Microtechnology enables endoscopic confocal microscopy

Single-axis confocal microscopes are commonly used to obtain high-resolution images of biological tissue. Traditionally, excised tissue samples are fixed with preservative chemicals and inspected under a tabletop microscope. However, there is now rapidly growing interest in in vivo imaging, which will permit observation of biological processes that were previously inaccessible. The biggest challenge in realizing this goal is miniaturizing the microscope while maintaining high resolution and image quality.

Extended efforts toward constructing small versions of single-axis confocal microscopes have involved either shifting the optical fiber movement or using the lens to scan the sample. But the results are neither fast enough for real-time imaging nor sufficiently practical to scale down to endoscopic sizes. Microelectromechanical systems (MEMS) scanners provide a scalable approach for miniaturizing single-axis confocal scanning endoscopes.^1,2 Yet the high numerical aperture (NA) lens required to ensure high resolution has led to microscopes with limited working distance (WD) and field of view (FOV). Our approach decouples resolution and WD by using a dual-axes confocal architecture,³ and shrinks the system while maximizing imaging performance with a new MEMS scanner.

Figure 1. Schematic of the dual-axes confocal microscope architecture.

Dual-axes confocal microscopy uses two low-NA objectives with the illumination and collection axes crossed at an angle, θ, from the midline, as shown in Figure 1. This design offers advantages over conventional single-axis architecture for in vivo imaging. First, subcellular resolution can be achieved in both transverse and axial dimensions using low-NA objectives that facilitate miniaturization. Second, a long WD allows for post-objective scanning, which minimizes aberrations. Third, scattered light along the illuminated beam has a low probability of being collected outside the focus, thus increasing the detection sensitivity and dynamic range. The 2D MEMS scanner is the key optical element that enables post-objective scanning in a diminutive package while maximizing FOV.

We designed and fabricated a 2D MEMS scanner,⁴ as shown in Figure 2. It is made from double silicon-on-insulator (SOI) wafers and is actuated by self-aligned vertical combs for large scanning angles. The scanner is designed to be integrated in a 5mm-diameter endoscope, and achieves maximum DC optical deflections of ±4.8 and ±5.5° for the outer and inner axes, respectively. The corresponding torsional resonant frequencies are 500 and 2.9kHz. These mirror characteristics allow real-time imaging with a large FOV. The scanner is also metallized with a 10nm-thick aluminum layer to increase mirror reflectivity.⁵

Figure 2. Scanning electron micrograph of the 2D microelectromechanical system (MEMS) scanner includes two mirrors: one for illumination and one for collection.

Figure 3. The macroscopic dual-axes breadboard imaging setup. PMT: Photomultiplier tube.

The imaging capability of the 2D scanner in the dual-axes configuration was first demonstrated in a breadboard setup for both reflectance and fluorescence modes. A schematic of the imaging arrangement is shown in Figure 3. The target sample is illuminated by a 488nm-wavelength laser beam. The signal from the volume where the beams overlap is collected by a single-mode fiber and acquired by a computer. For fluorescence imaging, a long-pass optical filter is inserted into the collection path to selectively transmit the desired fluorescent signal.

Figure 4. Reflectance image of ex vivo human colon tissue from the breadboard system taken at eight frames per second. (The scale bar is 30μm.)

Figure 5. Fluorescence image of muscle tissue removed from a mouse that produces green fluorescent protein. The image was obtained with the handheld probe prototype. (The scale bar is 100μm.)

Results from the breadboard setup^4,5 demonstrate the feasibility of high-resolution real-time imaging with a 2D MEMS scanner. A reflectance image of excised human colon tissue is shown in Figure 4. In the image, we can discern crypt structures with lumen and surrounding colonocytes. The measured full-width-half-maximum (FWHM) transverse resolution is 3.9 and 6.7μm for the horizontal and vertical dimensions, respectively.

We are now downsizing the breadboard system to a 10mm-diameter handheld probe using 488nm-wavelength light. Figure 5 shows a fluorescence image from the first probe prototype. The results of the efforts described here hold great promise for high-resolution real-time imaging that will enable functional in vivo studies on small animal models of human disease and early diagnosis of cancer for patients in the clinic.

The authors would like to thank Daesung Lee, Gordon S. Kino, Michael J. Mandella, Thomas D. Wang, and Christopher H. Contag for useful discussions, and Pei-Lin Hsiung and Jonathan T. C. Liu for technical assistance. This research is supported by the National Institutes of Health under grant U54 CA105296.