Open Access
Add to BibSonomy
Abstract
Optical coherence tomography (OCT) is a non-invasive imaging modality with high spatial resolution suitable for early embryonic heart imaging. However, the most commonly used OCT systems cannot provide direct 4-D imaging due to acquisition speed limitations. We proposed a retrospective gating 4-D reconstruction method based on spectral domain OCT. A special circuit was designed to measure the impedance change of chick embryos in response to the heart beating. The impedance signal was acquired simultaneously with the OCT B-scan image sequence at several different locations along the heart. The impedance signal was used as a gating for 4-D reconstruction. The reconstruction algorithm includes cardiac period calculation, interpolation from multi-cardiac cycle image sequence into one cardiac cycle, and cardiac phase synchronization among the different locations of the heart. The synchronism of the impedance signal change with the heartbeat was verified. Using the proposed method, we reconstructed the cardiac outflow tract (OFT) of chick embryos at an early stage of development (Hamburger–Hamilton stage 18). We showed that the reconstructed 4-D images correctly captured the dynamics of the OFT wall motion.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
In the early stage of development, the embryonic heart undergoes a complex morphogenetic process [1,2]. This critical process is a coordination of many events where small deviations can lead to abnormal development and congenital heart defects (CHDs) [3–8]. Though strong evidence exists to suggest that cardiac function and biophysical forces are important factors affecting the development, the inner mechanism is poorly understood [9]. 4-D imaging (imaging of three-dimensional (3-D) structures over time) can provide image sets of the heart built up over many heart cycles, which can then be used to calculate various anatomical and physiological parameters. Thus, it is critical for the mechanism investigation. However, 4-D imaging of embryonic heart is challenging due to the diminutive size (typically, 2 mm) and the rapid cardiac motion (around two heartbeats per second) [10,11].
OCT uses back-reflected near-infrared light to image internal structures of the sample noninvasively. OCT offers cross sectional image up to a depth of ∼2-3mm in tissue with high resolution. Compared to other non-invasive imaging modalities, such as confocal microscopy and high-frequency ultrasound, OCT combines the high-resolution (micrometer scale) capability of confocal microscopy with the large penetration depth (millimeter scale) capability of ultrasound [12–14]. Furthermore, OCT can measure blood flow velocity by calculation of Doppler frequency shift, i.e. Doppler OCT. Thus, OCT is suitable for imaging the embryonic heart at early developmental stages. OCT has been used to image the embryonic hearts of chick, frog, and mouse [15–18].
Though OCT has many advantages, the acquisition speed of commercial OCT is insufficient for 4-D imaging of embryonic heart in vivo. The direction solution to achieve 4-D imaging is to improve the imaging speed of the OCT system, i.e. the acquisition speed is fast enough for volumetric imaging and the motion of heart beat is negligible [19–21]. W. Wieser et al. demonstrated a Fourier domain mode-locked (FDML) laser based OCT system with an A-line rate of 20MHz [22]. This extraordinary speed increase can enable direct 4-D imaging with sufficient spatiotemporal sampling. Direct 4-D OCT imaging is attractive for several reasons, including decreased acquisition time, reduced post processing, and the ability to capture nonperiodic events like arrhythmias [23]. However, an OCT system with this acquisition speed is very expensive. The corresponding data acquisition and handling are extremely challenging. Another strategy for 4-D imaging is prospective gating, which used a physiological signal to trigger data acquisition at specific phases of the cardiac period. In 2006, Jenkins et al. captured 4-D OCT image using an external electrical pulse to pace heartbeat and trigger data acquisition simultaneous. However, the experiment was performed on an excised heart [24]. Later Jenkins et al. performed prospective-gated OCT imaging using signals obtained by a laser Doppler velocimeter (LDV) [25]. Since the output of LDV cannot be used for triggering directly, additional setups were required, such as a R-wave detector and a field programmable gate array (FPGA). Obviously, the trigger system is complex and difficult for implementation. Currently, most of research groups investigating embryonic cardiac dynamics prefer retrospective gating scheme, which records a physiological signal simultaneously with the image data and utilizes a post processing algorithm to reconstruct 4-D image. Mariampillai et al. implemented a two-OCT system for 4-D imaging of Xenopus laevis hearts, i.e. one time domain OCT to capture a Doppler optical cardiogram used as the gating signal and the second swept source OCT for imaging [26]. Image-based retrospective gating methods obviates the need for extra signals to synchronize data (signal-free) and rearranges out-of-order data based upon image similarity. Although the system complexity is greatly reduced, the gating algorithm requires extra processing time. We can see that extracting a steady physiology signal from embryonic heart is crucial for retrospective gating technique. On the other hand, physiology signals of embryonic heart are weak and is difficult to extract (often requires extra complex setups). An easy implemented physiology signal extraction technique can promote the development of 4-D imaging on embryonic heart.
Bio-electrical impedance is a physical quantity reflecting the electrical properties of biological tissues, organs, cells, or the whole organism [27–29]. In clinic, impedance pneumography is a commonly used technique to monitor a person’s respiration rate. Electrical impedance tomography has been used as a diagnostic tool for breast cancer detection. For the beating embryonic heart, periodic pumping can cause impedance change corresponding to cardiac period. Thus, impedance signal of the embryonic heart has the potential to be used to gate 4-D imaging.
Chicken embryo has been extensively used as an animal model of cardiac development studies because it is easy to access and manipulate at early stage, the cardiovascular circulation of chicken embryo is similar to that in humans [1,30–33]. Outflow tract (OFT) is the distal region of the embryonic heart connecting the ventricle with the arterial sac, which is a crucial cardiac segment to study since a large portion of congenital heart defects originate in the OFT.
In this study, we present an impedance-signal based retrospective gating method for 4-D imaging of chicken embryonic OFT. Impedance signal of chicken embryo was extracted and acquired simultaneously with the OCT B-scans. The impedance signal was utilized in a post processing algorithm to parse the data into the appropriate phases of the cardiac period creating a 4-D data set. Two-dimension (2-D) and 3-D images at different cardiac phase were achieved. 3-D movies reproduce beating of OFT.
2. System and method
2.1 Experimental system
2.1.1 in vivo chick embryo impedance detection unit
The embryonic impedance detection was implemented with two-electrode mode, i.e. the two electrodes served as both drive electrode and receive electrode. Figure 1(a) shows the configuration of impedance detection, where ${R_P}$ (∼1 kΩ) is the protection resistances and ${C_Z}$ (0.3nF) is blocking capacitor. During the detection, the two electrodes dipped into the egg white across the embryo (on the two side of the embryo). The distance between the two electrodes was ∼5 mm according to our experience. A high-frequency ac current (IAC, 10 mA, 68kHz) was injected into the embryo through the electrodes. The ac current caused a potential difference between the two electrodes (V in Fig. 1(a)) because of the intrinsic resistivity of embryo. Figure 1(b) shows the electrical model of Fig. 1(a), where the electrode impedance is modeled as a resistor (${R_D}$) in parallel with a capacitor (${C_D}$). The electrical impedance between the two electrodes consisted of two impedance components: a relatively constant value and a varying value. The relatively constant impedance was the baseline impedance (${R_B}$). The varying impedance was corresponding to the heart beating ($\varDelta R$). The varying component of impedance generated a varying voltage component ($\varDelta V$) when current was injected. This varying voltage component was the parameter of interest because this component can be used to determine the heart beating cycle.
Fig. 1. Impedance detection circuit. (a) Arrangement of two-electrode impedance detection. (b) Electrical model of (a). IAC, drive ac current. CZ, blocking capacitor. RP, protection resistance. RD and CD, modeled electrode impedance. RB, baseline impedance. ΔR, varying impedance generated as a result of heart beating. ΔV, varying voltage caused by ΔR.
For the detection circuit, the high-frequency ac signal acted as a carrier that is amplitude-modulated by the low-frequency signal ($\varDelta V$) generated as a result of heart beating. On the receiver side, this modulated signal must be demodulated in order to extract the low-frequency heart beating signal. After demodulation, the signal was low-pass filtered to the 4Hz bandwidth level to remove unwanted noise. The demodulated and filtered output was acquired by a DAQ card (NI USB-6341). The acquisition rate is 12 kHz. Figure 2 shows the entire signal chain concept in a block diagram. The carrier at frequency ${F_C}$ was a square wave. The demodulation was accomplished with a square wave signal at the same frequency (${F_C}$). Capacitor ${C_Z}$ blocked any dc current in the detection circuit. Gain stage was used to amplify original signal on the receiver side.
Fig. 2. Conceptual block diagram of the signal chain. IAC(FC), drive ac current with frequency FC. CZ, blocking capacitor. RB, baseline impedance. ΔR, varying impedance generated as a result of heart beating.
2.1.2 Spectral domain OCT (SDOCT)
The light source of our custom-made SDOCT was a superluimnescent diode (1325 nm center wavelength, 52 nm bandwidth) offering ∼14µm axial resolution in air [34]. The light was coupled into a fiber-based Michelson interferometer via a broadband optical circulator. A fiber-based Michelson interferometer was utilized for the interference of light from reference and sample arms. The sample arm contained a collimator, an X-Y galvanometer scanner providing 2-D scan, and a 50 mm focal objective lens to achieve ∼16um lateral resolution. The optical interference signal between the light backscattering from the sample and reference arms was routed into a home-built spectrometer, which had a collimating lens (f = 50 mm), an 1145 line/mm transmission grating, and a Fourier lens (f = 100 mm) that imaged the interference spectrum into a line scan camera (1024 pixels). The spectral resolution of the spectrometer was ∼0.128 nm, resulting in an imaging depth of ∼2.95 mm in air. The maximum rate of the line scan camera was 92 kHz. The power of the light source is 10 mW and the irradiation on the sample is measured to be ∼3 mW [18]. The schematic of SDOCT system is shown in Fig. 3.
During image acquisition, the camera worked in an external trigger free run mode. The line scan rate is 66 kHz. Each B-scan consisted of 400 A-lines which were acquired continuously. The B-scan rate was 120 frames per second. There was a certain time interval between two adjacent frames. During the time interval, the B-scan data were transmitted from the camera to the computer, simultaneously, the X-scanner was driven back to the initial position for the next B-scan. Totally, 300 image sequences were acquired at different location with ∼4µm interval. Thus, scanning range in Y-direction was ∼1.2 mm, which was sufficient to cover chicken embryonic OFT (∼800um in length). Totally, the data acquisition time was ∼15 mins.
2.1.3 Synchronous acquisition of OCT and the impedance signal
In order to realize 4-D reconstruction, impedance signal was used for gating. Synchronous acquisition of OCT image and impedance signal was required. This was realized via an analog output (AO) card (NI, PCI-6713). For each location, counter (CT0) of AO card generated a pulse. This pulse triggered analog outputs (AO card) and impedance signal acquisition (DAQ card) simultaneously. The AO card outputted three waveforms: channel AO0 outputted a sawtooth wave to control X-scan; channel AO1 outputted a phase delayed squire wave to trigger camera acquisition; channel AO2 outputted a constant voltage to keep Y-scan stationary. The output voltage of AO2 changed according to different locations. Figure 4 shows the sequence diagram of synchronous acquisition.
2.2 Baseline drift correction of the impedance signal
For the impedance signal detection unit, a low-pass filter was adopted to remove high frequency disturbance from environment. Thus, the waveform of acquired impedance signal was smooth. On the other hand, the waveform was suffered the disturbance of baseline drift (Fig. 5 (a)). A clean drift-free signal will make the subsequent reconstruction results more accurate. We corrected the baseline drift with two steps [34]. Firstly, entire signal was shifted, point-by-point, in accordance with the median value ($\overline S $) calculated by overall average. Secondly, a second-degree polynomial fitting was performed on the shifted signal to acquire the central tendency (${\widehat x_k}$). Eventually, baseline drift was eliminated by:
where ${x_k}$ is the original impedance signal. Figure 5 shows the baseline drift correction result.Fig. 5. Baseline drift correction of impedance signal. (a) Original signal. (b) Signal curve was shifted with the median value. (c) Baseline drift correction result. Green line, median value. Red line, central tendency.
2.3 Post-acquisition synchronization algorithm
With the proposed data acquisition strategy, we achieved OCT B-scan sequence and corresponding synchronous impedance signal at different OFT location. To realize accurate 4-D reconstruction, the following steps are performed (Fig. 6):
- 1. Cardiac period (CP) calculation at each location based on impedance signal;
- 2. Interpolation of acquired multi-cardiac-period images into one cardiac period image sequence;
- 3. Absolute phase shift determination of each interpolated image sequence;
- 4. Phase synchronization and 4-D image reconstruction.
Fig. 6. Process of 4-D reconstruction algorithm. (a-c) Impedance signals with baseline drift correction (a), normalization (b) and synchronization (c). (d) Typical B-scan image of embryo. (e-g) OCT M-mode images extracted form original B-scan sequence (e), normalized 3-D image data set (f) and synchronized 3-D image data set(g). (h) 4D reconstruction of chick embryo. Yellow line, position of M-mode (e) on the B-scan.
2.3.1 Cardiac period calculation based on impedance signal
Basically, our 4-D reconstruction is a retrospective-gating method which relies on the periodicity of the cardiac period. Accurate cardiac period calculation is required for B-scan sequence interpolation, i.e. pooling acquired periodic B-scan sequences (including multi cardiac period) to a normalized cardiac period. 4-D image acquisition is time consuming, which takes several minutes. Cardiac period varies slightly at different location due to temperature fluctuations of chick embryo during the image acquisition. At each location (${L_i}$), we calculate cardiac period $CP({{L_i}} )$ based on the impedance signal.
To calculate the signal period, Fourier transform and multi-period averaging are methods which are easy to be implementation. However, these methods require signals contain sufficient cycles to achieve reliable results. Due to the acquisition time limitation, cardiac cycles contained in the impedance signal is insufficient. In order to increase accuracy of the cardiac period calculation, we use a string-length method (SLM) [35]. SLM is a typical algorithm for period calculation in case of relatively few observations (a few cycles).
Firstly, the original signal is smoothed and adjacent two peaks are selected. The approximate cardiac period is achieved:
where ${n_{p1}}$ and ${n_{p2}}$ is the sequence number of the two selected peaks; f is the sampling frequency (12 kHz). At each location, the acquired impedance signal is a continuous wave, if a period $CP^{\prime}({{L_i}} )$ is given, the phase (time position within one cycle) of each sample point can be determined:The correct cardiac period $CP({{L_i}} )$ is found by minimizing the length of $D({{L_i},CP^{\prime}({{L_i}} )} )$ within a certain range:
2.3.2 Image sequence and impedance signal normalization
As described in section 2.1, OCT image sequence and impedance signal were acquired simultaneously. Acquisition speed of OCT image is 120 frames per second, while impedance signal is 12 kHz. Thus, the acquisition duration of one OCT image is corresponding to 100 impedance signal points. Hence, acquisition start time (ast) of each OCT image is:
where m is the OCT image sequence number; t is the sample point time of impedance signal. Using the calculated cardiac period ($CP({{L_i}} )$), phase of each OCT image (${p_{im}}$)can be determined (Eq. (3)). Sorted by ${p_{im}}$ ascending, OCT image sequence is rearranged into one cardiac period. The rearranged phase is not evenly distributed. We resample the rearranged image sequence into a new sequence with 200 evenly distributed phases, which is realized by cubic spline interpolation. The result is a normalized one cardiac period 3-D (x, z, +phase) volume. Similar operation is performed on the corresponding impedance signal, i.e. sampling point rearrangement according to phase ascending and rearranged points resampling. Also, the normalized impedance signal contains 200 evenly distributed phases. Thus, we realize direct correspondence between normalized OCT images and impedance signal points, i.e. one image corresponds to one point.Figure 6 shows the process of normalization and synchronization for both impedance signal and OCT images. Figure 6(a) shows the impedance signal after baseline correction, which is composed of 36,000 sampling points (12,000 Hz (acquisition speed) × 3 second (acquisition time)). The signal experienced ∼4 cardiac cycles. Then, the signal was rearranged into one cardiac cycle and resampled into 200 evenly distributed phases (Fig. 6(b)). Figure 6(e) shows the M-mode image extracted from the corresponding OCT B-scan image sequence at certain depth (z = constant, yellow line in Fig. 6(d)). The M-mode image contained 360 lines corresponding to 360 images. After normalization, the image sequence was also rearranged into one cardiac cycle. We extracted M-mode image from the same depth and showed in Fig. 6(f) (200 lines corresponding to 200 phases).
2.3.3 Phase synchronization among different locations and 4-D reconstruction
At each location, data (image and impedance) acquisition starts at random cardiac phase. Phase synchronization among different location is essential for 4-D reconstruction. Here, we synchronize the phase with the normalized impedance signal (NIS). At the first location, NIS is circular-shifted to a state that NIS starts with point of minimum intensity (NIS1). The corresponding normalized image sequence is also circular-shifted according to the NIS. For other locations (${L_c}$), signal similarity between $NI{S_1}$ and circular-shifted current NIS ($NI{S_{{L_c}}}$) is calculated:
We then assembled 2-D images into 3-D image datasets at the same phases and reconstructed 4-D images of the OFT. Figure 6(c) shows the circular-shifted impedance signal which starts at minimum intensity. The corresponding image sequence was also circular-shifted simultaneously with impedance signal. The M-mode image was extracted (section 2.3.2) and shown in Fig. 6(g).
3. Experiment and results
3.1 Embryo preparation
Fertilized White Leghorn chicken eggs were incubated at 37.5° and 80% humidity until they reached the desired developmental stage. The eggshell was carefully opened from the blunt end to establish optical beam access to the chicken heart, and a small portion of the inner chorionic membrane was carefully removed with tweezers. For physiological measurement, each egg was placed in an electronic water bath for manipulation in all experiments.
3.2 Synchronism verification between changes of structure and impedance signal
For the proposed 4-D reconstruction method, a basic assumption was that the measured impedance signal changes corresponding to the heart beating. We verified the synchronism by comparing two cardiac periods calculated from structure (OCT image) and impedance signal. B-scan was performed iteratively for 6 seconds at a certain location ($y = {y_i}$). With 100 frame per second acquisition rate, we achieved an image sequence which contain 100*6 images. For the acquired image sequence, M-mode image was extracted along a horizontal line with $z = {z_k}$ (blue line in Fig. 7(a)). Figure 7(b) shows the extracted M-mode image. Two boundaries of the myocardial wall (in M-mode image) were segmented with edge detection algorithm (red lines in Fig. 7(b)). Here, the distance between the two boundaries was used to represent the OFT structure change, i.e. structure change caused by heart beating was corresponding to the periodical boundary distance change. Figure 7(c) shows the normalized boundary distance and impedance signal. We can see that the two curves are in good sync. Using the algorithm presented in section 2.3.1, accurate cardiac periods can be calculated with different signal, (i.e. CPI for impedance signal and CPS for boundary distance signal). In total, 10 image sequences were acquired at different location and the calculated result is shown in Table.1. At each location, the computed CPI and CPS were very close, with a maximum deviation of 0.003 seconds (relative deviation of 0.48%). The impedance change of chicken embryo was synchronous with the structure change which means the impedance signal can be used for 4-D reconstruction.
Table 1. Calculated cardiac period from impedance signal and structure in M-mode image. (r > 0.99)
Fig. 7. Synchronism verification between changes of structure and impedance signal. (a) M-mode image extraction. (b) Distance calculation between two myocardial wall Boundary in M-mode image. (c) Comparison of synchronous acquired impedance signal and OFT size change. Black line, OFT size (boundary distance (b)). Red line, impedance signal.
3.3 4-D reconstruction
Using the proposed system, 4-D image acquisition was performed on a chick embryonic OFT of stage HH18 in vivo. Simultaneously, impedance variation of the embryo was detected and recorded at each location. After 4-D reconstruction, we obtained 3-D transient structure of OFT of 200 phases. Figure 8 compare the structure of OFT at four phases, including minimum impedance, mid-value in impedance rising, maximum impedance, and mid-value in impedance falling (the first column). The second column of Fig. 8 shows the top view of the chicken embryo OFT. Comparing the first and second columns, it can be seen that the minimum impedance corresponds to the OFT contraction and the maximum impedance corresponds to OFT expansion. To further show the blood filling of OFT in different phases, we further erased myocardial wall and jelly of the OFT. Thus, only perfused blood was left (the third column of Fig. 8). Figure 8(i) shows that there seems to be more blood in OFT at the contraction state, which is an illusion caused by the observation angle. In fact, there was not much blood left in OFT at the contraction state, which was clearly shown on the cross-sectional image (Fig. 8(q), fifth column). The fourth column shows the shape of OFT under free angle, which can better display the twist structure of OFT. Overall, the blood flow of OFT started from the ventricular end and flowed to the arterial end, causing the sequential expansion of OFT at different position. Two videos were made to show reconstructed beating process of OFT. In the first video, we erased all the tissues around the OFT and showed OFT beating at a free angle. In the second video, we only erased tissues above OFT and observed OFT from top view. Periodical OFT beating can be seen in both videos. In the second video, we observed periodical blood flow in artery.
Fig. 8. 4-D reconstruction result of chick embryonic OFT (see Visualization 1 and Visualization 2). (a-d) Corresponding phase on impedance signal. (e-h) Top view of the OFT. (i-l) Top view of blood perfusion in the OFT. (m-p) View in free angle of the OFT. (q-t) Cross-section image of OFT (the position is corresponding to the blue line in (i)).
Figure 9 shows the change of OFT structure with pulsation in three different sections (the first column). Sections were selected according to the structure of chicken embryo. The first row is the longitudinal section of OFT, and different forms of OFT in a cardiac cycle can be observed. The second row is the horizontal section of ventricle and inlet of OFT. The third row is the cross section of the outlet of OFT. We selected 10 typical phases (10 columns) for section display. Since OFT is an irregular twist tube, it is difficult to obtain the full feature of OFT through one section. The sections of the first and second rows are mainly inlet of OFT connected to the ventricle; the third row is outlet of OFT which has a phase lag with the previous two rows. In Fig. 9, the structure of OFT at each section is smooth without jump, indicating that our reconstruction has good accuracy. In section image, jelly, myocardial layer and blood can be distinguished, and the boundary is clear. The results demonstrate that the proposed method is accurate and 4-D reconstruction affects little on the spatial resolution. High resolution reconstructed image gives the potential for accurate physiologic measurement on OFT, such as wall thickness and flow volume. For in vivo imaging, OFT of embryonic heat is irregular in shape and beating fast. On the other hand, scan direction of OCT probing beam is limited by the optics design. Thus, it was impossible to achieve arbitrary section of living OFT by direct OCT imaging. Using 4-D reconstruction techniques, the structure of arbitrary section at different phases can be reproduced, which provides a powerful tool for embryonic development research.
Fig. 9. Arbitrary section of OFT after 4-D reconstruction. (a-c) Position of section plane; each column of a1- a10, b1- b10, c1- c10 corresponding to the same cardiac phase.
4. Discussion
In this study, we propose a retrospective gating 4-D reconstruction method based on impedance information. Compared with LDV, double OCT and other retrospective gating methods, the measurement hardware of impedance signal is relatively simple. In 2007, Jenkins et al. used LDV signal to trigger 4-D imaging of quail embryonic heart [25]. To generate a proper trigger, signal processing includes LDV signal amplification, low-pass filtering, R-wave detection to produce two triggers and FPGA based one trigger ignoring. Obviously, the signal processing is complex. Thus, LDV signal is required to be stable. Small disturbance in the environment may cause unpredictable result. The proposed method corrected the base line drift and calculated cardiac period using SLM algorithm. These operations can partially eliminate the effect of small disturbance, making the reconstruction result reliable. For Doppler OCT based gating, the main problem is the increased Doppler noise floor that accompanies the higher flow velocity. Multi A-line averaging is a solution for noise floor reduction. However, averaging operation reduces the temporal resolution of the gating signal [26]. The proposed method measured impedance between the two electrodes as the gating signal. The impedance is mainly determined by inner structure change due to heart beating and blood flow velocity affected little on the signal. The impedance signal can provide high temporal resolution for 4-D reconstruction gating. At the same time, the heart of chicken embryo is very small, and the impedance change caused by its pulsation is very weak. Therefore, we specially designed a high-gain amplifier circuit, and the gain can be adjusted within a certain range to reduce the impact of individual differences on the impedance measurement. The designed impedance measurement unit is small in size and low in cost. The impedance measurement only requires two electrodes on both sides of the chicken embryo, which does not block the optical path and can be conveniently combined with the OCT imaging system. While amplifying the impedance signal, the high gain amplifier circuit is sensitive to external interference, especially vibration. For this reason, our samples were placed on the air floating platform, and the laboratory was kept quiet during the data acquisition.
For retrospective gating-related methods, a physiological signal of the embryonic heart is required for 4-D reconstruction generally. The physiological signal is used to parse OCT image data into the appropriate phases of the cardiac cycle, creating a 4-D data set. In recent years, image-based retrospective gating (or signal-free, i.e., not requiring an external signal) has become predominant for 4-D OCT imaging in developmental cardiology studies [36,37]. Obviously, the system complexity is greatly reduced, on the other hand, the gating algorithm requires extra processing time. The proposed method utilizes easily accessible hardware (impedance measurement circuit and DAQ card) and relatively simply algorithm, which is easy to implement. Furthermore, the extracted impedance signal has the potential to be used as a trigger for prospective gating, which can further simplify the algorithm of 4-D reconstruction and reduce data processing time.
At the early developing stage, myocardium and endocardium layers of the embryonic heart tube are not in direct contact with each other. Between them, there is a filling substance, cardiac jelly (CJ) [38,39]. Previous papers had demonstrated that CJ acts as a hydraulic skeleton facilitating end-systolic occlusion of the embryonic heart lumen to prevent the backflow of blood [40]. Transient shape of CJ at different cardiac phase is valuable for related studies. In this paper, the impedance signal is synchronously acquired as the gating, and clear 3-D images of different phases are reconstructed. Based on this, we can extract CJ by the image processing and analyze its changes with blood flow. Follow-up research will carry out related work.
In order to measure the impedance signal, a high frequency ac current was injected into the sample. The amplitude of the injected current was controlled to be small. The current is radio frequency signal (68 kHz). Due to the “skin effect”, radio frequency current flux can only penetrate a limited depth of the conductor surface [41]. The result is that most of the supply current flows through the surface of the sample (egg white) and only a small amount flows through the embryo. Thus, the injected current has little effect on heart rate. The theory analysis was supported by our experiment. We measured heart rates with and without current injection, and found no static differences.
In order to reconstruct the 4-D image, we acquired continuous B-scans at multiple locations. Data acquisition was time consuming (several minutes). During the acquisition, the chicken embryo floated on the surface and was exposed to the air, while the temperature of the environment was lower than the chicken embryo incubation temperature. Temperature fluctuation caused changes in cardiac cycle. The data in Table 1 were continuously acquired from different positions of the same embryo. It can be seen that, the cardiac period of the embryo increased slightly with the time. At early development stages, the central neural system of chick embryos has not yet developed, the heart rate is only a function of embryo temperature. The heart rate decreases as temperature drops [42]. In order to maintain similar physiological activities of chicken embryo, we used an automatic heating device to hold the temperature steady. The heating device kept the temperature fluctuation within 1°C to minimize the heart rate change. Increasing the data acquisition speed is another strategy to decrease the effect of temperature fluctuation on 4-D reconstruction. The proposed method contained normalization operation at each location, which can also reduce the influence of cardiac period fluctuations.
5. Conclusion
In this study, an impedance measurement circuit was designed to detect impedance change of chick embryo according to the heart beating. Impedance signal and OCT images were recorded simultaneously at each location of chick embryo OFT. The impedance signal is utilized as the gating signal for 4-D image reconstruction. Using the proposed method, we obtained 3-D images of OFT at 200 different phases. 2-D images of OFT of arbitrary sections at different phases were presented. The results showed that the proposed method is a valuable tool for studying early embryonic heart development.
Funding
National Natural Science Foundation of China (61771119, 61901100, 62075037); Natural Science Foundation of Hebei Province (E2020501029, F2019501132).
Disclosures
The authors declare that there are no conflicts of interest related to this article.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. A. F. Moorman and V. M. Christoffels, “Cardiac chamber formation: development, genes, and evolution,” Physiol. Rev. 83(4), 1223–1267 (2003). [CrossRef]
2. E. M. Olson, “A decade of discoveries in cardiac biology,” Nat. Med. 10(5), 467–474 (2004). [CrossRef]
3. A. Gittenberger-de Groot, M. Bartelings, M. Deruiter, and R. Poelmann, “Basics of cardiac development for the understanding of congenital heart malformations,” Pediatr. Res. 57(2), 169–176 (2005). [CrossRef]
4. B. Hogers, M. C. DeRuiter, A. C. Gittenberger-de Groot, and R. E. Poelmann, “Unilateral vitelline vein ligation alters intracardiac blood flow patterns and morphogenesis in the chick embryo,” Circ. Res. 80(4), 473–481 (1997). [CrossRef]
5. B. C. Groenendijk, B. P. Hierck, J. Vrolijk, M. Baiker, M. J. Pourquie, A. C. Gittenberger-de Groot, and R. E. Poelmann, “Changes in shear stress-related gene expression after experimentally altered venous return in the chicken embryo,” Circ. Res. 96(12), 1291–1298 (2005). [CrossRef]
6. D. Srivastava and E. N. Olson, “A genetic blueprint for cardiac development,” Nature 407(6801), 221–226 (2000). [CrossRef]
7. J. N. Topper and M. A. Gimbrone Jr, “Blood flow and vascular gene expression: fluid shear stress as a modulator of endothelial phenotype,” Mol. Med. Today 5(1), 40–46 (1999). [CrossRef]
8. J. Stypmann, “Doppler ultrasound in mice,” Echocardiography 24(1), 97–112 (2007). [CrossRef]
9. L. M. Peterson, M. W. Jenkins, S. Gu, L. Barwick, M. Watanabe, and A. M. Rollins, “4D shear stress maps of the developing heart using Doppler optical coherence tomography,” Biomed. Opt. Express 3(11), 3022–3032 (2012). [CrossRef]
10. J. Männer, “Cardiac looping in the chick embryo: a morphological review with special reference to terminological and biomechanical aspects of the looping process,” Anat. Rec. 259(3), 248–262 (2000). [CrossRef]
11. N. Hu and E. B. Clark, “Hemodynamics of the stage 12 to stage 29 chick embryo,” Circ. Res. 65(6), 1665–1670 (1989). [CrossRef]
12. M. Liebling, A. S. Forouhar, R. Wolleschensky, B. Zimmermann, R. Ankerhold, S. E. Fraser, M. Gharib, and M. E. Dickinson, “Rapid three-dimensional imaging and analysis of the beating embryonic heart reveals functional changes during development,” Dev. Dyn. 235(11), 2940–2948 (2006). [CrossRef]
13. K. Tobita, J. B. Garrison, L. J. Liu, J. P. Tinney, and B. B. Keller, “Three-dimensional myofiber architecture of the embryonic left ventricle during normal development and altered mechanical loads,” Anat. Rec. A Discov. Mol. Cell. Evol. Biol. 283A(1), 193–201 (2005). [CrossRef]
14. C. K. Phoon, “Imaging tools for the developmental biologist: ultrasound biomicroscopy of mouse embryonic development,” Pediatr. Res. 60(1), 14–21 (2006). [CrossRef]
15. Z. Ma, Y. Ma, Y. Zhao, J. Liu, J. Liu, and L. Wang, “Measurement of the absolute velocity of blood flow in early-stage chick embryos using spectral domain optical coherence tomography,” Appl. Opt. 56(31), 8832–8837 (2017). [CrossRef]
16. R.C. Haskell, M. E. Williams, D. C. Petersen, B. M. Hoeling, A. J. Schile, J. D. Pennington, M. G. Seetin, J. M. Castelaz, S. E. Fraser, C. Papan, H. Ren, J. F. de Boer JF, and Z. Chen, “Visualizing early frog development with motion-sensitive 3-D optical coherence microscopy,” Conf. Proc. IEEE Eng. Med. Biol. Soc. 2004, 5296–5299 (2004).
17. S. H. Syed, K. V. Larin, M. E. Dickinson, and I. V. Larina, “Optical coherence tomography for high-resolution imaging of mouse development in utero,” J. Biomed. Opt. 16(4), 046004 (2011). [CrossRef]
18. Z. Ma, A. Liu, X. Yin, A. Troyer, K. Thornburg, R. K. Wang, and S. Rugonyi, “Measurement of absolute blood flow velocity in outflow tract of HH18 chicken embryo based on 4D reconstruction using spectral domain optical coherence tomography,” Biomed. Opt. Express 1(3), 798–811 (2010). [CrossRef]
19. R. A. Whipple, M. A. Matrone, E. H. Cho, E. M. Balzer, M. I. Vitolo, J. R. Yoon, O. B. Ioffe, K. C. Tuttle, J. Yang, and S. S. Martin, “Epithelial-to-mesenchymal transition promotes tubulin detyrosination and microtentacles that enhance endothelial engagement,” Cancer Res. 70(20), 8127–8137 (2010). [CrossRef]
20. J. M. Boyd, L. Huang, L. Xie, B. Moe, S. Gabos, and X. Li, “A cell-microelectronic sensing technique for profiling cytotoxicity of chemicals,” Anal. Chim. Acta. 615(1), 80–87 (2008). [CrossRef]
21. C. Khanna, X. Wan, S. Bose, R. Cassaday, O. Olomu, A. Mendoza, C. Yeung, R. Gorlick, S. M. Hewitt, and L. J. Helman, “The membrane-cytoskeleton linker ezrin is necessary for osteosarcoma metastasis,” Nat. Med. 10(2), 182–186 (2004). [CrossRef]
22. W. Wieser, T. Klein, D. C. Adler, F. Trépanier, C. M. Eigenwillig, S. Karpf, J. M. Schmitt, and R. Huber, “Extended coherence length megahertz FDML and its application for anterior segment imaging,” Biomed. Opt. Express 3(10), 2647–2657 (2012). [CrossRef]
23. Z. Al-Qazwini, Z. Ko, K. Mehta, and N. Chen, “Ultrahigh-speed line-scan SD-OCT for four-dimensional in vivo imaging of small animal models,” Biomed. Opt. Express 9(3), 1216–1228 (2018). [CrossRef]
24. M. W. Jenkins, F. Rothenberg, D. Roy, V. P. Nikolski, Z. Hu, M. Watanabe, D. L. Wilson, I. R. Efimov, and A. M. Rollins, “4D embryonic cardiography using gated optical coherence tomography,” Opt. Express 14(2), 736–748 (2006). [CrossRef]
25. M. W. Jenkins, O. Q. Chughtai, A. N. Basavanhally, M. Watanabe, and A. M. Rollins, “In vivo gated 4D imaging of the embryonic heart using optical coherence tomography,” J. Biomed. Opt. 12(3), 030505 (2007). [CrossRef]
26. A. Mariampillai, B. A. Standish, N. R. Munce, C. Randall, G. Liu, J. Y. Jiang, A. E. Cable, I. A. Vitkin, and V. X. Yang, “Doppler optical cardiogram gated 2D color flow imaging at 1000 fps and 4D in vivo visualization of embryonic heart at 45 fps on a swept source OCT system,” Opt. Express 15(4), 1627–1638 (2007). [CrossRef]
27. A. S. Khalafalla, S. P. Stackhouse, and O. H. Schmitt, “Thoracic impedance gradient with respect to breathing,” IEEE Trans. Biomed. Eng. BME-17(3), 191–198 (1970). [CrossRef]
28. W. G. Kubicek, E. Kinnen, and A. Edin, “Calibration of an impedance pneumograph,” J. Appl. Physiol. 19(3), 557–560 (1964). [CrossRef]
29. J. M. Ernst, D. A. Litvack, D. L. Lozano, J. T. Cacioppo, and G. G. Berntson, “Impedance pneumography: noise as signal in impedance cardiography,” Psychophysiology 36(3), 333–338 (1999). [CrossRef]
30. V. M. Christoffels, P. E. Habets, D. Franco, M. Campione, F. de Jong, W. H. Lamers, Z. Z. Bao, S. Palmer, C. Biben, R. P. Harvey, and A. F. Moorman, “Chamber formation and morphogenesis in the developing mammalian heart,” Dev. Biol. 223(2), 266–278 (2000). [CrossRef]
31. E. B. Clark, N. Hu, J. L. Dummett, G. K. Vandekieft, C. Olson, and R. Tomanek, “Ventricular function and morphology in chick embryo from stages 18 to 29,” Am. J. Physiol. 250(3 Pt 2), H407–H413 (1986). [CrossRef]
32. D. Sedmera, T. Pexieder, V. Rychterova, N. Hu, and E. B. Clark, “Remodeling of chick embryonic ventricular myoarchitecture under experimentally changed loading conditions,” Anat. Rec. 254(2), 238–252 (1999). [CrossRef]
33. K. Tobita, E. A. Schroder, J. P. Tinney, J. B. Garrison, and B. B. Keller, “Regional passive ventricular stress-strain relations during development of altered loads in chick embryo,” Am. J. Physiol. Heart Circ. Physiol. 282(6), H2386–H2396 (2002). [CrossRef]
34. V. S. Chouhan and S. S. Mehta, “Total removal of baseline drift from ECG signal,” Int. Conf. Comput.: Theory Appl. IEEE, 2007.
35. D. H. Roberts, J. Lehar, and J. W. Dreher, “Time series analysis with clean. I. Derivation of a spectrum,” Astron. J. 93(4), 968–989 (1987). [CrossRef]
36. R. F. Stellingwerf, “Period determination using phase dispersion minimization,” Astrophys. J. 224, 953–960 (1978). [CrossRef]
37. L. Wang, Y. Wang, S. Guo, J. Zhang, M. Bachman, G. P. Li, and Z. Chen, “Frequency domain phase-resolved optical Doppler and Doppler variance tomography,” Opt. Commun. 242(4-6), 345–350 (2004). [CrossRef]
38. A. Nakamura and F. J. Manasek, “An experimental study of the relation of cardiac jelly to the shape of the early chick embryonic heart,” J. Embryol. Exp. Morphol. 65, 235–256 (1981).
39. Y. Nakajima, M. Morishima, M. Nakazawa, K. Momma, and H. Nakamura, “Distribution of fibronectin, type I collagen, type IV collagen, and laminin in the cardiac jelly of the mouse embryonic heart with retinoic acid-induced complete transposition of the great arteries,” Anat. Rec. 249(4), 478–485 (1997). [CrossRef]
40. A. Barry, “The functional significance of the cardiac jelly in the tubular heart of the chick embryo,” Anat. Rec. 102(3), 289–298 (1948). [CrossRef]
41. H. A. Wheeler, “Formulas for the Skin Effect,” Proc. IRE 30(9), 412–424 (1942). [CrossRef]
42. J. Wispé, N. Hu, and E. Clark, “Effect of environmental hypothermia on dorsal aortic blood flow in the chick embryo, stages 18 to 24,” Pediatr. Res. 17(12), 945–948 (1983). [CrossRef]
