshold value were defined as lung perfusion regions, denoted as Q [18].

Ultimately, the index regions for the lungs were extracted based on the matching of regional ventilation/perfusion ratios (V/Q) [19]. The V/Q match area, that is to say the area of both ventilation and perfusion, is denoted as RV+P. The Dead Space area, that is to say the area of lung ventilation only, is denoted as RV. The Shunt area, that is to say the area of blood perfusion only, is denoted as RP. The formula is as follows:

$$\eqalign} \right.\kern-\nulldelimiterspace} Q}\;Match\% \cr & = }/( + + }) \times 100\% \cr}$$

(6)

$$\eqalign/( + + }) \times 100\% \cr}$$

(7)

$$\eqalign/( + + }) \times 100\% \cr}$$

(8)

Extraction of lung perfusion indicators based on hypertonic saline method

Figure 2B illustrates a series of lung perfusion-enhanced contrast images obtained through the systematic administration of 5 ml of a 10% NaCl solution into the animal via a central vein during an end-expiratory pause [20]. In this case, as illustrated in Fig. 2B, the onset of the dilution curve was identified as the point of saline entry into the animal, designated as P0. One cardiac cycle following this was used as the point of saline entry into the pulmonary vasculature, marked as P1. The lowest point of global impedance was then taken as the endpoint of the initial saline passage through the pulmonary circulation, labelled as P2 [20].

Ultimately, the dilution curves of the P1-P2 interval were employed for the reconstruction of the EIT image sequence and the extraction of perfusion indexes, utilising the pulmonary impedance data at the moment of P1 as a reference point [20]. On this basis, the corresponding lung perfusion regions were selected and perfusion metrics, including amplitude, maximum slope and waveform area, were extracted based on the method proposed by Nguyen et al. [21]. The extraction of lung perfusion indexes and V/Q indexes was conducted in accordance with the pulsation method, with the objective of estimating the relative distribution of lung perfusion in the animal model of acute pulmonary embolism.

Fig. 2

The division of the region of interest (black area) and the extraction of the corresponding region, ARC. (A) Schematic diagram of the extraction of ARC waveform indicators by the pulsation method, where crest indicates the peak, trough indicates the bottom, and crest + 1 indicates the second peak; (B) Schematic diagram of the extraction of ARC waveform indicators by the hypertonic saline method, where P0, P1, and P2 correspond to the perfusion points at different times during the hypertonic saline method

Data analysis and statistics

The EIT data acquisition and online image reconstruction were implemented by the EC-100 pro system. The offline data analysis and image reconstruction were conducted in the Matlab R2022b (Mathworks, Natick, Massachusetts, USA) environment, based on EIDORS V3.10, with the imaging algorithm GREIT [22,23,24].

The pulmonary embolism models of 12 pigs were all successful, resulting in a total of 36 sets of valid data. These included 12 sets of control monitoring in the healthy state, 12 sets of control monitoring in the pulmonary embolism state, and 12 sets of monitoring in the validation state. The main effects were the PE-induced changes in pulmonary perfusion evaluation indexes and V/Q. The data were subjected to statistical analysis using SPSS 27.0 software. Given that the measured data did not adhere to a normal distribution, a Kruskal-Wallis test was performed, with P < 0.05 was considered statistically significant [25, 26].