4.1.1.

Gated image acquisition

Due to the longer acquisition time of TDOCT systems, it was well known that functional imaging of a beating heart can cause motion artifacts. To minimize these artifacts, gating was introduced. Defined as image acquisition synchronized to the heart cycle, gated cardiac imaging was able to provide sufficient data to produce 3-D images of the beating heart without motion artifacts, thus helping in evaluating and analyzing parameters, such as ejection fraction and stroke volume. Jenkins et al.⁸⁹ demonstrated the first combination of direct gating and OCT for four-dimensional (4-D) representations of embryonic hearts in 2006, utilizing excised and externally paced chick and murine hearts. The OCT image acquisition was gated to an external stimulus, which was used to pace the heart at 1 Hz. After acquiring images over many cardiac cycles, a single 3-D image was constructed by combining all B-scans from a given phase of the cardiac cycle (explained in Fig. 4). Finally, by generating similar 3-D images at different phases, a 3-D movie was generated and 3-D measurements of the ejection fraction were made. The TDOCT system captured 4000 A-scans/second and B-scans were acquired at $20\mu \mathrm{m}$ intervals. Four frames were captured at each position and at each cardiac cycle, which were averaged to reduce noise. Capturing 16 equally spaced phases in the cardiac cycle resulted in 16 3-D volumes and the full 4-D dataset was recorded in less than 5 min. They demonstrated 4-D representations of embryonic mouse and chick hearts, digital sections of the beating heart, and visualizations of 3-D internal embryonic structures. Since this was a method of direct gating, parameters such as wall thickness, ventricular volume, and ejection fractions were quantified.

Fig. 4

Gated reconstruction: B-scans were acquired at the same position at different phases of the cardiac cycle before moving to the next position. The 3-D volumes were then reconstructed at each phase. Adapted from Ref. 89.

Extending this technology to live hearts and in vivo studies, the same group demonstrated the first gated 4-D imaging of avian embryonic hearts in vivo using TDOCT in 2007.⁹⁰ The system acquired 4000 A-scans/s with axial and lateral resolutions of $10\mu \mathrm{m}$ and used a trigger derived from a laser Doppler velocimetry signal, which was used to trigger image acquisition in lieu of an electrocardiogram used in other imaging modalities. This made the process noninvasive and easily repeatable as it did not rely on the placement of electrodes on the heart. The gated 4-D image consisted of $\sim 864,000$ A-lines accumulated over multiple heartbeats and represented 3-D volumetric images at eight phases of the cardiac cycle with a total imaging time of $\sim 7\min$. The images demonstrated the transition from systole in phase 1 to diastole in phase 8 (Fig. 5). The volumes at phase 8 and phase 1 were used to measure the ejection fraction. The OCT slices clearly resolved anatomical features, such as the myocardium and endocardium. This technique opened doors for simultaneous assessment of structural and functional development due to the increase in temporal resolution and superior image quality.

Fig. 5

Eight phases of the beating embryonic quail heart from three different orientations acquired in vivo using gated OCT. Each time series progresses from systole to diastole and each slice is separated by 95 ms. (a) En face 2-D OCT images (sagittal to the body). (b) Eight phases from the normal OCT view (coronal to the body). (c) Transverse view of the heart. Images on the far right of each row show a 3-D surface reconstruction of the heart in phase 8 (diastole). The white plane indicates the position of the corresponding 2-D OCT images in that row. Adapted from Ref. 90.

Phenotyping of transgenic embryonic murine hearts using TDOCT was demonstrated by the same group in 2007.⁹¹ Traditional histological sectioning for phenotypic analysis has a number of disadvantages including the lack of dynamic information, large number of samples required, and long processing times. Jenkins et al. used OCT to distinguish and quantify morphological differences between embryonic murine hearts from HEX1M1 mutants and their wild-type counterparts at stages E12.5 (embryonic day 12.5) and E13.5, as shown in Fig. 6. The mutants had increased ventricular volumes and thinner compact myocardia. At E13.5, the mutants showed asymmetric morphology as the left ventricle and atrium showed growth retardation. The results from this work showed the potential of OCT as a noninvasive high-throughput screening tool.

Fig. 6

(a) The 2-D en face image of HEX1M1 wild-type heart, (b) corresponding mutant heart, (c) 3-D volume from a wild-type heart, and (d) corresponding mutant heart. RA, right atrium; RV, right ventricle; LA, left atrium; LV, left ventricle; IVS, interventricular septum. Adapted from Ref. 91.

4.2.1.

Spectral-domain optical coherence tomography

Kagemann et al.⁹² utilized SDOCT to perform the first in vivo 3-D imaging of zebrafish morphology in 2008. Utilizing the noninvasive nature of OCT, they developed an imaging protocol without the use of anesthetizing agents to avoid sedative side-effects on the embryos. Embryos were imaged in three different preparations. In preparation 1, the embryos were embedded in 1% low melting agarose and immersed in E3 egg water while live OCT imaging was performed. In preparation 2, the same imaging procedure was followed at 72-h postfertilization (hpf). The embryos were then returned to E3 egg water and re-embedded in agarose at 120 hpf. Preparation 3 was similar to preparation 1 but involved nagie oko (m520 allele) mutant embryos. The SDOCT system used in this study had an axial resolution of $3.5\mu \mathrm{m}$ in air, and the embryos were imaged at different transverse spatial dimensions depending on the stage of the embryo. At a frame rate of 47 Hz, 180 sequential frames were obtained, each consisting of 501 A-scans acquired at 24 kHz. The unused embryos were fixed for histological processing after OCT imaging. After image processing and averaging to reduce noise, quantitative tissue measurements and statistical analyses were performed. A number of internal structures of the embryo were visible from the SDOCT images (Fig. 7) and were correlated with published data on anatomic sections for identification. In addition to noninvasively imaging zebrafish organogenesis of the eye, ear, and spine, this study also visualized and assessed the beating heart at different stages of development. Quantitative measurements of the heart and retinal thicknesses were shown to significantly increase with age, and there was good correspondence between the OCT images and histological sections. This investigation clearly showed the ability of SDOCT to perform longitudinal studies and distinguish developmental defects in mutant zebrafish embryos, demonstrating the potential of OCT to detect developmental changes at the tissue microstructural level.

Fig. 7

Visualization of developing internal morphology of zebrafish embryos at different developmental stages. Adapted from Ref. 92.

More recently, SDOCT was used to image spinal development and deformities in zebrafish.⁹³ Both wild-type and Chd7⁹⁴ knockdown zebrafish embryos from 5 to 29 days postfertilization (dpf) were imaged using an SDOCT system with a central wavelength of 930 nm, bandwidth of 100 nm, and axial and lateral resolutions of 7 and $8\mu \mathrm{m}$, respectively. OCT images from wild-type embryos were compared with histological sections, and the results clearly showed that OCT was capable of differentiating the spine and the surrounding muscles. Furthermore, individual spinal cells were resolved, revealing the cellular-level imaging capabilities of OCT. The highly deformed notochord and many of the malformed organs were clearly visible in the OCT images of the Chd7 knockdown zebrafish embryos. OCT was able to monitor the spinal development of zebrafish embryos noninvasively, which helped overcome the difficulty of obtaining a suitable section of the curved spine during histological sectioning. However, the limited imaging depth of OCT inhibited full body imaging of embryos older than 30 dpf. Moreover, OCT was able to identify skeletal structures that had not yet been mineralized and could not be identified with calcein staining. However, OCT could not distinguish between mature and immature bones, which is possible with calcein staining, as depicted in Fig. 8. Additionally, OCT was used to image developmental defects in zebrafish embryos due to ethanol exposure.⁹⁵ Similar to bright-field microscopy images, abnormal features, such as edema around the developing heart and an enlarged yolk sac, were observed in the OCT images of ethanol-exposed embryos when compared to unexposed embryos. Other organs, such as the eye and notochord, were also imaged to visualize the effects of ethanol on embryonic development.

Fig. 8

(a) Sagittal view of an OCT 3-D reconstruction of a 21 dpf wild type zebrafish in vivo. (b) 2-D cross-section where the spine is indicated by the white arrow. (c) Calcein stained image of the same specimen, taken an hour later. Scale bar is 1 mm. Adapted from Ref. 93.

In one of the first studies to assess the Drosophila melanogaster heart using SDOCT, Bradu et al.⁹⁶ used a combination of OCT and laser scanning fluorescence microscopy (LSFM). Due to the optical transparency of the heart and relative opacity of the insect cuticle, live in vivo imaging of the heart was difficult. SDOCT provided axial profiles, but LSFM was utilized to obtain en-face images. Thus, the hearts were not imaged consecutively with both imaging modalities, but rather sequentially. Because the SDOCT system provided depth-resolved information at once, the LSFM system was focused such that fluorescence emission from a broad depth was collected. The complementary information from both systems provided more sensitive and specific information for localizing and monitoring protein function and tissue structure than either of the two systems alone.

Parameters such as the shape of an action potential, activation sequence, and conduction velocity play an important role in normal cardiac development. Optical mapping (OM) is a technique, which collects information about a 3-D structure from a two-dimensional (2-D) projection map. Since the information is collected from a 2-D map, the curvature of the 3-D structure is often ignored, leading to errors in the conduction velocity map. Moreover, the 3-D structure is critical for cardiac development, especially in early stages, and can vary between embryos and between developmental stages. OCT can accurately measure embryo morphology in 3-D at early developmental stages. Recently, Ma et al.⁹⁷ demonstrated simultaneous OCT and OM imaging and evaluated both the 3-D structure as well as the conduction velocity. Incorporating the 3-D structure as imaged by OCT increased the accuracy of the conduction map. Consequently, conduction velocities in embryonic quail hearts were corrected to account for the 3-D morphology as obtained from OCT images. This 3-D correction reduced errors introduced from traditional 2-D projection map-based conduction velocities. The results showed the benefit of using OCT to integrate the true 3-D morphology when investigating cardiac electrophysiology.

Lopez et al.⁹⁸ used 4-D SDOCT to differentiate control and Wdr19 mutants. They used a SDOCT system employing a Ti:sapphire laser source with a central wavelength of 808 nm, bandwidth of 100 nm, and an A-line acquisition speed of 250 kHz. The system had an axial resolution of $5\mu \mathrm{m}$ and transverse resolution of $4\mu \mathrm{m}$. A smaller looping angle was seen in the Wdr19 mutants when compared to the control mice, thus indicating that 4-D OCT imaging can be a useful tool for cardiovascular phenotyping of early stage embryos.

In addition to label-free imaging, exogenous contrast agents can be used to obtain structural and functional information of embryos via microinjection. A real-time depth-resolved high-resolution imaging technique could enable more accurate microinjection into different embryonic organs and compartments. Most recently, Syed et al. demonstrated the first use of OCT for image-guided microinjection. The SDOCT system utilized was the same as in Ref. 98 but with an A-scan rate of 50 kHz. They visualized microinjection of gold-silica nanoshells into embryo yolk sac vasculature to study blood flow dynamics.⁹⁹ In addition, quantitative measurements were taken and showed the capability of OCT to measure injection volumes.

Davis et al.¹⁰⁰ employed Doppler OCT with an SDOCT system to image developing chicken hearts at very early embryonic stages where the heart fuses into a tube and begins the looping process. Utilizing a previously developed technique called spectral Doppler velocimetry (SDV),¹⁰¹ hemodynamic measurements were made as the heart develops from stages HH 11 to 14. SDV is considered an SDOCT analog to pulsed Doppler ultrasound imaging and is capable of producing depth-resolved hemodynamic measurements. This technique calculates the angle of flow from a volume, which increases the accuracy of blood flow velocity measurements. Davis et al. also solved the problem of phase wrapping that occurs if the Doppler phase shift is greater than $2\pi$ by implementing the cellular-automata method described in Ref. 102, thus greatly increasing the range of Doppler blood flow velocity measurements.

Similarly, a number of other groups have studied the effect of blood flow on morphogenesis of the heart using SDOCT systems and have investigated various techniques to calculate the absolute blood flow velocity. In 2010, Ma et al.¹⁰³ quantified the absolute blood flow velocity distributions in the heart of a chicken embryo by calculating the slope of the centerline at the outflow tract (OFT) of the heart at different positions and the Doppler angle needed for accurate measurements. This proved to be a faster and more reliable method to calculate velocities at the inlet and outlet of the OFT than earlier techniques. A number of forces act on the walls of the heart, such as the wall shear stress and blood pressure. Studies making use of modeling, instead of the Doppler signal, have also been performed for analyzing these forces.^104–106

Zhenhe et al.¹⁰⁷ utilized a high-speed SDOCT system to quantify blood flow velocity in embryonic chicken heart in vivo. The SDOCT system was comprised of an ultrahigh speed line-scan CCD camera, $1145\text{lines}/\mathrm{mm}$ diffraction grating, and a long focal length lens. This system had a maximum A-scan rate of 92 kHz resulting in a measureable velocity range up to $24\mathrm{mm}/\mathrm{s}$. Recently, Midgett et al.¹⁰⁸ evaluated blood flow in the OFT of a chick embryo to characterize hemodynamic changes during early stages of cardiac development. Several parameters were measured such as OFT blood flow velocity with respect to lumen diameter (from Doppler OCT images), peak velocity, peak flow rate, stroke volume, wall shear rate, cardiac cycle length, and time of flow. These measurements were supported by computational fluid dynamic simulations, thus providing a greater understanding of blood flow during cardiac looping.

Following the work on strain measurements in embryonic chick hearts, where the cardiac wall motion and blood flow were quantified with OCT in both normal and abnormal hemodynamic condition,^109–111 Ma et al.¹¹² proposed a method to measure the wall strain in the OFT of chick hearts at early developmental stages, in vivo. The strain was calculated from a segmented myocardial wall from a cross-section of a 4-D dataset. Since hemodynamic forces play an important role during cardiac development, understanding the mechanical properties of cardiac tissues at different developmental stages is crucial. These studies have demonstrated the potential of using OCT to study the biomechanical properties of the embryonic hearts by quantitative analysis of the wall strain.

4.2.2.

Swept-source optical coherence tomogrpahy

Larin et al.¹¹³ demonstrated the overall ability of SSOCT to perform 3-D structural imaging of live embryos at different stages. Structural and functional assessments of embryos from stage E7.5 to E10.5 were performed. Figure 9 shows a snapshot of a 3-D structural image of an E10.5 embryo. Furthermore, nongated 4-D reconstructions of live embryos were acquired and successfully captured the embryonic heartbeat. Figures 9(b) and 9(c) show snapshots from the first OCT images of a beating mammalian heart in 4-D (3D+time).

Fig. 9

(a) Snapshot from a 3-D structural image of a mouse embryo at E 10.5. (b) Snapshot from a gallery of denoised and synchronized time lapses acquired at different positions and used for 4-D reconstruction of the heartbeat. (c) Snapshot from a 4-D reconstruction. Adapted from Ref. 113.

In 2011, our group demonstrated the first in utero OCT imaging of live mouse embryos starting from E12.5 until the end of embryogenesis.¹¹⁴ The embryos were imaged with a commercial SSOCT system that consisted of a “broadband” source with an A-scan rate of 16 kHz, output power of 12 mW, central wavelength of 1325 nm, and bandwidth of 100 nm. The total acquisition time for a 3-D dataset was $\sim 20\mathrm{s}$. The system had lateral and axial resolutions of $8\mu \mathrm{m}$. This study focused on which embryonic structures could be imaged with OCT through the uterine wall. The embryos were kept alive by maintaining the body temperature of the anesthetized mother. However, motion artifacts caused by maternal respiration reduced the image quality and demonstrated the need for additional techniques for motion compensation.

Figure 10 shows 3-D reconstructions of mouse embryos at E13.5, E15.5, and E17.5, respectively. The images clearly show the morphological and size differences between the three stages. Fine details, such as craniofacial details, are clearly visible in stages E15.5 and E17.5.

Fig. 10

3-D reconstructions of mouse embryos in utero at different developmental stages: 1, head; 2, forelimb; 3, hind limb; 4, pinna of the ear; 5, eye; 6, yolk sac; 7, uterine wall; 8, follicles of vibrissae. (a) 13.5 dpc, (b) 15.5 dpc, and (c) 17.5 dpc. Scale bars correspond to 1 mm. Adapted from Ref. 114.

Limb development in embryos at various developmental stages as imaged using in utero OCT is shown in Fig. 11. The OCT images correlated well with histology and electron micrographs of embryos at corresponding embryonic stages. In addition to imaging the gross morphology of the embryos, the ability of OCT to provide cross-sectional images was also demonstrated by imaging the forelimb. The digits and half of the forelimb could be visualized, suggesting that OCT is capable of providing detailed information for characterizing internal and external structures of the limbs. Next, in utero brain imaging was performed, and the cerebral cortex and ventricles were clearly visualized until E16.5. After this stage, only the cortex was still visible. This study provided an effective protocol to perform in utero longitudinal imaging of mouse embryos.

Fig. 11

3-D reconstructions of the embryonic forelimb at different stages of development of a mouse embryo. Insets show scanning electron micrographs of the embryonic forelimb at the corresponding embryonic stages. 1, Cartilage primordium of distal phalangeal bone; 2, cartilage primordia of phalangeal bones; 3, cartilage primordia of metacarpal bones; 4, follicles of vibrissae. Scale bars correspond to (a–c) $500\mu \mathrm{m}$, (d–g) 1 mm. Adapted from Ref. 114.

In addition to gross anatomical investigations, a number of examinations of ocular developments in murine models were performed using SSOCT. Making use of the in utero imaging protocol developed in Ref. 114, Larina et al.¹¹⁵ studied ocular structures of mouse embryos at different stages of embryogenesis and also showed the ability of OCT to differentiate normal and mutant mouse models based on differences in their ocular structures. The mutant model (Pax6-SV40 T-antigen transgenic mouse line) spontaneously forms lens and retinal tumors during embryonic development. Longitudinal imaging was performed on wild-type and mutant mice at E13.5, E15.5, and E17.5. OCT was also able to capture the 3-D structure of the hyaloid vasculature, which is a capillary network of the embryonic eye, proving that OCT can assess ocular vascular remodeling during embryonic development. Moreover, OCT successfully distinguished the wild-type embryos from the mutant embryos.

As an extension to this study, our group recently quantified embryonic eye development at different developmental stages of a murine model.¹¹⁶ An increase in size and morphological changes were clearly visible (Fig. 12), and the results correlated well with the values obtained using UBM. However, the OCT images provided significantly higher spatial resolution than UBM, enabling more accurate quantification of eye volumes.

Fig. 12

2-D SSOCT images of the mouse eye in utero at different developmental stages: (a) E 13.5, (b) E 15.5, and (c) E 17.5. Adapted from Ref. 116.

The brain of a developing embryo is highly susceptible to changes caused by a number of factors, both genetic and environmental. A high-resolution noninvasive imaging technique would provide valuable information to aid in understanding the causes and outcomes of various brain developmental disorders. Thus, OCT is gaining popularity for embryonic brain imaging. Our group imaged the developing brain structure in utero using an SSOCT system while developing a protocol for high-resolution in utero imaging of live mouse embryos.¹¹⁴ Sudheendran et al.¹¹⁷ utilized this protocol to investigate the effects of prenatal alcohol exposure on the developing brain of mouse embryos using OCT and compared the results with ultrasound imaging. The results of this pilot study showed that the volume of the lateral ventricles of the ethanol-exposed fetuses was significantly larger than those from the control group. The volume of the brain obtained from SSOCT was verified by measuring the volume from the ultrasound images. However, due to inferior spatial resolution and contrast in the ultrasound images, the ventricular margins were not clearly defined, unlike in the OCT images. This resulted in a twofold decrease in variance of the ventricular volumes imaged by OCT, as compared to ultrasound. Figure 13 shows OCT, ultrasound, and H&E stained images of the control and ethanol-exposed fetal brains. It is readily apparent that the ventricles of the embryos exposed to ethanol are dilated in comparison to the controls.

Fig. 13

Effect of ethanol on mouse fetal brain development at E14.5. (a and b) OCT images in the transverse plane, (c and d) ultrasound images in the coronal plane, and (e and f) H&E images in sagittal plane of fetal brains from both (a,c,e) control and (b,d,f) ethanol-treated embryos. The ethanol-induced increase in ventricular dilation can be observed with all the imaging techniques, and in all three planes of orientation. US, ultrasound; LV, lateral ventricles. Scale bars are $500\mu \mathrm{m}$. Adapted from Ref. 117.

Doppler SSOCT has also been used widely to study the murine embryonic cardiovascular system. In vivo structural imaging and hemodynamic measurements were performed in early stage mouse embryos by Larina et al.¹¹⁸ in 2008. This was the first study that evaluated the normal blood flow in externally cultured mouse embryos using a Doppler SSOCT system. Spatial and temporal blood flow velocity profiles were acquired from the dorsal aorta and yolk sac vessel. Figures 14(a) and 14(b) show structural and Doppler images of the yolk sac and the dorsal aorta, respectively, from a E9.5 embryo. As seen from Fig. 14(b), strong Doppler signals were detected from the selected regions of the yolk sac vessel and dorsal aorta. Here, blue and red depict the direction of blood flow. A peak blood flow velocity of $3.2\mathrm{mm}/\mathrm{s}$ and a heart rate of $2\text{beats}/\mathrm{s}$ was observed. These results correlated well with previously obtained values using confocal microscopy,¹¹⁹ thus showing the ability of Doppler SSOCT to perform accurate blood flow measurements. Due to the greater imaging depth of OCT as compared to confocal microcopy, measurement of flow velocities and heart rates from the dorsal aorta were possible.

Fig. 14

SSOCT Doppler measurements of blood flow in live embryo culture. (a) Structural image of a E9.5 embryo showing fragments of a yolk sac and an embryonic trunk. (b) Corresponding color coded Doppler image showing strong signals produced by blood flow in the yolk sac and dorsal aorta. Adapted from Ref. 118.

In a similar study, Larina et al.¹²⁰ made hemodynamic measurements of individual blood cells in murine embryos at E8.5. At this stage, the heart begins to beat and most blood cells are found in the blood islands with very few circulating erythroblasts. Hence, this study was described as an OCT signal detection technique from single circulating blood cells. Hemodynamic measurements of individual blood cells were made using Doppler OCT. Color-coded Doppler velocity maps from the dorsal aorta at different phases of the cardiac cycle were also acquired. Blood flow dynamics clearly indicated pulsatile flow (Fig. 15). The heart rate and flow velocity obtained from this study correlated well with other published data, supporting the physiological relevance of the data and ability of Doppler SSOCT to obtain accurate hemodynamic measurements. Assessing individual time frames, measuring the Doppler velocity shift from each visible blood cell, and distance of the cell from the vessel wall suggested the presence of laminar flow. These results showed the ability of SSOCT to provide spatially resolved hemodynamic measurements at early stages of embryonic development without the need of any additional image processing techniques.

Fig. 15

Doppler velocity signals from blood cells on E8.5. (a) Structural and corresponding color-coded Doppler velocity images acquired at different phases of the heartbeat cycle. Green corresponds to zero velocity. Individual blood cells are distinguishable in the dorsal aorta. (b) Magnified view of the same area showing Doppler signal from single cells as well as a small group of cells. Adapted from Ref. 120.

Doppler OCT has mostly been used for imaging the vasculature of live samples. However, Doppler OCT is highly insensitive to blood flowing in the transverse direction to the probe beam. Hence, blood flow in vessels orthogonal to the probe beam cannot be easily quantified with Doppler OCT. In addition, the phase stability of the system determines the dynamic range of the Doppler OCT imaging. A number of systems with superior phase stabilities have been implemented to combat this problem. Although phase contrast methods have also been applied,¹²¹ they are similarly affected by the phase sensitivity of the OCT system. Time varying speckle analysis has been proposed as an alternative technique to overcome this limitation.^47,48,122 Mariampillai et al.⁴⁸ utilized speckle analysis to reconstruct the vasculature in the skin of nude mice. Sudheendran et al.¹²³ utilized SSOCT in conjunction with speckle variance OCT (SVOCT) to obtain 3-D reconstructions of the vasculature in live embryos in culture at E8.5 and E9.5. The SVOCT results were compared with Doppler OCT, and the results demonstrated that both methods perform equally well when the blood flow is in the axial direction, but SVOCT performs better when the vasculature is almost perpendicular to the OCT probe beam. Most recently, Kulkarni et al.¹²⁴ demonstrated that sparsity-integrated speckle anomaly detection algorithms further increase the quality of 3-D reconstructions of the network of blood vessels in developing embryos by mitigating problems associated with the presence of vessel structures at multiple spatial scales, thin blood vessels with weak flow, and artifacts resulting from bulk tissue motion.

The majority of embryonic investigations have utilized avian, fish, or mouse models due to the wide availability of transgenic models and ease of development of novel mutants. Rats have not been used as widely as mice for embryonic studies due to a lack of transgenic libraries. After the derivation of rat embryonic stem cells,^125,126 the use of rats as a mammalian model to study diseases gained momentum. Larina et al.¹²⁷ showed the first live imaging of rat embryos in 2009, where the beating heart was captured using SSOCT and blood flow velocity was quantified with Doppler OCT. Rat embryos of stages E10.5 and E11.5 were studied, which correspond to stages E8.5 and E9.5 in mouse embryos. Compared to mouse embryos, the rat embryos exhibited stronger and more stable heartbeats. The 3-D reconstructions of the embryo with the yolk sac were depicted and seemed to have a higher resolution due to the larger size of the internal structures as compared to mouse embryos.

4.2.3.

Nongated image acquisition

In gated acquisition, reconstruction errors are introduced because of the strong dependence of time stamps obtained from the trigger signals, which are often weak in live embryonic imaging. Nongated image acquisition eliminates this problem by removing the need for any postacquisition synchronization. One of the first studies of embryonic hearts using nongated OCT was done by Yazdanfar et al.⁸⁵ utilized nongated acquisition with retrospective gating OCT to study the blood flow throughout a beating X. laevis heart and the surrounding vasculature. Mariampillai et al.¹²⁸ also used retrospective gating on X. laevis embryo hearts by utilizing an optical Doppler cardiogram to obtain signals from the heart instead of using an ECG and demonstrated 4-D imaging at 45 fps. A TDOCT system was used to obtain the Doppler optical cardiogram, whereas an SSOCT system was used to image the embryos. The Doppler data were then used to synchronize the B-mode image sequences. Since the image sequences are out of phase in nongated image acquisition, Liu et al.¹²⁹ adjusted the phase lags by acquiring a longitudinal image covering the entire region of interest in addition to the series of transverse images. The transverse or parallel slices were first aligned with respect to each other and then aligned with respect to the longitudinal slice, thus synchronizing the transverse images.

In the above case, the longitudinal slice should be chosen carefully such that it does not contain the beating heart. To avoid such situations and also to overcome the limitation of error accumulation, Larina et al.¹³⁰ introduced a method called sequential turning acquisition and reconstruction (STAR) for imaging periodically moving structures. As seen in Fig. 16(a), the STAR approach acquired 2-D+T images along planes which share a common axis, which was unlike other nongated imaging approaches that acquired consecutive images in parallel imaging planes. This common central 1-D+T line from all images served as a temporal gating signal for postacquisition synchronization. After synchronization, the images were represented in 3D+T by transforming them into orthogonal coordinates [Fig. 16(e)].

Fig. 16

Sequential acquisition and reconstruction (STAR) method. (a) Image sequences acquired along radial planes (sharing a central axis). (b) Intensity of the central lines of the slice sequences versus time (kymograph) acquired sequentially at different orientations ${\theta }_i$. (c) Synchronization of the sequences by time warping the kymographs. (d) Synchronized radial sequences covering one period resulting from application of the warping function. (e) Visualization of the synchronized sequences as cutting planes or (f) interpolated into a Cartesian grid for volume rendering. Adapted from Ref. 130.

During dynamic acquisition, there is a tradeoff between acquisition time and spatial resolution. When the acquisition time is too long, the motion of the sample blurs the image. If the frame rate is increased to reduce this blurring, the SNR of the image can decrease as fewer backscattered photons are captured. When imaging cardiovascular systems, a frame rate of 100 Hz or more is required due to the velocity of the moving structures to be imaged. Building up on techniques of tiled imaging followed by mosaic reconstruction, Yoo et al.¹³¹ proposed a method to improve the field of view and resolution of dynamic 2-D OCT by combining tiled image sequences of cyclically moving structures in the mammalian embryonic heart. Live rat embryos were imaged with an OCT system similar to the one used in Ref. 118. The image acquisition protocol and reconstruction algorithm imaged cardiovascular structures with cyclical motion. Dynamic mosaicing overcame limitations of frame rate, resolution, and field of view during fast cardiac imaging. Figure 17 shows the principle of tiled acquisition and mosaicing used in this study.

Fig. 17

Tiled acquisition and automatic mosaicing procedure for dynamic images. (a) Sequentially acquired overlapping cardiac image series. (b) Static image acquired by collapsing each sequence into one. (c) Spatially aligned static images. (d) Spatially transformed and zero padded image sequences. (e) Temporally synchronized sequences. (f) Blended image sequences. Adapted from Ref. 131.

Bhat et al.¹³² proposed another method of nongated acquisition, where two sets of parallel 2D+T image sequences were acquired orthogonal to each other in 2013. A high-resolution 4-D (3D+T) volume was reconstructed accurately by making use of the similarities at the points of intersection of the two slices.

4.3.1.

Optical pacing

One drawback with studying cardiodynamics is the weakening of the heartbeat of the embryo during in vivo experiments. Optical pacing (OP) is a method that uses pulsed laser light to control the heart rate in embryos. Jenkins et al.¹⁴¹ demonstrated the first use of pulsed infrared light to optically pace intact quail embryos hearts. This technique was later adapted by Peterson et al.¹⁴² in 2012 to alter the heart rates of intact quail embryos to measure the resultant shear forces using OCT. Their results showed that OCT is an effective technique for evaluating OP protocols to study the effects of altered cardiodynamics. More recently, Alex et al.¹⁴³ captured and monitored a drosophila heart at different stages of its lifecycle (from larva to adult stages) by optogenetic pacing of the heart using an OCM system.