Simple ultraviolet microscope using off-the-shelf components for point-of-care diagnostics

Abstract

At the primary care setting, where there are often no or minimal laboratories, examinations often consist of self-testing and rapid diagnostics. Because of this, medical devices must be simple, robust, and easy to operate. To address these concerns, an alternate fluorescence microscope design uses ultraviolet (UV) excitation, since fluorescent dyes that are excitable in the visible region are also excitable by UV. This may allow for the removal of typical excitation, emission, and dichroic filters as optical components absorb UV wavelengths and UV is not detected by silicon based detectors. Additionally, UV has a very low penetration into samples, which may allow for controlling the depth of excitation, and thus the imaging volume. Based on these ideas, we developed a simple fluorescence microscope built completely from off-the-shelf components that uses UV to image fluorescently stained samples. The simple opto-mechanical design of the system may allow it to be more compact and easy to use, as well as decrease the overall cost of the diagnostic device. For biological validation, we imaged whole blood stained with acridine orange and performed a two-part white blood cell differential count.

Fig 1. Optical schematic and image of the UV fluorescence microscope system.

(a) Optical schematic of the designed UV microscope and (b) image of the fully assembled system. Samples were mounted on an xyz stage to enable focus adjustment and lateral scanning.

Fig 2. Performance of the assembled fluorescence microscope.

(a) MTF plot calculated for the image plane in both the tangential and sagittal directions showing the nominal performance of the system. (b) Image taken with a 455 nm LED of a high resolution 1951 USAF target presenting the measured performance of the system. The smallest resolvable feature on the target was group 8 element 1 (shown enlarged and outlined in white), demonstrated by the intensity profiles through the vertical and horizontal elements indicated by the red lines and arrows. Resolution target image has been contrast enhanced for visualization purposes.

Fig 3. Fluorescence image of WBCs stained with AO.

(a) Image taken using the 280 nm excitation source with an exposure time of 500 ms and detector gain of 0 dB without excitation or emission filters. The large, central white box represents the 1 mm x 1 mm FOV where WBCs are in focus. Insert image shows a pair of arbitrarily selected WBCs. (b) Image of the same FOV with an inexpensive absorption foil, which served as an emission filter, that only allowed the fluorescence signal from the WBCs. Color images were reconstructed from raw intensity values given by the Bayer mask of the camera.

Fig 4. Segmenting WBCs from the FOV.

(a) Image of AO stained whole blood taken under UV excitation in the 1 mm x 1 mm FOV, (b) WBCs after applying a Sobel filter, and (c) the segmented WBCs, displayed as the red circles, used for analysis based on the size criteria. Scale bars represents 100 μm.

Fig 5. Example histogram of the R/G ratios in a single FOV.

The red line shows the bimodal shape of the histogram, delineating what appears to be two distinct populations of WBCs (agranulocytes in the left population and granulocytes in the right population).

Fig 6. Comparison of calculated WBC values to commercial hematology analyzer reported values.

Bar graphs show the (a) total WBC/μl, (b) percent agranulocytes, and (c) percent granulocytes for five separate venous draws. Purple bars on the left show the mean value of 12 FOVs as calculated by the algorithm (with error bars indicating ± one standard deviation) while turquoise bars on the right show the gold standard value as reported by the commercial hematology analyzer (with error bars indicating ± 5% error, given in the hematology analyzer specifications on accuracy [16]).