Open Access
Add to BibSonomy
Abstract
Optical microscopy is limited to shallow interrogation depths as high-resolution imaging in scattering media is challenging. Current methods require complex and expensive experimental setup or suffer from low resolution. Through gating of photons exiting the scattering media using a restricted numerical aperture (NA) fiber optic plate (FOP), we establish a novel spatio-angular filter imaging device that allows deeper imaging in scattering media. Using dilutions of Intralipid (1-4 v/v%) and a USAF resolution target, it is shown that by reducing the NA of the FOP from 0.55 to 0.17, the interrogation depth improves ~2 times using trans-illumination.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Imaging with high resolution of an object embedded in a turbid medium requires controlling the effect of scattered light within the turbid medium so that only ballistic (no or insignificantly scattered) light that has traveled via region of interest is detected. The existing methods for gating such “ballistic” and “quasi-ballistic” photons include diffuse optical tomography [1], time gating methods [2], interference methods [3,4], polarization gating [5], and physical barrier gating [6,7]. When these methods are implemented in diffuse reflectance imaging applications, and in microscopic imaging systems, gating the relevant photons may become complex and costly. Interrogation depth can be significantly improved but at a cost of resolution.
One of the established methods in the field of biomedical imaging is Diffuse Optical Tomography (DOT) [8–10]. The technique utilizes near infra-red light to image large tissue volumes such as the breast or brain [11,12]. DOT relies on a model of light scattering to reconstruct the paths of photons as they travel through the tissue. This uncertainty manifests as blurring in the reconstructed images, such that resolutions between 0.5 and 10 mm are typical [1]. As the near infra-red range of light is significantly less absorbing and scattering as compared to the visible range, this may also limit the ability of DOT to detect many of the object’s intrinsic absorption-based features, for which the main absorption region lies is in the visible range (e.g. Hemoglobin).
Another technique that has been proposed for improving interrogation depth in scattering media is Angular Domain Imaging (ADI) [6,13,14]. It is an optical tomography method which introduces spatial filtering of the detected photons to reject the photons, which do not fall within a specified angle by Angular Filter Array (AFA) device for trans-illumination. AFA is a fine hole (tunnel or channel) array that provides separation of scattered versus non-scattered photons to improve resolution and contrast of diffuse images. The aspect ratio of each micro channel (length/width) determines the acceptance angle, which is typically ≈0.3 degree (NA = 0.005) [15]. Resolution of the system depends on the physical parameters of the tunnels and limited by the width of the tunnels after which the light will not pass unobstructed. Following from that, the diffraction of the light occurring on the edges of very narrow tunnels cause significant blurring of the image. Another significant limitation of this method is its restriction to scale the tunnel size as manufacturing such tunnels close to a micron diameter is very challenging. Recently, FOPs have also been used to perform extended depth of field imaging in scattering media [16].
In this work, imaging in highly scattering media is approached from a novel prospective of a lens-free architecture of the camera exploiting Spatio-Angular Filtering (SAF) [17]. The term “Spatio-Angular filter” is referred to the numerical aperture or acceptance angle gating using the lensless fiber optic plate (FOP) coupled to CMOS imaging platform. SAF consists of a low numerical aperture, coherent fiber optic bundle of small pixel size (<10 µm), directly mounted on a sensor, and an illumination unit. The experiments were performed using transillumination setup for ease of characterization. This approach can prove to be beneficial for biomedical applications where the major source of the contrast can be the light absorbed by hemoglobin in red blood cells in visible range. The hardware design of the camera provides an instant 2D imaging of the FOV of several mm2 without scanning. The data collection mode for SAF imaging can be both contact and non-contact. The lensless design of the hardware renders the device compact, portable and durable for usage in clinical settings, with no movable parts that are prone to artefacts caused by the object’s motility. Application of the flexible fibers also permits operation of endoscopic applications.
2. Methods
2.1 Numerical aperture gated imaging device
A Raspberry Pi camera module v2 (Sony IMX219 8-megapixel sensor) was custom fitted with 2 FOPs ((1) 0.55NA FOP, 13µm core, 3.2mm diameter, 25.4mm length, Schott, USA and (2) 0.17NA, 15µm core, 3.6mm square, 25.4mm length, Collimated Holes, USA) after removal of the cover glass/filter protecting the sensor surface. The proximal surface of the FOP was placed in contact with the surface of the image sensor using mineral oil (Life, Canada n~1.46-1.47) to reduce the refractive index mismatch between the both surfaces. A custom 3D printed case was designed to hold the assembly firmly in place (Fig. 1). The distal surface of the FOP was used as the imaging inlet to allow conduction of light from the sample surface to the image sensor. The numerical aperture prescribed by the refractive index mismatch of the core and cladding of the FOPs (0.55 and 0.17 NAs) was responsible for numerical aperture gating. A 1951 USAF target was used as a test sample and was imaged using two FOP based camera setups and a lens-based camera (Microscope objective, 0.25NA Motic, China) setup, so 3 imaging setups were compared to each other. The face of the USAF target that contained the reference lines was facing the FOP/lens. To emulate a scattering environment, varying concentrations (1-4%) of Intralipid (Sigma Aldrich, USA) was used in a Petri dish to mimic low and the high range of scattering in human tissues (e.g. dermal layers and mucosal layers - µs’ (1-4mm−1) [18,19]). The USAF target was immersed inside the Intralipid and the camera was placed on a 3D translation stage. Zero interrogation depth was at the point when the distal end of the FOP was in direct contact with the surface of the target and the Petri dish was filled with Intralipid. The FOP camera was translated incrementally upwards by 100µm steps increasing the layer of Intralipid between the distal end of the FOP and the target. For lens-based setup, a square holder was designed with a window made of silica microscopic cover slip. Zero interrogation depth for this setup was considered when the cover slip was in direct contact with the target and the lens was focused onto it. The holder and lens were translated with increment of 100µm like the above described setup. The translation was continued until the distance between the target and the distal end of the FOP surface/cover slip was 2mm. A white light collimated LED source (MCWHL5, Thorlabs, USA) was used to illuminate the target from the bottom within the framework of the trans-illumination geometry. Group 3 element 3 (~50µm size of the grooves) of the 1951 USAF target was chosen as a region of interest for all three setups. The integration time for the images was kept at 60ms for each acquisition of the image and all the experimental conditions, the intensity of illumination, integration time and scanning distances were the same for all setups.
Fig. 1 a) Experimental setup where a broadband light source (BLS) was used to project collimated light onto a transparent petri-dish with a USAF target glued to the bottom. The FOP camera (LC) was translated in the Z direction with a resolution of 100µm in the range 0-2mm. The microscope objective (MO) was focused onto the USAF target and a holder (PG) was translated in the Z direction like LC. The base of PG that was immersed in intralipid had a glass coverslip attached in the center to create an imaging window. b) the FOP coupled directly to the exposed sensor surface using matching refractive index medium (mineral oil). The FOP was attached to the sensor while continuously acquiring images to confirm that the FOP was at a minimal distance away from the sensor. The outer casing had a screw to hold the FOP in place once a desired distance was achieved. This setup proved to be robust for the experiments and the FOP placement was secure for a long period of time even after the experiments.
2.2 Image processing
Due to the non-uniform illumination conditions and hexagonal arrangement of cores inherent in the design of the FOP-camera setup, the image acquired would always have a variable background and presence of repetitive cladding pattern. Hence, a background correction needs to be performed to improve the image quality. This was achieved by using a differential method where two Fourier filters were employed to estimate the background/cladding pattern and to suppress the high frequency spatial noise. The filter created was a low-pass Butterworth filter as described below [20].
where , represent the spatial frequency components, cut-off was used to define a normalized radial extent from the origin and n represents the order of the filter and is the radial distance from the origin in the frequency domain. The cut-off was manually selected to obtain a background intensity variation and for reducing the high frequency components (noise) and it ranged from 0.001 to 0.008 for filter 1 and 0.03-0.08 for filter 2. For all image processing, a first order filter was used as it performed adequately without introducing artefacts. All the images acquired from the FOP setup and lens setup were filtered through the same process.2.3 Contrast ratio
To characterize the interrogation depth, a contrast ratio was calculated for each image. Two regions (3x3 pixels) were selected inside the image where the target was located (the black lines in Fig. 2). A similar window was chosen for the background and the average intensity for both was calculated. The resultant contrast ratio was defined as
Fig. 2 a) Optical setup for ray tracing simulation in Zemax. A point source with light emanating parallel to the axial transmission direction is impinged upon a 1mm layer of scattering media with a mean free path of 0.25mm and g (Henyey-Greenstein scattering function) value of 0.85 simulating a 1% intralipid solution [21]. The simulation had no absorption.
A threshold of 10% was used to define as the imaging interrogating depth. The standard deviation of the image without the presence of any target was calculated and this value was found to be around 7% for FOP based images. Hence, 10% was chosen as a threshold to compare all the imaging setups. This value was used to compare interrogation depths for all the setups.
2.4 Theoretical simulations
To obtain the angular distribution of the light exiting a scattering media, a point source (PS) with parallel beam to the axis of transmission was used to incident light into a scattering slab (S) with dimensions of . The scattering slab was modelled using mean free path and Henyey-Greenstein scattering function in Zemax. A radial detector was placed around the optical geometry to collect the light scattered in all directions. rays were used to do the ray tracing analysis and an angular cross section was analyzed to detect scattering in the all directions for 550nm wavelength.
3. Results and discussion
A visual comparison of the images obtained through all the setups is provided in Fig. 3. The rows of images correspond to an increasing concentration of Intralipid. The 0.17 NA FOP depicted in an increased interrogation depth when compared to the other imaging setups. For 0.17 NA FOP, the visibility of the lines is lost around 1700-1800µm compared to 1100-1200µm for 0.25NA lens and 800-900µm for 1% Intralipid concentration. The objective images and 0.55NA FOP images have similar interrogation depths. All the images were processed using the method described in section 2.2.
Fig. 3 Summary of images acquired through all 3 setups. The rows represent increasing concentration of Intralipid from 1 to 4% and a comparison of 0.17NA FOP, 0.25NA lens and 0.55NA FOP is presented. The columns represent the depth of target immersed inside a scattering layer of Intralipid. The depth was controlled using a translation stage to change the distance between the distal surface of the FOP and the surface of the target.
The angular distribution of a point source transmitting through a 1mm slab with (mean free path of 0.25mm and g value of 0.85 corresponding to 1% intralipid solution [21]) is depicted in Fig. 4. The green region represents −12° to 0° (0.25NA equivalent) and the red region represents −13° to −66° (66° corresponds to 0.55NA). Integration of the radiant intensity was performed from 0° to −66° (0.55NA) and the green region corresponds to 81.6% of radiant intensity and the red region corresponds to 18.4% radiant intensity. This indicates the highly forward scattering nature of the scattering media. Since the FOP has a core size of 10µm, at 1mm from the USAF target, the angle of light entering each individual fiber is 0.3° due to the collection geometry. Since majority of photons are forward scattered, the high NA FOP can still perform similar to a low NA microscope objective lens.
Fig. 4 a) Radiant intensity for a point source transmitting through a scattering slab of 1mm thickness depicting largely forward scattering. b) A schematic showing the angle of light entering each optical fiber when the FOP is placed 1mm away from the USAF target.
A contrast ratio threshold of 10% (black dashed horizontal line – 0.1 in Fig. 5(a), (b) & (c) was chosen to compare the interrogation depths for all the setups. The slope of contrast drop was gradual for 0.17 NA FOP compared to the other setups. This gradual decrease in slope corresponds to a greater interrogation depth for low NA FOP when compared against a lens-based setup as well as a higher NA FOP. Figure 5(d) depicts the interrogation depth at 10% contrast for all the setups at varying scattering parameters. For the lower scattering, the lower NA FOP performed considerably better at resolving the target at deeper locations. Reducing NA of the FOP also results in a substantial improvement in imaging interrogating depth but, the dependence of the interrogation depth on NA proved to be non-linear. The interrogation depth for 0.17 NA FOP for 1% Intralipid was around 1150µm compared to 450µm and 400 µm for 0.25NA lens and 0.55 NA FOP respectively, this corresponds to an approximate ratio of 3 for the lower scattering. This ratio decreases as the scattering increases and becomes closer to 1 for higher scattering conditions (4% Intralipid). This can be likely explained by the fact that the higher scattering results in an increased crosstalk between the optical fibers (highly diffuse photons are incident on the fiber at angles greater than the acceptance angles prescribed by the NA of the fiber) of the FOP as well as a chance of increased light coupling through the cladding of the FOP. The resultant contrast is affected decreasing the quality of the image and thus the imaging interrogating depth. It was also observed that for the 0.55NA FOP, the contrast at zero interrogation depth proved to be close to 0.5 when compared to the other setups where this contrast was close to 1. Due to the higher NA of this FOP, the crosstalk between the fibers is relatively large thus providing a reduced contrast even at zero interrogation depths. It was also noticed that in time due to the viscous nature of mineral oil and the current camera case design, the FOPs drifted from the original positions. This is planned to be remedied in the next generation using optically curing glue allowing for the FOPs to be attached firmly to the detector.
Fig. 5 Contrast ratio for a) 0.17NA FOP b) 0.25NA objective lens and c) 0.55NA FOP. The horizontal dashed lines in the figures represent the 10% contrast threshold. d) Imaging interrogating depth at 10% contrast threshold for all the setups, blue line represents 0.17NA FOP, red line represents 0.25 NA lens and black lines represents 0.55 NA FOP.
4. Conclusions
A novel Spatio-Angular Filter approach based on numerical aperture gated design was evaluated in terms of the imaging interrogation depths achieved in comparison to standard lens-based setup. A low NA was essential for deeper interrogation in scattering media exploiting fiber optic plates (FOP). At 1% Intralipid concentration (comparable to scattering in human mucosa) the imaging interrogating depth was found to be ~2 times larger than the other two setups. As the scattering increased (to 2-4% of Intralipid concentration, comparable with scattering in human skin), this ratio decreased prompting a non-linear dependence of the interrogation depth to the scattering. Further experiments need to be performed to elucidate and analyze this dependence. Another factor that could have affected the results was the coupling between the FOPs and the image sensor. The current design of the commercial FOPs used was found to be inefficient at reducing the crosstalk between fibers thus restricting the interrogation depths achieved through the setup. A customized FOP must be designed to significantly improve imaging interrogation through reducing the crosstalk between the pixels in the future experiments. Such a device can offer a compact and a robust imaging setup that can be used for imaging in scattering environments such as skin or oral mucosal layers when used in the reflection mode. Since the imaging of the device occurs mainly in contact mode, it can be a drawback in some applications where contact is undesired. The current generation of the device underperforms in terms of spatial resolution when compared to a lens-based system, this can be improved by using smaller core optical fibers. It can also be introduced as an endoscopic imaging device through a needle or current endoscopic imaging architecture by using controlled numerical aperture FOPs. The data presented in the study were acquired in the transillumination geometry and we plan to extend this study in the epi-illumination geometry to test the performance in reflectance mode.
Funding
Ryerson - Health Research Grant; Ryerson - Office of the Vice-President Research and Innovation (OVPRI): Ryerson - NSERC Discovery Grant and Ontario Graduate Scholarships.
Acknowledgements
The authors would like to thank the support by Kevin Liu (Physics Department, Ryerson University), Graham Pearson (Physics Department, Ryerson University) for assistance with mechanical design.
Disclosures
Aditya Pandya: Ryerson University (P), Irina Schelkanova: Ryerson University (P), Zeev Zalevsky: Ryerson University (P), Alexandre Douplik: Ryerson University (P)
References
1. A. Konovalov, E. Genina, and A. Bashkatov, “Diffuse optical mammotomography: state-of-the-art and prospects,” J. Biomed. Photonics Eng. 2, 2 (2016).
2. E. M. Hillman, J. C. Hebden, M. Schweiger, H. Dehghani, F. E. Schmidt, D. T. Delpy, and S. R. Arridge, “Time resolved optical tomography of the human forearm,” Phys. Med. Biol. 46(4), 1117–1130 (2001). [CrossRef] [PubMed]
3. A. M. Zysk, F. T. Nguyen, A. L. Oldenburg, D. L. Marks, and S. A. Boppart, “Optical coherence tomography: a review of clinical development from bench to bedside,” J. Biomed. Opt. 12(5), 051403 (2007). [CrossRef] [PubMed]
4. G. Indebetouw and P. Klysubun, “Imaging through scattering media with depth resolution by use of low-coherence gating in spatiotemporal digital holography,” Opt. Lett. 25(4), 212–214 (2000). [CrossRef] [PubMed]
5. S. P. Schilders, X. S. Gan, and M. Gu, “Microscopic imaging through a turbid medium by use of annular objectives for angle gating,” Appl. Opt. 37(22), 5320–5326 (1998). [CrossRef] [PubMed]
6. F. Vasefi, B. Kaminska, G. H. Chapman, and J. J. L. Carson, “Image contrast enhancement in angular domain optical imaging of turbid media,” Opt. Express 16(26), 21492–21504 (2008). [CrossRef] [PubMed]
7. F. Vasefi, M. Najiminaini, A. Chamson-Reig, M. Brackstone, B. Kaminska, and J. J. L. Carson, “Angular domain spectroscopic imaging for breast cancer margin assessment after lumpectomy,” in Optical Biopsy X (2012), 8220, p. 822003.
8. D. A. Boas, D. H. Brooks, E. L. Miller, C. A. Dimarzio, M. Kilmer, R. J. Gaudette, and Q. Zhang, “Imaging the body with diffuse optical tomography,” IEEE Signal Process. Mag. 18(6), 57–75 (2001). [CrossRef]
9. B. W. Pogue, T. O. McBride, U. L. Osterberg, and K. D. Paulsen, “Comparison of imaging geometries for diffuse optical tomography of tissue,” Opt. Express 4(8), 270–286 (2009). [PubMed]
10. V. V. Tuchin, Tissue Optics: Light Scattering Methods and Instruments for Medical Diagnosis (SPIE, 2015)
11. A. Li, E. L. Miller, M. E. Kilmer, T. J. Brukilacchio, T. Chaves, J. Stott, Q. Zhang, T. Wu, M. Chorlton, R. H. Moore, D. B. Kopans, and D. A. Boas, “Tomographic optical breast imaging guided by three-dimensional mammography,” Appl. Opt. 42(25), 5181–5190 (2007). [PubMed]
12. L. Becerra, W. Harris, D. Joseph, T. Huppert, D. A. Boas, and D. Borsook, “Diffuse optical tomography of pain and tactile stimulation: Activation in cortical sensory and emotional systems,” Neuroimage 41(2), 252–259 (2008). [CrossRef] [PubMed]
13. G. H. Chapman, B. Kaminska, P. K. Y. Chan, F. Vasefi, and N. Pfeiffer, “Angular filters for optical tomography of highly scattering media,” U. S. patent US20080177169A1 (2008).
14. Y. Zhang, F. Vasefi, E. Ng, A. Chamson-Reig, B. Kaminska, and J. J. L. Carson, “Mesoscopic reflectance angular domain spectroscopic imaging,” J. Biomed. Opt. 19(7), 076010 (2014). [CrossRef] [PubMed]
15. E. Ng, “Two dimensional angular domain optical imaging in biological tissues,” Western University thesis (2013).
16. A. Orth, M. Ploschner, I. S. Maksymov, and B. C. Gibson, “Extended depth of field imaging through multicore optical fibers,” Opt. Express 26(5), 6407–6419 (2018). [CrossRef] [PubMed]
17. I. Schelkanova, A. Pandya, Z. Zalevsky, and A. Douplik, “Spatial Phase Filter and Illumination Device for Deep Interrogation of Strongly Scattering Media and Uses Thereof,” U.S. patent Provisional 62/696,519 (2018).
18. H. Assadi, R. Karshafian, and A. Douplik, “Optical scattering properties of intralipid phantom in presence of encapsulated microbubbles,” Int. J. Photoenergy 2014(471764), 1–9 (2014). [CrossRef]
19. S. L. Jacques, “Optical properties of biological tissues: a review,” Phys. Med. Biol. 58(11), R37–R61 (2013). [CrossRef] [PubMed]
20. Rice University, “2D Frequency Domain Filtering and the 2D DFT,” https://www.clear.rice.edu/elec301/Projects01/image_filt/matlab.html.
21. H. J. van Staveren, C. J. M. Moes, J. van Marie, S. A. Prahl, and M. J. C. van Gemert, “Light scattering in Intralipid-10% in the wavelength range of 400-1100 nm,” Appl. Opt. 30(31), 4507–4514 (1991). [CrossRef] [PubMed]
