The achievement of super-resolution microscopy was widely celebrated with Nobel prizes in chemistry several years ago. While super-resolution imaging is becoming routine in many laboratories, there are still many challenges and limitations that must be overcome for these imaging approaches to be broadly adopted for biomedical imaging. Some of the key limitations are limited speed, field-of-view, and penetration depth. Equally importantly, most successes in super-resolution are based on fluorescence probes, the extension of super-resolution approach to other optical contrast mechanism is also important (see also TRD2.3). LBRC is developing several new approaches for mapping 3D volume at super-resolution limit based on structured light illumination. The advantage of our approach is that super-resolution information from 3D volume can be acquired in parallel with no additional illumination power as compared with just imaging a 2D plane. With the successful development of these techniques, we hope to use this technique to map the connection diagram over a large brain volume ex vivo and to monitor synaptic junction formation and dissociation in vivo.
For low scattering, ex vivo specimens, in a collaborative project with Prof. Ed Boyden, we are developing stimulated emission depletion (STED) microscopy with 3D structured light in order to rapidly map neuronal connectivity using visible light that requires resolution on the 50 nm level. For in vivo studies, light scattering degrades image resolution as we image deep into tissue specimens. In a collaborative project with Prof. Elly Nedivi, high-resolution mapping of synapatic distribution is needed to understand the plasticity of memory. While temporal focusing multiphoton wide-field excitation enabled high-speed imaging at shallower depths, structured light temporal focusing approaches are needed to overcome resolution degradation due to emission photon scattering.
3D high-throughput, high-resolution imaging based on structured light STED
STED microscopy is able to image fluorescent labeled samples with nanometer scale resolution. It is typically a point-scanning method [1], limited by the high intensity requirement of the depletion beam. With the development of high peak power lasers, two-dimensional (2D) parallel STED microscopy was realized [2]. Here, we develop the theoretical basis for extending STED microscopy to three dimensions in parallel by combining with structured illumination (SI) that generates a three-dimensional (3D) depletion pattern.
The pattern scans in 3D by shifting the relative phases of the five beams in the Fourier plane. This design modulates the out-of-focus light to form structure illumination along the propagation direction, which increases the axial resolution and reduces the photobleaching at the same time. Thus, the power requirement of the 3D SI-STED is the same as 2D parallel STED. In addition, the common path interference design is more robust to external environmental perturbations than incoherent designs. Widefield microscopy has a finite support in kz of the frequency domain. Other than kz = 0 point, the frequency components are zeros [3,4]. Three-dimensional interference pattern has multiple spatial frequency components in the Fourier domain. Thus, widefield 3D SIM can fill the "missing cone" by convolution of these two modulation transfer functions (MTF) [3]. The comparison of widefield and SI-STED after reconstruction shows in Fig.3. Under our simulation conditions, the lateral resolution of widefield microscopy is 305.2nm, and the axial resolution is 657.2nm. After the reconstruction, the SI-STED microscopy reaches 65.7nm lateral resolution and 88.5nm axial resolution that are 5-fold and 7-fold narrower than the widefield microscopy, respectively. The MTF of SI-STED microscopy is broader as well (Fig. 1).
Overcoming resolution loss using mosTF
LBRC's collaboration with Dr. Elly Nedivi requires high resolution imaging (better than 1 micron) over large volume at high speed to study memmory plasticity in animal models. Temporal focusing two-photon microscopy is a promising solution. A limitation of temporal focusing is that signal-to-background ratio and resolution degrade rapidly with increasing imaging depth. This degradation is a result of scattered emission photons that are widely distributed across the camera, resulting in a strong background. We have developed Multiline Orthogonal Scanning Temporal Focusing (mosTF) microscopy [5] that overcomes this problem by capturing a sequence of images at each scan location of the excitation line, followed by a reconstruction algorithm that reassigns scattered photons back to the correct scan positions. By reassignment of scattered photons, mosTF remarkably improves SBR and preserves fine structures when imaging through turbid media compared to lineTF. As a proof-of-principle demonstration, we imaged the same area of 200 nm fluorescent beads using water as our control condition (Fig. 2) and then 2 mm thick 0.2% intralipid solution corresponding to an imaging depth approximately 400 µm in the brain (Fig. 3). The working distance of the objective is 2 mm. The full field-of-view (FOV) is 205 x 205 µm2. The mosTF images were generated by a custom algorithm reassigning scattered emission photons to the correct locations. The images without reconstruction, that is, lineTF images, were generated by directly summing the intermediate images together without photon reassignment. mosTF is able to image eight times faster than TPLSM while keeping the same spatial resolution and high SBR. While the lineTF image cannot distinguish any beads, the mosTF image can distinguish individual beads with high SBR.