1, Dispersion compensation for optical coherence tomography:
We implemented dispersion compensation to enhance OCT image resolution. The presence of dispersion causes the point spread functions to broaden, leading to reduced resolution. To address this, we employed the Short-Time Fourier Transform (STFT) and iterative method to extract dispersion information using a single A-line spectrum mirror data. A method was required to align these peaks to the same location and narrow down their width. The positions of each peak were plotted against wavenumber. By applying densely overlapped windows, we obtained a 2D depth-wavenumber plot.
Now the question becomes: how can we change this spatial-spectral distribution of the mirror spectrum data from figure A to figure B? The basic idea for achieving this is to minimize the wavenumber-dependent variance of the zig-zag ridge, and the algorithm for doing so is presented here.
Below, you'll find core MATLAB code that demonstrates the details of obtaining the dispersion compensation term. Please note that your own code may differ significantly from what is provided here, as you may not require all the input arguments.
% Part of Main function
ord = 2;
fun = @(a)(obj_stft(aline,a,ord,k_linear,k0,Fs,winSize,overlapLength,rmvDcRows));
[a,J,exitflag,output] = fminsearch(fun,0);
disp_STFT(:,xi) = -a.*(k_linear-k0).^ord;
% if you'd like to do further the 3rd order or higher orders dispersion compensation please repeat the previous steps.
% in my application for the iridocorneal angle imaging, 2rd order is enough to me.
% here is the object function
function J = obj_stft(aline,a,ord,k,k0,Fs,winSize,overlapLength,rmvDcRows)
delda_phi = -a.*(k-k0).^ord;
aline = aline.*exp(1i*delda_phi);
[S,~] = stft(aline,Fs,'Window',hanning(winSize,'periodic'),...
'OverlapLength',overlapLength,'FFTLength',length(aline));
S = abs(S(size(S,1)/2+rmvDcRows+1:end,:));% exclude the DC area by jumping 'rmvDcRows' rows of data,...
...making sure for correctly finding the mirror position peaks
[~,indz] = max(S);
J = var(indz);
end
2, Iridocorneal angle OCT imaging:
To achieve higher imaging resolution, we combined 3 SLD laser sources. We performed a cross-line scan, simultaneously conducting both radial and circumferential scans to capture detailed information about the angle structures. You can see how the cross-line scan works and the imaging region, covering a part of Schlemm's canal from a top view.
To enhance depth differentiation in our images, we employed a color encoding technique. This involved a semi-automated, graph-based segmentation method, after which allowed us to assign specific colors to different depth ranges. To automate the process, we utilized the segmented result from the current image as an estimate of the search region for the next image, taking into account the similar boundaries between adjacent images.
This is a 3D OCT reconstruction. By performing image slicing in different axes, we can examine the details of the trabecular meshwork (TM), Schlemm's canal (SC), and collector channels (CCs) at the angle. SC is shown in the minimum enface projection image.
Stitching together the serial circumferential frames allows us to create a panoramic view that provides comprehensive details of the angle structures.
From the videos below, you can observe visible changes in the Schlemm's canal, the distribution and branching of collector channels, the morphology of the TM/SC/CC region, as well as nearby vessels.
3, Evaluating the Effect of Pulse Energy on Femtosecond Laser Trabeculotomy (FLT) Outflow Channels for Glaucoma Treatment in Human Cadaver Eyes: Current methods for treating glaucoma, except selective laser trabeculoplasty, are all invasive to some degree, as they require opening the eye either to insert surgical instrumentation or an implant. In our approach, we developed a technique called femtosecond laser trabeculotomy (FLT), which is a non-invasive, non-incisional method to bypass aqueous humor outflow resistance through the trabecular meshwork (TM) in the anterior part of human eyes.
In this study, we use OCT to investigate the influence of pulse energy on the creation of outflow channels within the TM during FLT surgery in human cadaver eyes.
Fig. 1: FLT surgical system: Femtosecond laser is guided by a visible aiming beam to scan across the entire trabecular meshwork (TM), starting from inside Schlemm's canal (SC) and extending into the anterior chamber (AC). Surgical size: 500µm x 200µm (blue box on TM; see the monitor)
Fig. 2: Schematic illustrating an FLT channel (W=500 µm, H=200 µm). The red spots indicate the focal points of the femtosecond laser arranged in a raster pattern, effectively photodisrupting the TM tissues.
Fig. 3: Sample (human cadaver eyes, $2.5k/pair) preparation for FLT surgery.
Fig. 4: FLT channel evaluation using 3D OCT images.
Fig. 5: Illustration of FLT channel evaluation.
Fig. 6: FLT channels under a light microscope. Unfortunately no depth information. Let’s go to use 3D OCT !
Fig. 7: Use of OCT images to evaluate/optimize the FLT pulses.
Fig. 8: Statistics of cutting width (Wc) in TM at different energies.
Fig. 9: 3D OCT representation of the FLT channel at different viewing perspectives.
Table. 1: OCT evaluation of FLT channels created at varying pulse energies.
4, Modeling meibum secretion: Alternatives for obstructive Meibomian Gland Dysfunction (MGD): (This is the very first finite element modeling paper in MGD)
The paper delves into meibum secretion in obstructive MGD, a prevalent cause of dry eye. Our research ideas come from daily life. Squeezing toothpaste in the morning reminds me that meibum, the oil expressed on our ocular surface, is actually a non-Newtonian fluid.
Modeling non-Newtonian fluid dynamics in meibum secretion can be a complex issue. But there is beauty in simplicity. This initial and simple modeling paper can guide you through the development of the model and help you understand the novel mechanisms behind obstructive MGD.
The paper describes the finite element modeling of meibum secretion and the effects of changes in meibum viscosity, duct diameter and eyelid pressure on measured meibum flow in the terminal duct. The modeling results suggest that increased meibum viscosity, decreased duct diameter, and weakening of the eyelid pressure on the meibomian glands all show potential for dramatically affecting the secretion of meibum through an alternative obstructive mechanism compared to the more conventional Hyperkeratinization plugging mechanism that is more widely accepted.
A), Pressure dependency: Flow rate ∝ eyelid pressure
B), Viscosity effects: minimum yield stress ↑, plastic viscosity↑ leads to flow rate ↓
Cross-sectional distribution of the velocity maps (μm/s) and flow rates (nL/s) of meibum secretion from a meibomian gland orifice at different meibum viscosities and eyelid pressures:
C), Geometric effects: Flow rate ∝ R4
Supplement: A simplified physical model using free body diagram (Analytical solution)
In summary, the results are of great interest as they support a novel pathophysiologic theory of Meibomian gland dysfunction, which also has an impact on therapeutic issues.