1. Introduction

Microscopic examination of tissue through biopsy and histopathology remains the gold standard for the diagnosis of the majority of diseases and, in turn, often determines the management and prognosis of the patient. Tissue biopsy is not practical, however, for surveillance of large surface areas and in some organ systems is hazardous. Confocal microscopy is an optical imaging technique that provides depth-sectioned images of bulk tissue with a resolution rivaling that of histopathology. This capability opens the possibility of optical biopsy, the non-excisional investigation of tissue pathology. Techniques such as tandem scanning and laser scanning confocal microscopy have been developed to provide sufficient image acquisition rates to avoid motion artifacts in clinical imaging.1–3 These techniques generate two-dimensional images by scanning the angle of incidence of a collimated beam on the objective in two orthogonal planes. In raster-scanning strategies, the angle in one plane is swept rapidly (~15 kHz) while the orthogonal angle is scanned slowly (30 Hz). Sufficiently fast scanners such as spinning polygons are difficult to miniaturize, however, and limit the utility of these approaches to the fields of dermatology and ophthalmology. Several attempts have been made to miniaturize confocal microscopes but at present they suffer from either increased complexity and decreased field of view (micro-machine-based devices) or image pixelation (fiber-bundle-based devices).4,5

An alternative, fiber-optic based confocal microscopy technique, spectrally encoded confocal microscopy (SECM), has recently been demonstrated.6 This single optical fiber technique allows reflectance confocal microscopy to be performed through a flexible, compact probe, such as a catheter or endoscope. SECM employs the concept of wavelength division multiplexing (WDM) to encode one spatial dimension in wavelength. The fast scanning axis of standard scanning confocal systems is replaced by a series of focused points with each location being represented by a different wavelength of light. The reflected confocal signal as a function of spatial position is determined by measuring the spectrum of the light returned from the sample. Scanning the wavelength-encoded axis in an orthogonal direction with a slow mechanical actuator creates a two-dimensional image. Such a device can produce a confocal image devoid of pixelation artifacts with a large field of view and a high number of resolvable points. Since the mechanically actuated axis is scanned at a low frequency, probe complexity can be significantly reduced. Prior work performed using SECM demonstrated the feasibility of this technique.6 This first description of SECM, however, presented a system that was too large to be incorporated into an endoscope. In this Letter we demonstrate a novel design for a 10 mm diameter SECM probe that can be used in a variety of clinical settings, including hand-held operation, intraoperative diagnosis, and laparoscopy.

2. Design

The design of imaging probes appropriate for laparoscopic or endoscopic use is severely restricted by the limitations in size imposed by the instruments’ narrow access ports. For diagnostic medical devices, the diameter is limited to the maximum bore available for the application (6 mm for large bore endoscopes, 12 mm for laparoscopes.) Currently, multiple-element, high-NA microscope objective lenses are required to perform adequate optical sectioning in tissue. Since the length of these objective lenses typically exceeds the diameter of the probe, the optical axis of the objective lens must coincide with the axis of the probe. As a result, all of the components must be designed to ensure beam propagation along one axis. This restriction is problematic for SECM as diffraction by a grating necessitates off-axis illumination and/or transmission.

To enable endoscopic SECM, we have developed a novel probe that enables on-axis spectral dispersion for SECM, based on a dual prism grating combination (dual prism GRISM or DPGRISM). The DPGRISM allows on-axis illumination to be highly diffracted while maintaining on-axis propagation of the center wavelength (Figs. 1 & 2). The characteristics of the DPGRISM are defined by tradeoffs in wavelength, spectral range, desired resolution, efficiency and complexity of design. Our first prototype was optimized for a large field of view (FOV) and sub-micron resolution. To achieve these goals, a high periodicity grating was required. The resulting large diffraction angle necessitated the use of high index of refraction prisms (silicon, n = 3.51) to maintain on-axis incident and output beams. For the current design, we used an 1100 line/mm holographic, Dickson grating which had >95% efficiency for all polarization states when operated at Littrow’s angle (47°).7 In order to avoid optical losses due to refractive index mismatches, the prisms were antireflection coated (AR) coated to match the air and glass interfaces. The theoretical resolution of this design was calculated to be 0.81 µm using a 0.9 NA objective lens and the transmission efficiency was calculated to be 73%. This design is flexible enough to allow further miniaturization without significant design changes.

With SECM, the axis perpendicular to the wavelength encoded axis can be scanned using a variety of schemes to form a two-dimensional confocal image. For this prototype a stepping micro-motor was chosen due to the advantages it offers in low voltage requirements, linearity and availability of components. This motor was used to tilt the fiber and collimator, which were housed in a metal pivoting cone, in a direction perpendicular to the wavelength encoded axis (i.e. in and out of the page in Fig. 1). The above components, along with a 0.9 NA water immersion objective (LOMO 30x, which was machined to 10 mm diameter) were housed in a stainless steel tube. The probe was subsequently incorporated into an SECM system that consisted of a 1.3 µm, multiple-quantum well, semiconductor optical amplifier source with 65 nm FWHM (AFC Inc, BBS 1310) and fiber-optic beam splitter. Light back-reflected from the tissue was directed to an optical spectrum analyzer (Hewlett Packard 70950B) for the detection of the wavelength encoded reflectance.

 

Fig. 1. Outline of the dual prism GRISM (DPGRISM) based SECM probe consisting of a slow axis scanning motor (not shown), a pivoting cone holding the fiber and collimator, a DPGRISM and a modified 0.9 NA water immersion objective.

Download Full Size | PDF

 

Fig. 2. A photograph of the DPGRISM (A) that is enclosed within the 10 mm diameter SECM probe (B).

Download Full Size | PDF

3. Results

Images of an electron microscopy grid (6 µm bars separated by 25 µm), shown in Fig.3A, were acquired to test the resolution of the SECM probe. Measurements of the cross-section of individual bars of the grid (Fig. 3B) were used to determine lateral resolution.6

The combination of optics and light source used in this experiment resulted in images with a FOV of 658 µm and a lateral resolution of 1.1 µm, with approximately 15% variation over the entire FOV. We attribute the degraded resolution (1.1 µm measured; 0.81 µm theoretical) to the use of 1.3 µm wavelength light in an objective which was optimized for use in the visible. The measured transmission efficiency on double pass of the DPGRISM was 63%. Images of onion at various depths were acquired with the probe and are presented in Fig. 4. Microscopic features including cell walls and intra-cellular details of the onion cells can be clearly visualized. The images have dimensions of 650 × 300 µm (1300 × 600 pixels), and are displayed using a gray scale look-up table.

 

Fig. 3. Image of an electron microscope grid (A), consisting of 6 µm bars separated by 25 µm spaces, and a cross-section of a single bar (B). Only a section of the full FOV is displayed here so that the bars can be more visible.

Download Full Size | PDF

4. Conclusions

The DPGRISM-based SECM probe design suffers from certain limitations. While the design is flexible, the DPGRSIM itself can only be operated at the designed wavelength and with a predefined FOV. Since the number of illuminated grooves determines the number of resolvable points, if the overall diameter of the DPGRISM is reduced, the number of resolvable points also decreases. This limitation can be partially overcome by modifying the geometry of the DPGRISM to allow illumination of the grating at an angle. In addition, the current SECM system, which decodes the image using an optical spectrum analyzer, is relatively slow (~10 minutes per image) and has a poor dynamic range. Future implementation of interferometric, heterodyne Fourier transform spectroscopic (HFTS) detection will improve the speed and dynamic range of the system to enable clinical application of SECM in patients.6

In this work, we have demonstrated a new SECM probe based on a DPGRISM, which allows on-axis diffraction by a transmission grating. The 10 mm diameter DPGRISM-based SECM probe is capable of performing confocal microscopy in a variety of clinical settings including handheld intraoperative use and laparoscopy. Moreover, this design can be scaled without significant modification to less than 5 mm to allow insertion through the accessory channel of large bore endoscopes. When combined with a rapid and sensitive spectral detection system, this device will provide physicians with a tool for performing non-invasive sub-cellular diagnostic imaging in internal organ systems.

 

Fig. 4. Images of an onion obtained with the handheld SECM probe. Each image is 600×1300 pixels, 300 × 650 µm and taken at successively different depths as shown on the labels. The wavelength-encoded axis is along the (x) axis.

Download Full Size | PDF