Electrically tunable lens integrated with optical coherence tomography angiography for cerebral blood flow imaging in deep cortical layers in mice

Abstract

We report the use of an electrically tunable lens (ETL) in a 1.3-μm spectral-domain optical coherence tomography (SD-OCT) system to overcome the depth of focus (DOF) limitation in conventional OCT systems for OCT angiography (OCTA) in mouse cerebral cortex. The ETL provides fast and dynamic control of the axial focus of the probe beam along the entire range of the mouse cortex, upon which we performed cerebral blood flow imaging of all cortical layers by stitching the OCTA images automatically captured at 6 focal depths. Capillary vasculature and axial blood flow velocity were revealed in distinctive cortical layers, and for the first time, in white matter. The implementation of ETL will enable many future OCTA applications in deep cortical layers to benefit a comprehensive investigation of the neurovascular function in brain activities and pathologies.

The cerebral cortex consists of six physiological layers along the cortical axis characterized by cellular types and density [1]. The major efferent pyramidal neurons in the mouse brain, for example, are largely populated within cortical layer III and V [2]. Such neurophysiologic variation leads to different metabolic requirements in these cortical layers during functional activities, which are supported by the spatial and temporal adjustment of blood flow. Previous studies using functional magnetic resonance imaging (fMRI) have revealed the highest changes in blood flow volume at the middle of the cortex in cat during visual stimulation [3], but the resolution of the imaging was not sufficient to identify the exact location of the peak volume changes. While other studies have used bolus tracking to observe the earliest and largest red blood cell content arrival in the layer IV during rat somatosensory cortex activation [4], only a few arterioles were captured for estimating the change in temporal signals. Until now, direct imaging of capillary vasculature and hemodynamics in distinctive cortical layers, particularly in the deeper cortex, has not been achieved, hindering the systematic study of neurovascular coupling and cortical functions.

Optical coherence tomography angiography (OCTA) has become an emerging tool in neurovascular research that provides high-resolution (μm), and three-dimensional (3D) blood flow imaging [5] and velocity mapping [6] within cortical tissue with large field of view (cm) in rodents without a need of contrast agents. The most recent development of OCTA, capillary velocimetry [7], [8], has revealed capillary flow homogenization in the activated mouse cortex during sensory stimulation and its contribution to cerebral oxygenation [9], [10]. The technique is also useful in investigating pathology-related capillary transit-time heterogeneity changes in aging and Alzheimer’s Disease [11].

However, a severe limitation in these previous studies was that the dynamic signals were only quantifiable within a shallow depth up to 300 μm below the cortical surface, merely reaching layer III in mouse cortex and far less than system ranging distance. This is due to the Gaussian optics used to deliver the OCT beam into cortical tissue, where the depth of focus (DOF) (~λ/NA²) and lateral resolution (~λ/2NA) are coupled. Using high NA and sharp focus, only a small depth range (characterized by Rayleigh length of OCT beam) exhibiting the desired lateral resolution of the system was previously used for capillary flow quantification.

A useful and practical solution to extend the DOF without sacrificing the lateral resolution is to shift the axial focus of the probe beam along the depth of the imaging subject. A challenge of using this approach in cerebral blood flow imaging is to precisely locate the focal plane among anatomical cortical layers, and/or to capture fast hemodynamic response during neurovascular coupling in multiple brain layers efficiently. Axially shifting the scanning optics or translational imaging platform, therefore, do not meet these requirements.

In this letter, we report a novel application of the electrically tunable lens (ETL) with a custom-built spectral-domain OCT (SD-OCT) system to perform multi-focal-plane cerebral blood flow imaging in mouse cortex. The system diagram is shown in Fig. 1, which includes a broadband super-luminescent diode (SLD) light source with a central wavelength of 1340 nm and a bandwidth of 110 nm at −3dB. The axial resolution provided by this light source was measured ~7.5 μm in air (~5.1 μm in mouse brain tissue). The output light from the interferometer was routed to a home-built spectrometer, which has spectral resolution of ~0.141 nm that provided a detectable depth range of ~3 mm. The line rate of the line-scan camera (1024 pixel detector-array, Goodrich Inc.) employed in the spectrometer was 92 kHz. The system sensitivity was measured to be 105 dB at the focus (~500 μm below the zero delay) with an incident light power of ~5 mW at the sample. In the sample arm, an ETL (EL-10-30 series, Optotune Inc) was placed before the paired Galvo mirror scanners, followed by a 10× objective lens with an effective focal length of 18 mm that was used to focus the OCT beam into the cortex, yielding a lateral resolution of ~11 μm and a DOF of ~200 μm. The ETL was equipped with a tuning range of −1.5 to +3.5 diopter (dpt) with an offset diverging lens (f =−150 mm).

Fig. 1.

Schematics of the custom-built spectral-domain optical coherence tomography (SD-OCT) system integrated with an ETL lens. SLD, super-luminescent diode; PC, polarizer controller; ETL, electrically tunable lens; The inset picture demonstrates the multi-focal-plane imaging scheme in the mouse cortex with one ETL tuning cycle. CTX, cerebral cortex; HP, hippocampus.

To measure the focus tuning characteristics from the ETL-equipped OCT system, the focus shift was measured by a beam profiler mounted on a linear translational stage, while two measurements were taken by setting the focal power to +0.5 and +1.4 dpt. The maximum intensity (at center of the beam) is plotted versus distance along the beam, shown in Fig. 2 (a). The two curves indicated the focus of the beam has a shift of 685 μm. So the range of +0.5 to +2.0 dpt used in this study was corresponding to ~800 μm focus tuning range. Together with the ~200 μm DOF provided by the objective, ETL system would be able to cover up to 1 mm of the entire mouse cortex thickness. The temporal response characteristics of the ETL was measured by OCT M-scan, while a mirror was placed at the focus of the OCT beam and a step driving signal was sent to the ETL to tune from +0.5 to +1.4 dpt. The position of the OCT signal from the mirror is used to represent the focal power, as shown in Fig. 2(b). The settling time was measured to be 5.46 milliseconds, which proved sufficient dynamic performance for rapid focus tuning to meet our requirements in the neurovascular coupling investigations.

Fig. 2.

(a) Beam intensity profile measured at +0.5 and +1.4 dpt focal power, indicating the shift of focus. (b) Step response of the ETL when focal power is tuned from +0.5 to +1.4 dpt.

To apply it for in vivo cerebral blood flow imaging over extended DOF in mouse cortex, 3-month-old male C57BL/6 mice (n=3) were used. All animal procedures in this study were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Washington. The animals were anesthetized with isoflurane in a mixture of 80% air and 20% pure oxygen. A standard 5×5 mm cranial window was created to eliminate the effect of cranial tissue scattering on the cerebral imaging results. The volumetric optical microangiography (OMAG) and Doppler OMAG (DOMAG) protocols were performed consecutively at different axial locations [11]. In the OMAG protocol, each B-scan had 400 A-lines covering a distance of ~2.5 mm in the x-axis. B-scans were repeated 8 times for angiography, and 400 steps were acquired in the slow scan direction, covering a distance of ~2.5 mm in the y-axis. Therefore, the data cube of one processed 3-D angiogram was composed of 1024×400×400 (z-x-y) voxels. In the DOMAG protocol, 25 A-lines were acquired consecutively at each depth location to yield one M-scan (z-axis). One B-frame had 380 M-scans covering a distance of ~2.5 mm in the x-axis, and each 3-D dataset had 300 B-frames. Therefore, one DOMAG data cube was composed of 1024×380×300 (z-x-y) voxels. In order to map an axial velocity range of ±6.1 mm/s in this study, Doppler processing of the complex signals was performed with 3-A-line intervals to increase the Δt between effective A-lines. A phase variance mask was also employed to segment the meaningful Doppler flow signals from the noisy phase background.

Figure. 3(a)–3(f) show six extracted OCT structure B-scan images in one tuning cycle that demonstrate the dynamic change of signal strength along the cortical depth. The EFL is tuned from +0.5 to +2.0 dpt at 0.3 dpt interval, so the focal plane shifts down sequentially from Fig. 3(a) to 3(f). The tissue around the focal plane location (yellow dashed lines) reflects/scatters more light due to higher laser power density at that region, resulting in higher signal strength. This is also documented with the A-line intensity profiles (Fig. 3(h)) extracted from these B-scan images, where intensity of the deeper cortical tissue improves dramatically when the focal plane moves deeper. Focus stacking were performed post image acquisition. Briefly, the B-scan images were masked by a set of Gaussian windows (indicated in Fig. 3(a)–3(f)) to weigh the out-of-focus regions and then summed to generate a composite image in Fig. 3(g). The axial intensity profile (Fig. 3(i)) of the composite image shows adequate OCT signal strength maintained along the cortical depth, with two major peaks at pia and white matter.

Fig. 3.

OCT structural B-scan images of mouse cerebral cortex with different focus settings. (a)-(f) Images obtained with 6 different focal planes tuned from pia to white matter (ranging from 0~1000 nm below the cortical surface). The yellow dashed lines indicate the focal plane center, and the white dashed squares enclose regions within depth of focus (DOF). A Gaussian window (white curve) is applied to each image filter out-of-DOF regions and fused to generated (g). (h) The normalized axial intensity profile for images in (a)-(f). (i) The axial intensity profile for the composite image in (g), two peaks are identified for strong scattering at pia and white matter. The scale bar represents 500 μm for both x and z directions.

To demonstrate the improved lateral resolution for imaging capillary flow in the entire mouse cortex, optical microangiography (OMAG) B-scans (Fig. 4(a)–4(f)) and 3D enface projections (Fig. 4(g)–4(l)) are compared in one tuning cycle. Due to the limited DOF (~200 μm) of the imaging objective, the regions outside of the DOF are severely blurred and the blood vessels are visually enlarged and unsolvable. It was not possible to simultaneously realize both pial and deep cortical (and white matter) blood flow with the single-focal-plane regime. By focused volume stacking using Gaussian window, we were able to generate a fused 3D volume (Visualization 1). Compare to the single volume (Visualization 2), the multi-focal-plane imaging has resolved capillary flow in all cortical depths in the single composite video. It can also be seen in the 3D enface images (Fig. 4(g)–4(l)) projected from the DOF regions, that the detailed vasculatures are revealed in as deep as white matter (Fig. 4(l)). These results show that the ETL can effectively extend the DOF for OCTA and maintain optimized lateral resolution for deep cortical blood flow imaging.

Fig. 4.

OMAG images at different focal planes. (a)-(f) B-scan flow images obtained with 6 different focal planes tuned from pia to white matter. The dashed squares enclose the regions within the 200-μm DOF, which was used to make the enface maximum intensity projection (MIP) images as shown in (g)-(l). The yellow dashed lines indicate the location of the B-scans. The scale bars represent 500 μm for x and z in (a)-(f), and for x and y in (g)-(f). The same Gaussian window filters were applied to the flow as to the structure images in Fig. 3, and the fused blood flow volume can be viewed in the supplementary movie.

To show the continuity of blood flow mapping from the focused volume stacking approach, we here presented five consecutive OMAG cross-sections (Fig. 5(a)–5(e)) from the fused 3D volume (the volume of Visualization 1), each showing blood flow signals over an extended DOF in the entire depth of the cortex. These five representative cross-sectional images have indicated vertically well-connected blood flow signal across six stitched volumes, and particularly, continuous penetrating blood vessels indicted with the yellow squares.

Fig. 5.

Selected OMAG cross-sections from the multi-focal-plane fused blood flow volume. (a)-(e) correspond to B-frame number 282 – 286 in the volume, respectively. The yellow squares encircle a continuous penetrating blood vessel from consecutive image frames.

Lastly, we calibrated the axial depth of the imaging with the cross-sectional tissue thickness of the mouse S1 cortex according to histology [12] and generated enface images of OMAG and DOMAG according to the axial location of six cortical layers and white matter (Fig. 6(a)–6(f)). OMAG was used to map vascular perfusion and DOMAG was used to detect the axial velocity of red blood cell movement in the vessels. In the high-resolution blood perfusion maps and the bi-directional axial velocity maps, some flow signals (cyan squared) are identified in specific cortical layers only (Fig. 6(d)–6(f)), which was checked from the 3-D data and exclude the possibility of projection tailing artifacts from its above. Distinctive white matter blood supply and its corresponding axial velocity (cyan arrows) are clearly shown in Fig. 6(f). To our knowledge, it is the first in vivo documentation of capillary-resolved blood perfusion with functional velocity measurements in mouse white matter with OCTA.

Fig. 6.

OMAG and DOMAG images in different cortical layers and white matter. Each image is generated with enface MIP projection of regions within DOF. The color bar represents RBC axial velocity of the flow descending to (green) and rising from (red) the deeper brain in a range of ±6.1 mm/s. The cyan squares and arrows indicate distinctive blood flow detected in specific layers (as opposed to projection tail artifacts). The scale bar represents 500 μm for x and y.

With focus volume stacking, continuous vascular maps from pial layers to white matter were successfully revealed. The results have demonstrated the practicality of using ETL in OCT/OCTA systems to extend the DOF for cerebral blood flow imaging in deep cortical layers. We note that a good quality linear translation stage would be able to serve the purpose of effectively shifting the focus for multi-focal plane imaging in the mouse brain [13]. However, the ETL approach has important advantages in the design and future practical applications. The compact and light-weight quality of the ETL makes the miniaturization of optical device possible, thus useful for applications where the hand-held or intra-cavity probes are required. Yet, there are limitations to use ETL in the current application. Regardless of the fast response and settling time (5.46 milliseconds) of the ETL (Fig. 2), the current data acquisition time for each 3D dataset of OMAG and DOMAG are 15 sec and 50 sec, respectively, at each focal plane. The data acquisition for six focal planes can take up to several minutes, which makes no guarantee for stable neurological condition and hemodynamic state during the time period. Therefore, the approach is more suitable to evaluate the hemodynamics of individual cortical layers of interest, for example, vascular transit time and heterogeneity in the most active layer IV during brain functional activation. Additionally, the system speed is upon further improvement by taking the advantage of high-speed swept-source OCT (SS-OCT). It is possible in the future to perform volumetric angiography in a fraction of a second with dynamic focusing. At this point, programed software control to deliver automatic fast transitioning of the focal plane is also desirable to streamline the multi-focal-plane imaging.

In conclusion, we have demonstrated the innovative use of an ETL with OCTA to perform multi-focal-plane cerebral blood flow imaging in mouse cortex. The ETL-integrated system provides dynamic focusing of the optical scanning at different depth along the cortical axis that has successfully extended the DOF. The results have shown an important application of the system to perform functional blood flow imaging in the deep cortex in mice to conveniently investigate the hemodynamics at targeted cortical location of interest. More importantly, the compact integration of the ETL can benefit the future design of hand-held or intra-cavity OCT probes to benefit a wide range of applications in research and clinical fields.