Brief introduction for the non-invasive, inexpensive and fast Purkinje image -based method for measuring the spectral transmittance of the human crystalline lens density in vivo.
Alternative download link:
https://www.dropbox.com/s/588y7epy13n34xo/purkinje_imaging.pdf?dl=0
Purkinje imaging for crystalline lens density measurement
1. Petteri Teikari, PhD
Singapore Eye Research Institute (SERI)
Visual Neurosciences group
http://petteri-teikari.com/
Version “Wed 28 March 2018“
Purkinje Imaging
Quantifying spectral
transmittance of the
ocular media
4. Purkinje
Images
Deviceasaresearch
tooldevelopedback
inthe90s(At College of
Optometry, UniversityofHouston,
Houston, Texas(GLS) & Department of
Ophthalmology, UniversityofCalifornia at
Davis, Davis, California (CAJ, DLH)
READ DETAILS FROM https://www.ncbi.nlm.nih.gov/pubmed/11444627.00: “The illumination portion included a
small light source (10-W quartz-halogen bulb) housed behind a collimating lens (L1). In this collimated light path was placed a
filter wheel (F) containing various narrow-band (10- to 12-nm bandwidth half-height) interference filters. A 450-Hz synchronous
motor was used to rotate the filter wheel rapidly and at a constant rate. The data for four of these filters, 410, 430, 450, and
550 nm, were collected for comparison with the results for the same wavelengths using the psychophysical bipartite (BIP) device.
Light passing through each interference filter was converged by relay lens L2 sothat a focused image of the light source formed the
IVth Purkinje image.
The image-capturing portion of the system included a solid-state Intensified Charge-Coupled Device (CCD) video camera
(Philips Amperex Electronic Corporation) fitted with a close-up lens assembly. The angular separation between the illuminant
housingand the videocamerawasfixed at just under 30°.”
9. IMAGING
High
Dynamic
Range
Needed
Figure above shows an example of linear fit to the response curve for very low light intensity using the HDR
function. This would provide higher linear dynamic range for systems with low light conditions.
In light-starved systems (read noise limited), you need a linear response in the dark portion while maximum Full Well is of less importance. In systems with plenty of
light (shot noise limited), you want a linear responsein a brighter portion of the scale but without saturation of the image. This allows you to still have linear dynamic
range but just for the conditions your system operates in. The camera/image sensor can be optimized automatically with the HDR function in this way to have
greaterdetailswithoutaffecting repeatability. info.adimec.com
10. IMAGING
High
Dynamic
Range
Needed
RaspberryPi
notenough
most likely
The OmniVisionOV5647 CMOS sensing chip of Raspberry Pi camera module is composed of 1944x2592 =
5,038,848 base pixels and is overlaid with a 2x2 GBRG Bayer pattern array (BPA). Thus, there are 972x1296 =
1,259,712BPAs,eachproducing auniquered,uniqueblue,andtwouniquegreen 10-bitpixelvalues.
https://doi.org/10.1177/1477153516649229
15. Software
Python/C++ image
processing requiredfrom
OpenCV or/and
Tensorflow thenfor
detecting3rd
and 4th
Purkinjeimage
And then compare the
intensitities.
Or domanually inImageJ
atprototyping phase
Fourthimageisformedafteradoublepassthroughthecrystallinelens
1) REGION PROPOSAL
https://medium.com/@smallfishbigsea/
faster-r-cnn-explained-864d4fb7e3f8
2)Pixel-by-pixelsegmentation,
a.k.a.“Semantic Segmentation”
https://github.com/GeorgeSeif/Semantic-Segmentation-Suite
The raw A/D output was then converted to a radiance profile
using calibration data for the camera/frame-grabber system.
Custom software allowed automated analysis of the averaged
radiance profiles for each interference filter. A least squares
linear regression line was fitted to the flanks of each radiance
profile (representing back-scattered light from various layers in
the crystalline lens, etc.), and the radiance for the point on the
regression line lying directly underneath the peak of the IVth
image was calculated. The radiance for this position on this
downward sloping scatter shoulder was subtracted from the
gross radiance at the peak of the radiance profile. Thus, the net
reflected radiance of the IVth Purkinje image alone, or output,
was computed separately for each wavelength.
The subsequent calculation of individual lens
density consisted of taking the common log of the ratio of the
input radiance to the net reflected output radiance of the IVth
image. The result was divided by two to compensate for the
double-pass through the anterior segment of the eye. Several
such density measurements were made, and the average
density was calculated for each eye of every subject. No
adjustment was made for the path length difference (about 3%)
caused by the beam being an average of about 15° off-axis or
for pupillaryarea.