Hyperspectral microscopy for avian retina analysis
A fast and innovative hyperspectral microscopy approach for advanced biological tissue characterization
At a glance
The study of avian retinas is fascinating in the biological, ecological, and physiological fields. Particular attention has been paid to oil droplets, the miniature lenses within avian retinal cones which have a specific transmittance regulated by the type of carotenoid present.
Hyperspectral microscopy of retinal sections allows researchers to correlate the spatial distribution of oil droplets with their spectral features. This technique complements behavioral studies and opens new research frontiers for studying the visual systems of different species, while also advancing fields such as comparative vision science and biomimetic sensor design.
Reconstructed RGB image from the hyperspectral dataset of retina sample.
In this application note, we present the characterization of avian retina samples by hyperspectral imaging using HERA VNIR hyperspectral camera coupled with a commercial microscope.
In detail, we show the advantage of correlating spatial and spectral information to map and quantify the distribution of oil droplets in chicken retina samples.
Introduction
The chicken visual system consists of various types of photoreceptors, including 5 types of cones, each distinguished by the presence of oil droplets. These droplets serve a dual role: (i) acting as microlenses to enhance light delivery to the photosensitive compartment of the cone and (ii) as an optical cut-off to modulate spectral sensitivity. Their specific transmittance is determined by the carotenoid composition and content and unambiguously correlates with the cone spectral type.
Consequently, the analysis of the distribution and spectral characteristics of oil droplets inherently reveals information about the corresponding cone types.
Further investigation into their quantification, shape, size, and composition may shed light on their influence on visual function, offering insights that could complement behavioral studies.
Measurement setup and sample
Hyperspectral images were acquired in the visible range (400-700 nm) on retina tissue isolated from the chicken eye.
The measurements were performed with an Olympus microscope (CX43) coupled with our HERA VNIR hyperspectral camera.
The samples were studied in transmission mode and illuminated with a white LED (bright field mode). A 20× objective was used, obtaining a 375 × 300 µm field of view (FOV). The total measurement time was approximately 60 seconds.
Chicken retina samples were prepared in flat mounts and fixed in 4% paraformaldehyde for 30 min, then washed in PBS and sealed under the coverslip with Vectashield medium (H 1000-10, Vector Labs). To preserve the cone mosaic as intact as possible, 200 mm spacers were placed under the coverslip. Retinal quadrants (Figure 1) representing the nasal (N), ventral (V), temporal (T), and dorsal (D) parts were dissected from the retina. In this application note, we report the analysis of two portions in the nasal and temporal quadrants.
Figure 1: Graphical representation of chicken retina flat mount preparation, showing the nasal (N), ventral (V), temporal (T), and dorsal (D) quadrants.
Experimental results and data analysis
The reconstructed RGB image from the hyperspectral dataset (Figures 2a and 2b) reveals a distinct pattern of red, yellow, and pale green oil droplets, corresponding to the single long-wavelength sensitive (LWS or ‘red cones’), medium wavelength sensitive (MWS or ‘green cones’), and double cones, respectively. The difference between “clear” and “transparent” oil droplets instead, corresponding to short wavelength sensitive cones (blue and ultraviolet), lies within the UV range of the spectrum (< 400 nm).
The transmission spectra of selected red, yellow, and pale green oil droplets (Figure 2c) show distinct spectral features: red oil droplets have a cut-off wavelength of around 525 nm, yellow oil droplets of around 475 nm, and pale green oil droplets of around 455 nm. These spectral differences improve the ability to map their distribution across the entire investigated area and enable their quantification through an automated quantification protocol.
Figure 2: (a) Reconstructed RGB image from hyperspectral dataset of retina chicken sample in the nasal quadrant of the left eye. (b) Zoomed red edged ROI evidencing red, yellow, and pale green oil droplets. (c) Average spectra (dotted lines) and standard deviation (shaded areas) of different red, yellow, and pale green oil droplets, and of the background cornea.
An automated quantification protocol was applied to the nasal and temporal quadrants of chicken retina samples to quantify the number of the different oil droplets. This protocol involved first the extraction of the meaningful endmembers (Figure 3a). Then, pixels were classified employing the Spectral Angle Mapper (SAM) algorithm and maps of the classified pixels were created (Figure 3b). This allowed to assign unique labels to the droplets and enable their quantification.
Figure 3: (a) Normalized endmember spectra of the different oil droplets (1: red; 2: yellow; 3: pale green oil droplets); (b) example of a SAM map of the yellow oil droplet.
The histograms reported in Figure 4 show the quantified oil droplets, in two portions of the left eye quadrants (nasal and temporal). This quantification may vary depending on the individual or specific environmental or experimental conditions. In particular, the tissue fixation process and the focal plane chosen can influence these values. However, in this application note, we have demonstrated the ability of hyperspectral imaging to efficiently and effectively characterize and quantify oil droplets in retina samples in the image scene.
Figure 4: Histogram of the numbers of the different oil droplets in the nasal and temporal quadrants
Conclusion
This application note highlights the capabilities of HERA VNIR hyperspectral camera coupled to a microscope for the characterization and quantification of certain oil droplets in chicken retina samples. Its versatility offers significant benefits while supporting multimodal approaches.
When integrated with an optical microscope, the system showcases its potential for spatial and spectral analysis, offering new insights into complex biological structures across various applications, including ophthalmology, biomedical research, pharmaceuticals, and nanoscale materials.
Discover the HERA Hyperspectral Cameras
