Description:FIELD OF THE INVENTION
The present invention relates to a system for detection and removal of eyeblink artifacts from aqueous flare measurements and a method thereof. More particularly, the present invention discloses a system and a method for removal of eye blink artifacts from aqueous flare measurements by interfacing an electrooculogram (EOG) device to an objective ophthalmologic device and detecting the blink-induced aqueous flare using a machine learning tool for assisting healthcare professionals in grading Uveitis.

BACKGROUND OF THE INVENTION
Anterior uveitis is a type of eye inflammation that affects the anterior chamber, specifically the iris and choroid. Regular monitoring of this medical condition is crucial for its effective management. Aqueous flare is a phenomenon where the normally clear aqueous humor in the eye becomes turbid due to the presence of inflammatory cells and proteins. This turbidity can be quantified using light scatter measurements, as a continuous measure of the intensity of light scatter (ILS). The quantification of aqueous flare in the form of ILS measurements thus can aid in accurately assessing the degree of inflammation providing valuable insights into the severity of anterior uveitis.

Aqueous flare measurements have several other applications such as but not limited to assess level of inflammation in the eye in order to monitor the progression of a multitude of ocular diseases such as Uveitis, diabetic retinopathy, age-related macular degeneration (AMD), retinitis pigmentosa (RP), Behçet disease, Vogt-Koyanagi-Harada disease (VKH) etc. to assess how well an eye treatment is working, to measure inflammation after ophthalmic surgical procedures related to these diseases etc.

However, a significant challenge in accurately measuring aqueous flare is the interference caused by the subject's blinking. When the eye blinks, the tear film can disrupt the light scatter pattern. Eye blinks are therefore one of the sources of artifacts while measuring the intensity of light scatter (ILS) using an objective ophthalmic device causing noisy aqueous flare leading to artificially elevated readings of aqueous flare intensity. This can result in an overestimation of inflammation and in turn inappropriate grading of uveitis or anterior chamber inflammation, potentially hindering the effectiveness of its treatment.

A number of literature have been published including patents and non-patents documents in said domain.

Conventionally, clinicians use the Standard Uveitis Nomenclature (SUN) grading system to count the cells and flare in the anterior chamber of the eye to assess the level of inflammation. A non-patent literature by Kallirroi Konstantopoulou, et al. titled, “A comparative study between clinical grading of anterior chamber flare and flare reading using the Kowa laser flare meter”, published in Int Ophthalmol in 2015 suggests that clinical evaluation of aqueous flare by SUN grading system is subjective. Therefore, the prior arts teach the method of evaluation of aqueous flare by performing objective quantification of ILS measurements followed by its validation by comparing it with the subjective SUN grading system. This method is able to classify mild-to-moderate uveitis, however, provides results that are inconsistent. In case of lower-grade uveitis, though semi-auto grading is possible with the contemporary objective ophthalmic flare meters, there is still a challenge in achieving reliable and consistent scoring. This is due to the patient's eye blinking and the observer's shaky focus, which may collect unwanted light scatter from the cornea, iris, and crystalline lens, causing variance in the ILS measurements. These artifacts thus tend to mislead the scoring of uveitis inflammation and result in clinicians prescribing overdosage of topical drugs. Therefore, the conventional methods of classification and grading are lacking in reliability and consistency especially in case of lower grade Uveitis due to blink artifacts.
Another non-patent literature by Tadepalli, S., et al. titled, “ Reliability of Aqueous Flare Measurements During Uveitis by a Spot Fluorometer”, published in J Ocul Pharmacol Ther in 2022 describes how advanced equipment such as the spot fluorometer are used to examine conditions such as post-cataract and uveitis. This equipment evaluates the eye conditions by quantifying the level of aqueous flare by measuring the ILS. The spot fluorometer is highly immune to noises and is known to have a good signal-to-noise ratio (SNR) necessary for the accurate measurement of ILS. However, the ILS measurements are likely to be impacted by blinking, eye movements, and focusing jitter. A major hindrance to ILS measurements is the persistence of ILS after blinks. Additionally, the blinking confounds the ILS measurements along with the increased amplifier time constants leading to improper grading of the mild forms of uveitis using the spot fluorometer. The study emphasizes that the impact of these factors must be reduced before using the fluorometer to distinguish mild-to-moderate uveitis. However, this document has not disclosed any techniques to detect and remove eye blink artifacts from the ILS measurements.

According to a study by Li, Y., et al., titled "Automatic removal of the eye blink artifact from EEG using an ICA-based template matching approach," published in Physiol Meas in 2006, it was found that eye blinking is the most common cause of distorted electroencephalogram (EEG) data. Various methods have been suggested to improve the reliability of EEG recordings by filtering out artifacts such as electrical noise, heartbeats, muscle spasms, and eye movements. Regression models and wavelet treatments are commonly used to detect and eliminate artifacts in EEG data. However, these techniques have limitations, particularly in identifying spectrum artifacts. In a non-patent literature written by Islam et al. titled "Methods for artifact detection and removal from scalp EEG: A review," which was published in Clin Neurophysiol Pract in 2016, various blind source isolation techniques are described, including independent component analysis (ICA), artifact subspace reconstruction (ASR), empirical mode decomposition (EMD), and canonical correlation analysis (CCA) for signal disintegration. These techniques have been used for EEG and blink artifact signals to recover their components. Another technique, known as adaptive filtering, involves iteratively modifying weights to eliminate physiological artifacts from EEG signals. While these techniques aid in blind source isolation for EEG signals, they do not address ILS readings for measuring aqueous flare. None of these previous studies provide any method or technique for using EOG signals to remove artifacts from aqueous flare measurements.

The non patent document by Golrou, A.et al., titled “Wheelchair Controlling by eye movements using EOG based Human Machine Interface and Artificial Neural Network” published in 2022 reiterates that the orbicularis oculi muscle that contract during blinking, are detectable by electrooculography (EOG) . As the EEG measure brain function by detecting the biopotentials using the electrodes placed on the scalp. These electrodes may detect blink artifacts and motion artifacts, both of which can interfere with the EEG signal. Nevertheless, their method fails to provide any practical or clear methodologies to utilize EOG data for removal of blink artifacts from aqueous flare data or intensity of light scattering (ILS).

A patent document CN-106485208-A, titled, “The automatic removal method of eye electrical interference in single-channel EEG signals” relates to the automatic removal method of eye electrical interference in single channel EEG signals. It comprises of steps of collecting anti-phase single channel EEG signals, interval detection of eye electrical interference based on MSDW, obtaining an electrical interference interval, then electro-ocular signal estimation is carried out to the single channel EEG signals in each interference interval using the eye electricity method of estimation based on wavelet transformation, deducting, from single channel EEG signals, the electro-ocular signal estimating is obtained, obtaining the EEG signals after an electrical interference removal, realizing the removal of eye electrical interferences. This literature teaches the removal of eye blink artifacts using elaborate signal decomposition methods that are not user friendly or easy to understand and depends on labelled information requiring large number of resources.

Therefore, there is a need for a cost effective, user friendly system that is computationally efficient, to detect and remove eye blink artifacts from aqueous flare measurements providing results that are accurate and consistent to aid clinicians in grading of Uveitis apart from other applications in the field of ocular diseases.

OBJECT OF THE INVENTION
In order to overcome the shortcomings in the existing state of the art the main object of the present invention is to provide a system for detection and removal of eyeblink artifacts from aqueous flare.

Another objective of the present invention is to provide a system for detection and removal of eyeblink artifacts from aqueous flare measurements for grading of uveitis.

Another objective of the present invention is to provide a system for detection and removal of eyeblink artifacts from aqueous flare measurements for assessing level of inflammation in the eye to monitor the progression of a multitude of ocular diseases, for assessing how well an eye treatment works, for measuring inflammation after ophthalmic surgical procedures etc.

Another objective of the present invention is to provide a method for removal of eye blink artifacts from the aqueous flare measurements by interfacing the Electrooculogram (EOG) sensor to an objective ophthalmic device and further detecting the blink-induced aqueous flare using an unsupervised machine learning tool.

Another objective of the present invention is to provide a system that enables a single data acquisition device to record eye blinks and ILS measurements simultaneously.

Yet another objective of this invention is to provide an upgraded objective ophthalmic equipment that can be used to grade mild-to-moderate uveitis objectively and precisely.

Another objective of the present invention is to use an EOG-based objective ophthalmologic device to examine the influence of eyeblink artifacts on ILS outcomes.

Another objective of the present invention is to provide a method for detection and removal of eyeblink artifacts from aqueous flare measurements for grading of uveitis.

Yet another object of this invention is to provide a method /technique for effective classification of anterior uveitis that will aid clinicians in establishing the most effective dosage of ocular medications, hence reducing excessive and protracted treatment.

Yet another object of this invention is to provide a method for assessing the effects of blinking by healthy subjects or patients on results of ILS.

Yet another object is to provide a user friendly, cost effective, and reliable system that would assist health care professionals to detect and remove eyeblink artifacts from aqueous flare measurements and aid in accurate grading of uveitis besides other applications in the domain of ocular diseases

SUMMARY OF THE INVENTION
Accordingly, the present invention relates to a system for detection and removal of eyeblink artifacts from aqueous flare measurements and a method thereof. Particularly, the system of the present invention facilitates removal of eye blink artifacts from aqueous flare measurements by interfacing an electrooculogram (EOG) device to an objective ophthalmic device and detecting the blink-induced aqueous flare using a machine learning tool for assisting healthcare professionals in grading uveitis.

The system of the present invention enables simultaneous measurement of eye blinks and aqueous flare, followed by the k means clustering approach to separate the flare with and without blinks and clean up the aqueous flare data. The present invention can therefore aid in accurate grading of anterior uveitis inflammation by removing blink artifacts from measurements of aqueous flare.

The system comprises of an objective ophthalmologic module to include data acquisition submodule that can simultaneously measure eye blinks and aqueous flare such as intensity of light scatter (ILS); a processing module that applies machine learning approach such as k means clustering to separate the flare with and without blinks and clean up the aqueous flare data and a display unit (UI1) that allows the output to be analyzed and utilized by the user. The invention thus provides a single data acquisition device that is capable of recording eye blink and ILS measurements simultaneously. As a result, the upgraded objective ophthalmic equipment can be used to grade mild-to-moderate uveitis objectively and precisely in offline mode. The invention also finds application in assessing various ocular diseases and monitoring the progress of the treatment or surgical procedures thereof in this field.
The method of the present invention as per an embodiment comprises of steps not limited to acquiring aqueous flare and blink data simultaneously, preprocessing of the acquired data, applying machine learning tools such as k mean clustering, identifying data with blink and without blink, removing blink cluster from data etc. The present invention, therefore, provides a method to mitigate the visual disruptions resulting from blinking in the ILS data offline and evaluating uveitis especially mild-to-moderate uveitis precisely.

The present invention has significant advantages over the prior arts in the domain of processing of aqueous flare measurements. The invention for the first time presents a way of interfacing an EOG device with ophthalmic device enabling effective analyzing of blink-induced artifacts in the ILS measurements. The method is highly accurate wherein the variance in the ILS is significantly reduced after the removal of blink-induced artifacts. Therefore, blink-induced artifacts in the ILS measurements are rectified, leading to precise grading of uveitis inflammation.

Accordingly, the present invention provides a cost effective, reliable and consistent method to obtain blink artifact free aqueous flare measurements that can assist healthcare professionals in grading uveitis accurately besides having other applications of monitoring progression of several ocular diseases and assessing treatments or procedures thereof etc.

BRIEF DESCRIPTION OF DRAWINGS
Figure 1 displays a block diagram of a system and it’s working for detection and removal of eyeblink artifacts from aqueous flare measurements.
Figure 2 displays a flow chart of the blink artifact removal approach in the aqueous flare measurements.
Figure 3 displays processing of blink and aqueous flare data for removal of blink artifacts.
Figure 4 displays Scatter plot of ILS and EOG with and without blinks.
Figure 5 displays magnified view of a three clustered single ILS-blink signal before and after artifact removal: (A, B) The clustering of the blink-induced ILS at both amplifier time constants by applying k-means, agglomerative, and Gaussian clustering models is represented in three colors: navy blue, lavender, and pink, respectively. (C, D) The blink artifact-free ILS at both amplfier time constants 10 ms and 100 ms is achieved by removing the WB clusters.
Figure 6 displays mean ILS of normal subjects and synthetic uveitis patients’ data (With blink) WB and (Without blink) WOB artifacts: (A) Normal subjects ILS with and without blinks. (B) Synthetic uveitis WOB values are nearly identical to experimental uveitis data (P < 0.6955). (C) The pairwise comparison of the experimental WOB with cleaned WB and the WOB of healthy subjects at 10 ms and 100 ms is calculated using the Mann-Whitney nonparametric t-test. (D, E, F) The actual uveitis and cleaned data for all SUN grades showed no significant difference, with P values of 0.1439, 0.1193, and 0.1322, respectively.

DETAILED DESCRIPTION OF THE INVENTION WITH ILLUSTRATIONS AND EXAMPLES
While the invention has been disclosed with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made, and equivalents may be substituted without departing from the scope of the invention. In addition, many modifications may be made to adapt to a particular situation or material to the teachings of the invention without departing from its scope.

Throughout the specification and claims, the following terms take the meanings explicitly associated herein unless the context clearly dictates otherwise. The meaning of “a”, “an”, and “the” include plural references. Additionally, a reference to the singular includes a reference to the plural unless otherwise stated or inconsistent with the disclosure herein.
Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this disclosure belongs. The system, methods, and examples provided herein are only illustrative and not intended to be limiting.
The abbreviations used in the invention are represented in table 1 as below:
Table 1: Legend of abbreviations
S.no. Particulars Legend
1 Electrooculogram EOG
2 Intensity of light scatter ILS
3 Electroencephalogram EEG
4 Standard Uveitis
Nomenclature grading system SUN
5 Laser flare photometer LFP
6 Ocular flare analysis meter or a fluorometer OFAM
7 Signal-to-noise ratio SNR
8 Mean squared error MSE
9 Data acquisition system DAQ
10 Independent component analysis ICA
11 Artifact subspace reconstruction ASR
12 Empirical mode decomposition EMD
13 Canonical correlation analysis CCA
14 Cor-neoretinal potential CRP
15 Light-emitting diode LED
16 Without blink WOB
17 With blink WB
18 Gated recurrent unit GRU
19 Within-cluster sum-of-squares WCSS
20 Number of clusters n
21 Relative error RE
22 Standard deviation STD
23 Signal-to-artifact ratio SAR
24 Optical Coherence Tomography OCT
25 Age-related macular degeneration AMD
26 Retinitis pigmentosa RP

Some of the technical terms used in the specification are elaborated as below:
Aqueous flare- Aqueous flare is a phenomenon where the normally clear aqueous humor in the eye becomes turbid due to the presence of inflammatory cells and proteins. This turbidity can be quantified using light scatter measurements, which can provide valuable insights into the severity of anterior uveitis.
Blink artifacts- Eye blink artifacts are a common type of physiological artifact that can appear in some biomedical signals such as intensity of light scatter or electroencephalography data. They can negatively impact the analysis of the concerned data leading to inaccurate diagnosis.
Electrooculogram- Electrooculogram is a bio-amplifier that detects biopotentials due to eye blinking by positioning surface electrodes vertically above and below the eye.
Ophthalmic devices- Ophthalmic devices or ophthalmologic device are a broader category that encompasses all devices used in ophthalmology, including both objective and subjective devices. Subjective ophthalmic devices rely on the patient's responses or observations. Some of the examples of ophthalmic devices are Snellen charts, pinhole refractors, color vision tests etc.
Objective ophthalmic device- Objective ophthalmic or objective ophthalmologic devices are those designed to provide an objective measurement or assessment of the eye. These devices typically involve the use of technology or instrumentation to gather data that is not influenced by subjective factors like the patient's perception or interpretation. Objective ophthalmic devices provide quantitative, objective measurements. Some of the examples of objective ophthalmic devices are ocular flare meter, fundus cameras, visual field testers etc.
k-means clustering- K-means clustering is an unsupervised machine learning algorithm that groups objects into clusters based on their features.
Subjects- A person or organism that is the object of research, treatment, experimentation. Here in the description the term subjects include humans or animals who have been researched upon for detection and removal of blink artifacts from aqueous flare for assessment of ocular medical conditions such as uveitis. The term includes the patients who undergo tests or any screening for uveitis or such other ocular diseases.
Agglomerative hierarchical clustering - Agglomerative hierarchical clustering (AHC) is a hierarchical clustering method that groups items into clusters based on similarities. It's a bottom-up approach that starts with each item as its own cluster and merges pairs of clusters until all items are in one large cluster.
Gaussian mixture clustering- Gaussian mixture model (GMM) clustering is a data clustering technique that uses pseudo-probabilities to assign data to clusters. It's a soft clustering technique that's used in unsupervised learning. GMM clustering is designed for numeric data and is more complex to implement and interpret than other clustering techniques.

As per the present medical practice, aqueous flare is assessed as a continuous metric of the intensity of light scatter using objective ophthalmic device. Utilizing the ILS technique, the degree of aqueous flare is assessed to determine the extent of uveitis inflammation. The disturbance of the blood-aqueous barrier and an increase in light scattering are the hallmarks of anterior uveitis. Conditions such as post-cataract are evaluated by quantifying the level of aqueous flare, the permeability of the corneal endothelium, the dynamics of tear production, and the dynamics of aqueous humor.

The setup of the ophthalmic device results in unintentional capturing of the scattered light from blinking as well. Therefore, a major hindrance to ILS measurements is the persistence of ILS after blinks. Consequently, it is essential to identify and remove blink artifacts in any ILS data analysis.

The aqueous flare measurements have several applications such as but are not limited to assessment of level of inflammation in the eye in order to monitor the progression of ocular diseases such as uveitis, diabetic retinopathy, AMD and RP; assessment of how well a treatment is working; measurement of inflammation after ophthalmic surgical procedures like trabeculectomy, pars plana vitrectomy, argon laser pan retinal photocoagulation laser capsulotomy etc.

The present invention discloses a system (S) and a method for detection and removal of eyeblink artifacts from aqueous flare measurements. The present invention relates to removal of eye blink artifacts from aqueous flare measurements by interfacing an electrooculogram (EOG) device to an objective ophthalmic device and detecting the blink-induced aqueous flare using a machine learning tool for assisting healthcare professionals in grading uveitis or assessment of other ocular disease conditions.

The reference numerals used in the present invention are tabulated below in table 2.

Table 2: Legend of reference numerals
Ser no. Item description Reference numerals
1 Subject Su
2 User U
3 System S
4 Objective ophthalmologic module O
Objective ophthalmologic device /fluorometer cum slit lamp biomicroscope O1
Illumination source O11
Excited light/UV light O111
Illumination slit O12
Focused light O121
Scattered light O122
Photodetector O14
Amplifier O15
Intensity of light scatter data ILS
Data acquisition submodule O2
5 Bio amplifier module E
Bio amplifier device/EOG E1
Surface Electrodes E11
Reference electrodes E12
6 Processing module P
CPU P1
Machine learning tool P2
Clustering algorithm or tool P21
7 User interface module UI
Display unit/monitor UI1

The system (S) for detection and removal of eyeblink artifacts from aqueous flare measurements and its performance are as described below.

As per an embodiment of the present invention the block diagram of the system (S) is shown in figure 1. The system comprises of an objective ophthalmologic module (O), bio amplifier module (E), processing module (P) and user interface module (UI). The various modules are elaborated upon in the following paragraphs.

As per an embodiment of the present invention the system (S) for detection and removal of eyeblink artifacts from aqueous flare measurements of a subject (Su) comprises of an objective ophthalmologic module (O) that comprises of an objective ophthalmologic device (O1) integrated with slit lamp biomicroscope for detection of intensity of light scatter (ILS) from a subject’s eye, as an objective measure of aqueous flare, that in turn comprises of:
Illumination source (O11) for emitting an illumination beam of excited light (O111) towards subject’s eye.
Illumination slit (O12) placed between said subject’s eye and said illumination source (O11) to convert said excited light (O111) into focused light (O121) for onward passage to the subject’s eye.
Photodetector (O14), for detecting scattered light (O122) from the subject’s eye and
Amplifier (O15) to amplify the detected signal further and minimize electronic noise to obtain output in the form of intensity of light scatter (ILS).

The system (S) of the invention also comprises of a bio amplifier module (E) that comprises of an EOG/ bio amplifier device (E1) for measurement of biopotential due to blinking of eyes by said subject (Su), as blink data, that comprises of surface electrodes (E11) and reference electrodes (E12).

The system (S) of the invention also comprises of processing module (P) for processing of data acquired from objective ophthalmologic module (O) comprising of a CPU (P1) and a machine learning tool (P2) to include one clustering algorithm or tool (P21). The system (S) also comprises of a user interface module (UI) comprising of a display unit (UI1).

The system (S) of the present invention as described above is characterized with the following features that provide the distinct technical advantage over the prior arts. The bio amplifier device (E1) of the bio amplifier module (E) is interfaced with the objective ophthalmic device (O1) of the objective ophthalmologic module (O) for detection and removal of eyeblink artifacts from aqueous flare measurements. The interfacing of said bio amplifier device (E1) and said objective ophthalmic device (O1) takes place at said data acquisition submodule (O2) by simultaneous detection and recording of said blink data from bio amplifier device (E1) and ILS data from objective ophthalmic device (O1) to obtain blink induced aqueous flare measurements. Said blink data is captured as vertical EOG by said bio amplifier device (E1) during simultaneous measurement of said ILS data by said objective ophthalmologic device (O1) and subsequent application of machine learning tool (P2) of processing module (P) is performed offline to said blink induced aqueous flare measurements to remove blink artifacts. Said machine learning tool (P2) comprising of a clustering algorithm or tool (P21) is pretrained to detect blink artifacts in aqueous flare measurements. Said aqueous flare measurements after removal of blink artifacts by machine learning tool (P2) of processing module (P) are made available on display unit (UI1) for user (U) interaction and analysis. Said objective ophthalmologic device (O1) integrated with slit lamp biomicroscope provides an EOG based ophthalmologic device for detection and removal of blink artifacts from aqueous flare measurement.

The invention provides a cost effective and reliable system (S) for detection and removal of blink artifacts from aqueous flare measurements to assist health care professionals in accurate grading of uveitis besides other applications in the domain of ocular diseases.

According to an embodiment of the present invention, the objective ophthalmological module of the system (S) is an upgraded objective ophthalmological apparatus. The module comprises of a regular objective ophthalmological device an ocular flare meter that can measure the aqueous flare as an index of light scatter from the eye and is customized and designed in the present invention to integrate with a slit-lamp biomicroscope. The ophthalmological device can be selected from an LFP (laser flare photometer), an OFAM (ocular flare analysis meter), a fluorometer etc. that measures the light scattering with a data acquisition submodule (DAQ). Additionally, the halogen lamp of the slit-lamp biomicroscope is replaced is replaced with light source selected from UV light source, laser, LED lamp, white LED etc., preferably white LED that preferably has a power output of at least 10 watts. Thus, the light source of the slit lamp may be in the form of a white light-emitting diode (LED) that has a power output of at least 10 watts. The photodetector (O14) Is selected from photomultiplier tube (PMT) , avalanche photodiode etc..

An excited light such as UV light (O111) in the form of illumination beam from the UV light source (O1) is focused on the subject’s eye. The light scattered from the subject’s eye is collected by the photodetector (O14) that is placed at a specific angle such as 450 to the illumination beam. Further, the data in the form of intensity of light scatter (ILS) is recorded by connecting the photodetector (O14) to a data acquisition submodule(O2). The recorded ILS that is a continuous index of aqueous flare measurement, is an indication of the severity of inflammation in the subject’s eye.

The method for detection and removal of eyeblink artifacts from aqueous flare measurements of a subject (Su) comprises of the following steps. A first data set in the form of intensity of light scatter (ILS) is acquired and recorded by data acquisition submodule (O2) of objective ophthalmologic module (O) that comprises of steps of:
The excited light (O111) is focused in the form of an illumination beam from illumination source (O11) of objective ophthalmologic module (O) towards subject’s eye.
Said excited light (O111) is converted into focused light (O121) for onward passage to the subject’s eye by allowing it to pass through illumination slit (O12) placed between said subject’s eye and said illumination source (O11).
Said scattered light (O122) from subject’s eye is detected, to obtain output in the form of intensity of light scatter (ILS), by photodetector (O14) placed at an angle in the range of 35° to 90° preferably at 45° or 90° to the illumination beam.
Said output of photodetector (O14) is amplified using amplifier (O15) for amplifying the ILS and minimization of the electronic noise.
to obtain output in the form of intensity of light scatter (ILS).
Said intensity of light scatter (ILS) is acquired and recorded as first data set by said data acquisition submodule (O2) of said objective ophthalmologic module (O).

As per the method of the invention a second data set in the form of blink data of said subject (Su) is acquired and interfaced simultaneously with said first data set, by data acquisition submodule (O2) of objective ophthalmologic module (O), to obtain blink induced aqueous flare data. This in turn comprises of steps of:
The biopotential due to blinking of eyes by said subject (Su) is captured as blink data, as vertical EOG by bio amplifier device (E1).
Said blink data is interfaced with said first data set at said data acquisition submodule (O2) of said objective ophthalmologic module (O) to obtain blink induced aqueous flare data.
Subsequently as er the method of the invention said blink induced aqueous flare data, obtained from data acquisition submodule (O2) is processed by processing module (P), using clustering algorithm (P21) or tool of machine learning tool (P2) to remove blink artifacts. Said aqueous flare measurements after removal of blink artifacts by machine learning tool (P2) of processing module (P) is made available on display unit (UI1) for user (U) interaction analysis. The bio amplifier module (E) of the present invention comprises of bio amplifier that is an electrooculogram (EOG) bio amplifier selected from electrooculogram (EOG)- BioAmp XG pill, or any instrumentation amplifier. The eye blink artifacts in the light scattering measurements are detected by interfacing the bio-amplifier (E) to the objective ophthalmologic module (O) as shown in Figure 1.

Blinking of the human eye generates an electrical potential called the cor-neoretinal potential (CRP), which represents the EOG signal. The detection of eye blinks involves placement of two surface electrodes (E11) above and below the subject’s eye, together with an additional electrode (E11) positioned behind the ear to serve as a reference electrode (E12). The surface electrodes (E11) are selected from a group of surface electrodes (E11) that measure or cause electrical activity in the tissue under the skin, such as but not limited to disposable ECG gel electrodesetc. When a subject (Su) or a patient blink, biopotentials are produced due to the contraction of the orbicular oculi muscle. These potentials are amplified using the bio-amplifier (E1), which is further interfaced with the same data acquisition submodule (O2) of an ophthalmologic device (O1) to record the EOG blinks. Therefore, whenever the subject (Su) or patient blinks the variance induced in the light scatter measurements (or aqueous flare) due to the eye blinks is detected by the EOG-based ophthalmologic device.

An average eye blink signal will have an amplitude in the range of 3 to 5 mV, with a frequency in the range of 0.5 to 19 Hz. The noise in the eye blink signals is attenuated by applying a 4th order band-pass filter at a frequency range of 0.5-19Hz. The data acquisition submodule (O2) is a single data acquisition device that records eye blink and ILS measurements simultaneously.

The processing module (P) comprises of CPU ( P1), machine learning tool (P2),
clustering algorithm or tool (P21), etc. The ILS measurements received from the objective ophthalmological module (O) of the system (S) are processed in the CPU(P1), utilizing a machine learning tool (P2) offline. The machine learning tool (P2) such as but not limited to unsupervised learning models or tools ) selected from group of clustering algorithm or tool (P21) such as k-means, agglomerative, and Gaussian mixture clustering are utilized to eliminate the blink artifacts from the contaminated aqueous flare data. These clustering algorithms or tools (P21) are pretrained on corrupted data to effectively remove unwanted noise. The pretraining of the algorithm comprises of applying the blink-induced uveitis ILS-WB data (U-ILS-WB) and ILS data of healthy subjects (i.e., N-ILS-WB) to train machine learning tool (P2) for detecting and removing blinks in aqueous flare measurements as described in the Examples in the following paragraphs.

Figure 2 displays a flowchart illustrating the process of removal of blink artifacts : The unsupervised models cluster the synthetic uveitis data (US-ILS-WB) and the normal subject's data (NE-ILS-WB), selecting the cluster with the highest silhouette score. The clusters identified as WB are removed and the models' performance is evaluated. The model and the number of clusters at which the highest performance is attained are taken into consideration.

Unsupervised k-means clustering is used to group acquired blink and ILS data to distinguish aqueous flare with and without blink. After cleaning the data, the elbow approach selects the ideal number of clusters (k). The Elbow method's primary principle is that as the number of clusters (k) is increased, the within-cluster sum-of-squares (WCSS) or average distance between clusters is determined.

The elbow point is seen by the sharp decline in WCSS at the ideal k. The optimal k value chosen by the Elbow approach is also validated by the Silhouette scoring metric. Further, the ILS and blink data are grouped based on the selected k value.
These clusters are further classified as with and without blinks by considering larger and smaller k values as with and without blinks, respectively. Furthermore, the clusters of anticipated or projected blink artifacts are eliminated, and a performance study is carried out. The unsupervised clustering model or tool and the corresponding cluster number at which the maximum signal-to-noise ratio (SNR) and reduced mean squared error (MSE) is attained is considered for the removal of blink artifacts. Therefore, the ILS data (Figure 2) clusters with blinks are discarded to obtain the blink artifact-free aqueous flare measurements.

The pretraining of machine learning tool (P2) comprises of applying the blink-induced uveitis ILS-WB data (U-ILS-WB) and ILS data of healthy subjects (i.e., N-ILS-WB) to train machine learning tool (P2) for detecting and removing blinks in aqueous flare measurements.

The output in the form of blink artifact free aqueous flare data obtained from the processing module (P) is available in the display unit (UI1) of the interface module (U) of the system (S).

The system (S) of the present invention has the distinctive feature of simultaneous measurement of eye blinks and aqueous flare, followed by the k means clustering approach to separate the flare with and without blinks and clean up the aqueous flare data. Interfacing the EOG device with the objective ophthalmic device has been developed for the first time to analyze the blink-induced artifacts in the ILS measurements. The current state of the art systems in this domain are yet to verify the blink artifacts while measuring the aqueous flare. Therefore, by removing blink artifacts efficiently from aqueous flare measurements, the present invention is able to assist in accurate grading of anterior uveitis inflammation.

In particular, as per an embodiment of the present invention, the ophthalmologic device is integrated with a regular slit-lamp biomicroscope. Another characteristic feature of the present invention is the replacement of halogen light source of the slit lamp with a white light-emitting diode (LED) with a power output of at least 10 watts.

As per an embodiment, the system (S) of the present invention is distinctively designed to capture blink signal as vertical EOG during ILS measurements and then applying unsupervised learning methods to deconvolve blink artifacts. Additionally, the ILS is cleansed by excluding the cluster identified as blinks. The assessment focuses on the substantial decrease in the variation of the aqueous flare when blink artifacts are eliminated. The methodology of the present invention effectively mitigates artifacts resulting from blinking and enables accurate grading of uveitis using ophthalmologic device measurements. This technical advancement thus has the potential to be utilized for enhanced diagnosis of mild to moderate forms of uveitis and a further reduction in the necessary medication dosage.

Further the present invention removes blink artifacts from the recorded ILS data by applying an ML-based unsupervised clustering tool to predict ILS with and without blinks. The accuracy of predicting ILS with blinks using k-means clustering was > 85 %. The variance in the ILS is significantly reduced after the removal of blink-induced artifacts. Therefore, blink-induced artifacts in the ILS measurements are rectified, leading to precise grading of uveitis inflammation.
EXAMPLES
The present invention shall now be explained with accompanying examples. These examples are non-limiting in nature and are provided only by way of representation. While certain language has been used to describe the disclosure, any limitations arising on account of the same are not intended. As would be seeming to a person skilled in the art, various working alterations may be made to the method in order to implement the inventive concept as taught herein. The figures and the preceding description give examples of embodiments. Those skilled in the art will appreciate that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be split into multiple functional elements. Elements from one embodiment may be added to another embodiment. For example, order of steps of method or processes of data flow described herein may be changed and is not limited to the manner described herein. Moreover, the actions of any flow diagram need not be implemented in the order shown; nor do all of the acts need to be necessarily performed. Also, those acts that are not dependent on other acts may be performed in parallel with the other acts. The scope of embodiments is by no means limited by these specific examples.

In an exemplary embodiment, the various hardware, devices and software which together form the system (S) along with the working of the invention and the method thereof are illustrated below.

Hardware details
Objective ophthalmologic device (O1) includes ocular flare meter such as but not limited to fluorometer LFP or OFAM.
Slit lamp biomicroscope such as but not limited to Ex: Zeiss or Haag Streit slit-lamp.
Bio-amplifier such as but not limited to BioAmp EXG Pill.
Electrodes utilized are surface electrodes (E11) such as but not limited to disposable ECG gel electrodes.
CPU (P1) - such as but not limited to Pentium 4 G1 (or equivalent) or (64-bit)
Display unit (UI1)
Software details
Machine Learning Tool (P2) - Unsupervised Machine Learning Tool
Clustering algorithm or tool (P21) such as but not limited to k-means clustering, agglomerative hierarchical clustering, Gaussian mixture etc.

The model development of the present invention through research, experimental work etc. and training of the model were performed as described in the following paragraphs.

Blink detection using EOG : As per an embodiment of the present invention the subject's or patients' eye blinks are measured by connecting the EOG - BioAmp EXG Pill to the data acquisition submodule (O2). The biopotentials are detected by positioning the electrodes at precise places on the upper and lower areas of the left eye. In addition, there is an additional electrode placed behind the ear that serves as a reference electrode (E12). The corneal side has a positive charge, while the retinal side has a negative charge. During the downward movement of the upper eyelid, the electrode positioned beneath the eye (linked to the positive input of the bio-amplifier) makes contact with the positive section of the cornea. Conversely, when the upper eyelid is opened, the other electrode (connected to the positive input of the bio-amplifier) comes into contact with the negative section of the retina. Consequently, there exists a difference in voltage between the electrodes. This voltage appropriately represents the blinking. The EOG signal displays signal amplitudes ranging from 250 to 1,000 μV, with a frequency range of around 0 to 30 Hz.

Intensity of light scatter (ILS) measurement using ophthalmologic device: As per an embodiment of the present invention the objective ophthalmologic device (O1) of the objective ophthalmologic module (O) is an ocular flare meter, in particular a spot fluorometer or a laser flare meter, that measures the aqueous flare using the concept of Tindal effect. The ophthalmologic device is specifically designed to integrate with a regular slit-lamp biomicroscope. The halogen light source of the slit lamp was replaced with a white LED or a Laser diode that has a power output of at least 10 watts. The detection slit is aligned in a confocal manner with the illumination slit (O12) at the imaging port of the slit lamp. The output of the photodiode is amplified for amplifying the signal and minimizing electronic noise. The ultimate measurement parameter is acquired by connecting to the DAQ system (O2), with the measurement being expressed in millivolts.

Preparation of data of healthy individuals: In the present invention, empirical data was collected by quantifying the blink-induced and blink-artifact-free ILS in healthy individuals. In addition, unsupervised machine learning techniques including k-means clustering, agglomerative hierarchical clustering, and Gaussian mixture clustering were used to identify blink artifacts in both the healthy and synthetic uveitis data. It was observed that the Gaussian mixture model outperformed other models in predicting blink-induced ILS, resulting in the most substantial decrease in blink-artifacts. Furthermore, the ILS was successfully resolved by using artifact removal technique of the present invention resulting in a notable accuracy rate of 89%.

All the subjects were prepared by placing the electrodes above and below the right eye and one more electrode behind the ear as a reference. The subjects were instructed not to blink while recording blink artifact-free (or without blink - WOB) ILS. While recording blink-induced ILS (or with blink- WB), the subjects were instructed to blink at their own pace. Further, the optometrist focused the illuminating beam on the eye and recorded ILS by continuously clicking the recording button for a minute. The data was collected for each individual, measuring WB and WOBs in total, around 1200 data points were gathered within a minute.

Figure 4 illustrates Scatter plot of ILS and EOG with and without blinks: An episode recording of a subject without blink and blink-induced aqueous flare (ILS) using an EOG interfaced ophthalmologic device at two amplifier time constants 10ms (A, C) and 100ms (B, D), respectively. The artifact-free ILS at 10ms and 100ms (0.1426 ± 0.0499, 0.1084 ± 0.0361) values are smaller than blink-induced ILS (0.1310 ± 0.0971,0.1518 ± 0.1331), respectively, indicating low to high variance.

Removal of blink artifacts from ILS measurements : The fluctuation due to the eye blink in the ILS measurements has to be minimized to achieve a reliable grading of uveitis inflammation. Manually detecting the blink-induced ILS among ~ 14000 data points (12 participants’ x 1200 data points) will be difficult and produce inaccuracies. Therefore, the data is classified by utilizing unsupervised machine learning tools (P2) such as but not limited to k-means clustering, agglomerative hierarchical clustering, and Gaussian mixture clustering for categorizing the ILS data into categories with and without blinks. The k-means being simple to use and that reduces the sum-of-squares distance between data points and cluster centres, is frequently selected as the clustering model. Agglomerative clustering has a notable feature of a hierarchical structure, that provides a new perspective on the data's structure by methodically merging clusters according to their proximity. Also, Gaussian mixture models are a good way to show the different types of data patterns. These models create complex data distributions by combining Gaussian components. For both blink and non-blink ILS evaluations, these tools were selected for their scalability, probabilistic clustering capabilities, hierarchical modeling, and simplicity.

Before clustering, the data is preprocessed by setting the standard deviation to one and the mean to zero using the Scalerfit() method. This method effectively handles outliers while maintaining the data distribution. Furthermore, the elbow technique and Silhouette score are used to calculate the best number of clusters (n). The fundamental concept of the elbow technique is to calculate the within-cluster sum-of-squares (WCSS) or average distance between clusters as the number of clusters (n) increases. The elbow point is determined by the sudden reduction in the WCSS at the optimal value of n. Utilizing the Silhouette scoring system further confirms the selected ideal n value. The number of clusters for which the silhouette scoring score was highest (close to 1) was considered for further analysis.

Figure 5 illustrates magnified view of a three clustered single ILS-blink signal before and after artifact removal: (A, B) The clustering of the blink-induced ILS at both amplifier time constants by applying k-means, agglomerative, and Gaussian clustering models is represented in three colors: navy blue, lavender, and pink, respectively. The mean ILS-WB at 10 ms is 0.1310 ± 0.0971, and at 100 ms is 0.1518 ± 0.1331. (C, D) The blink artifact-free ILS at both amplifier time constants 10 ms and 100 ms is achieved by removing the WB clusters (i.e., at 10 ms plot: Gaussian labels ‘1’ and ‘2’ are discarded and, at 100 ms plot: Gaussian and agglomerative labels ‘0’ and ‘1’ are discarded), and the result cleaned ILS is 0.1153 mV and 0.1094 mV, respectively.

Figure 6 illustrates mean ILS of normal subjects and synthetic uveitis patients’ data WB and WOB artifacts: (A) Normal subjects ILS with and without blinks. (B) Synthetic uveitis WOB values are nearly identical to experimental uveitis data (P < 0.6955). The inclusion of blink artifacts in the normal and synthetic uveitis data led to an increase in the variance of the ILS. (C) The pairwise comparison of the experimental WOB with cleaned WB and the WOB of healthy subjects at 10 ms and 100 ms is calculated using the Mann-Whitney nonparametric t-test. The experimental WOB data is moderately statistically different from the cleaned WB data at 10ms and 100ms (P = 0.1363 and 0.1263). In contrast, the cleaned WOB is statistically different from the original WOB, represented by lower P values of 0.0008 and 0.0216, respectively. (D, E, F) The actual uveitis and cleaned data for all SUN grades showed no significant difference, with P values of 0.1439, 0.1193, and 0.1322, respectively.
With the chosen cluster number, the simulated uveitis data (i.e., time series corresponding to U-ILS-WB) of all uveitis patients for a given SUN grade were merged and subjected to clustering k-means, agglomerative hierarchical clustering, and Gaussian mixture clustering. Similar analysis was performed on the data from healthy subjects (i.e., time series corresponding to N-ILS-WB of all subjects). In the classified data, a cluster with the lowest label value was designated as data WOB, and those with higher values were assumed to be data WB. For example, labels 1 and 2 for a three-cluster dataset would be considered WB; the remaining cluster 0 with the lowest value was designated WOB. The average ILS value for a specific cluster model and the corresponding cluster number at which the greatest reduction is attained are recorded in a table 3 as presented below. The artifact reduction approach adopted by the present invention was tested on both blink-induced (i.e., N-ILS-WB and U-ILS-WB) and blink-artifact-free (N-ILS-WOB) data, as some subjects couldn't resist blinking while recording WOB data, violating the protocol for measuring blink artifact-free ILS using an ophthalmologic device.

Table 3: List of optimal cluster numbers for various model types and the accompanying enhanced ILS values after eliminating with-blink clusters
Subject/Patient Amplifier (Tc) ILS-WOB (mV) ILS-WB (mV) # Clusters Type of clustering model ILS-Cleaned (mV)
Normal (WOB) 10 ms 0.254 0.254 3 Agglomerative 0.115
100 ms 0.171 0.171 2 k-means, Gaussian 0.10
Normal (WB) 10 ms 0.254 0.463 3 Agglomerative, Gaussian 0.269
100 ms 0.171 0.556 4 k-means, Agglomerative 0.171
SUN 0 10 ms Nil 0.355** 2,4 Gaussian 0.164
100ms 0.159** 0.321** 4,3 k-means, Agglomerative, Gaussian 0.160
SUN 1+ 10 ms Nil 0.434** 4 K-means, Agglomerative, Gaussian 0.262
100 ms 0.205 ** 0.478* 4 Gaussian 0.206
SUN 2+ 10 ms Nil 0.643** 2,3,4 Gaussian 0.438
100 ms 0.431** 0.669** 3,4 Gaussian 0.443

The blink-artifact-free ILS is validated by assessing the performance of the applied unsupervised tools to predict the flare WBs. The cleaned signal is compared with the original or contaminated signal to evaluate the error. The measures of performance are relative error (RE), mean squared error (MSE), signal-to-artifact ratio (SAR), signal-to-noise ratio (SNR), and correction percentage [32]. The formula for these matrices is given below:
RE=‖Artifact free ILS signal-cleaned ILS signal‖/‖Artifact free ILS signal‖ (1)
MSE=1/n ∑_(i=1)^n▒〖(C〖leaned signal〗_i 〗-〖Contaminated signal〗_i) (2)
SAR=(σ (contaminated signal))/(σ(contaminated signal-cleaned signal)) (3)
〖SNR〗_dB= 10 〖log〗_10 (1/n ∑_i^n▒〖[〖Contaminated signal (i)]〗^2 〗)/(1/n 〖∑_i^n▒[ Contaminated signal (i)-Cleaned signal (i)]〗^2 ) (4)
Correction Percentage= ((Cleaned Value-Contaminated Value)/Contaminated Value)×100 (5)

The value of n in MSE (Eq. 2) and SNR (Eq.4) indicates the length of the data points. And the σ in the SAR (Eq. 3) represents the mean of the respective signals. The SNR represented in Eq. 4 is in evaluated in the units of decibels (dB). Alternatively, SNR can also be determined by taking the ratio of the mean to the standard deviation (STD) of a signal (i.e., SNR_artifact or SNR_cleaned= µ/σ or Mean/STD). In order to demonstrate the difference in the SNR values of the contaminated signal (ILS with-blinks) to the cleaned signal, the former formula was also utilized and tabulated in Table 4 as below.

Table 4: Performance analysis summary for normal subjects and synthesized uveitis data
Subject/
Patient Amp.(Tc) ILS-WOB (mV) ILS-WB
(mV) ILS after correction (mV) RE MSE SAR SNR
(dB) SNR_
Artifact SNR_
Cleaned Correction percentage (%)
Normal (WOB) 10 ms 0.254 0.254 0.1159 0.425 0.249 0.897 0.542 0.672 0.824 88.4
100 ms 0.171 0.171 0.1040 0.776 0.018 4.070 6.017 1.186 0.487 89.6
Normal (WB) 10 ms 0.254 0.463 0.2691 1.042 2.772 9.993 0.320 0.278 1.295 73.1
100 ms 0.171 0.556 0.1714 0.590 1.499 9.028 0.450 0.478 2.896 82.9
SUN 0 100 ms 0.159** 0.321** 0.1605 0.674 0.003 0.990 9.592 2.863 23.456 83.9
SUN 1+ 0.205** 0.478** 0.2068 0.906 0.007 0.992 8.199 2.378 42.537 79.3
SUN 2+ 0.431** 0.669** 0.4433 1.137 0.038 0.976 7.690 2.239 12.668 55.8

The experimental subjects' data without blinks aided as the artifact-free signal, enabling them to evaluate the tools' performance by comparing it with the cleaned signal obtained after removing blink-induced ILS. Unsupervised tools were utilized on WOB data to eliminate unintentional blink-induced variance in ILS measurements, specifically in routine measurements using ophthalmologic device . The evaluation of the performance analysis for synthetic uveitis data was compared to referential data that was obtained using a constant time interval of 100ms. Ultimately, the correction % is evaluated for each data type to quantify the degree of correction accomplished by the chosen clustering model in predicting the blink-induced flare.
, Claims:We claim:
1. A system (S) for detection and removal of eyeblink artifacts from aqueous flare measurements of a subject (Su) wherein said system (S) comprises of
- at least one objective ophthalmologic module (O01, O02,.. On), said objective ophthalmologic module (O) comprising of
• at least one objective ophthalmologic device (O101, O102,.., O1n) integrated with slit lamp biomicroscope for detection of intensity of light scatter (ILS) from the subject’s eye, as an objective measure of aqueous flare, said objective ophthalmologic device (O1) comprising of
o illumination source (O11) for emitting an illumination beam of excited light (O111) towards subject’s eye,
o illumination slit (O12) placed between said subject’s eye and said illumination source (O11) to convert said excited light (O111) into focused light (O121) for onward passage to the subject’s eye,
o photodetector (O14), for detecting scattered light (O122) from subject’s eye as output and
o amplifier (O15) for amplification of said output of photodetector (O14) and to minimize electronic noise to obtain output in the form of ILS,
• at least one data acquisition submodule (O201, O202,.. O2n) connected to said photodetector (O14) and amplifier (O15) for acquisition and recording of data in the form of said ILS from objective ophthalmologic device (O1),
- at least one bio amplifier module (E01, E02,.., En), said bio amplifier module (E) comprising of
• at least one EOG/ bio amplifier device (E101, E102,.., E1n) for measurement of biopotential due to blinking of eyes by said subject (Su), as blink data, said bio amplifier device (E1) comprising of

o surface electrodes (E1101, E1102,.., E11n) and
o reference electrodes (E1201, E1202,.., E12n),
- at least one processing module (P01, P02,.., Pn) for processing of data acquired from objective ophthalmologic module (O), said processing module (P) comprising of
• at least one CPU (P101, P102,.., P1n), and
• at least one machine learning tool (P201, P202,.., P2n) comprising of at least one clustering algorithm or tool (P2101, P2102,…P21n) and
- at least one user interface module (UI01, UI02,.., UIn), said user interface module comprising of at least one display unit (UI101, UI102,.., UI1n)

wherein
- the bio amplifier device (E1) of the bio amplifier module (E) is interfaced with the objective ophthalmic device (O1) of the objective ophthalmologic module (O) for detection and removal of eyeblink artifacts from aqueous flare measurements,
- said interfacing of said bio amplifier device (E1) and said objective ophthalmic device (O1) takes place at said data acquisition submodule (O2) by simultaneous detection and recording of said blink data from bio amplifier device (E1) and ILS data from objective ophthalmic device (O1) to obtain blink induced aqueous flare measurements,
- said blink data is captured as vertical EOG by said bio amplifier device (E1) during simultaneous measurement of said ILS data by said objective ophthalmologic device (O1) and subsequent application of machine learning tool (P2) of processing module (P) performed offline to said blink induced aqueous flare measurements to remove blink artifacts,
- said machine learning tool (P2) comprising of at least one clustering algorithm or tool (P21) is pretrained to detect blink artifacts in aqueous flare measurements,
- said aqueous flare measurements after removal of blink artifacts by machine learning tool (P2) of processing module (P) are made available on display unit (UI1) for user (U) interaction and analysis and
- said objective ophthalmologic device (O1) integrated with slit lamp biomicroscope provides an EOG based ophthalmologic device for detection and removal of blink artifacts from aqueous flare measurement
that provides a cost effective and reliable system for detection and removal of blink artifacts from aqueous flare measurements to assist health care professionals in accurate grading of uveitis besides other applications in the domain of ocular diseases.
2. The system (S) as claimed in claim 1, wherein said objective ophthalmologic device (O1) that is integrated with slit lamp biomicroscope for detection of intensity of light scatter (ILS) from the subject’s eye is an ocular flare meter that is selected from an LFP (laser flare photometer), an OFAM (ocular flare analysis meter), a fluorometer etc.
3. The system (S) as claimed in claim 1, wherein said illumination source (O11) that is originally a halogen lamp in slit lamp microscope is replaced with light source selected from UV light source, laser, LED lamp, white LED etc., preferably white LED.
4. The system (S) as claimed in claim 1, wherein the light that is scattered by subject’s eye is detected by a photodetector (O14) that is selected from photomultiplier tube (PMT), avalanche photodiode etc. preferably an avalanche photodiode.
5. The system (S) as claimed in claim 1, wherein said photodetector (O14) is placed at an angle in the range of 35° to 90°, preferably 45° or 90° to said illumination beam for detecting said scattered light (O122) from subject’s eye.
6. The system (S) as claimed in claim 1, wherein said output of the photodetector (O14) is amplified using said amplifier (O15) while minimizing electronic noise and thereby enhancing signal-to-noise ratio and dynamic range of the ILS.
7. The system (S) as claimed in claim 1, wherein said bio amplifier device (E1) is utilized to measure blink data and is selected from any bio-amplifier, such as single-channel BioAmp EXG pill.
8. The system (S) as claimed in claim 1, wherein said machine learning tool (P2) comprises of an unsupervised k-means clustering tool selected from k-means, agglomerative, Gaussian mixture clustering preferably Gaussian mixture clustering.
9. A method for detection and removal of eyeblink artifacts from aqueous flare measurements of a subject (Su) wherein said method comprises of steps of
- acquiring and recording of first data set in the form of intensity of light scatter (ILS) by data acquisition submodule (O2) of objective ophthalmologic module (O) comprising of steps of
• focusing excited light (O111) in the form of an illumination beam from illumination source (O11) of objective ophthalmologic module (O) towards subject’s eye,
• converting said excited light (O111) into focused light (O121) for onward passage to the subject’s eye by allowing it to pass through illumination slit (O12) placed between said subject’s eye and said illumination source (O11),
• detecting scattered light (O122) from subject’s eye by photodetector (O14) placed at an angle in the range of 35° to 90° preferably at 45° or 90° to the illumination beam, to obtain output,
• amplifying said output of photodetector (O14) using amplifier (O15) while minimizing electronic noise to obtain output in the form of intensity of light scatter (ILS) and
• acquiring and recording of intensity of light scatter (ILS) as first data set by said data acquisition submodule (O2) of said objective ophthalmologic module (O),
- acquiring simultaneously second data set in the form of blink data of said subject (Su), and interfacing of said second data set with said first data set, by data acquisition submodule (O2) of objective ophthalmologic module (O), to obtain blink induced aqueous flare data, comprising of steps of
• capturing of biopotential due to blinking of eyes by said subject (Su), as blink data, in the form of vertical EOG by bio amplifier device (E1) and
• interfacing said blink data with said first data set at said data acquisition submodule (O2) of said objective ophthalmologic module (O) to obtain blink induced aqueous flare data,
- processing of said blink induced aqueous flare data, obtained from data acquisition submodule (O2), by processing module (P), using clustering algorithm (P21) or tool of machine learning tool (P2) to remove blink artifacts and
- making available of said aqueous flare measurements after removal of blink artifacts by machine learning tool (P2) of processing module (P) on display unit (UI1) for user (U) interaction and analysis.
10. The method as claimed in claim 9 , wherein said capturing of biopotential due to blinking of eyes by said subject (Su), as blink data, in the form of vertical EOG by bio amplifier device (E1), comprises of steps of
- placing of two surface electrodes (E11) above and below one of the subject’s eyes, together with an additional electrode positioned behind the subject’s ear, to serve as a reference electrode (E12),
- detecting biopotential due to eye blinks of said subject (Su) and
- amplifying of said biopotential to obtain blink data using said bio-amplifier device (E1).
11. The method as claimed in claim 9, wherein said method of processing of said blink induced aqueous flare data obtained from data acquisition submodule (O2) of objective ophthalmologic module (O) using said clustering algorithm (P21) or tool of pretrained machine learning tool (P2) comprises of steps of
- cleaning of said acquired blink data and said ILS data by preprocessing,
- selecting the ideal number of clusters (k) as per elbow approach to determine within-cluster sum-of-squares (WCSS) or average distance between clusters,
- determining the elbow point from the sharp decline in WCSS at the ideal k,
- validating the optimal k value chosen by the elbow approach by the silhouette scoring metric,
- grouping of said ILS and said blink data based on the selected k value,
- classifying said clusters by considering larger and smaller k values as data with and without blinks, respectively and
- discarding of the ILS data clusters with blinks to obtain the blink artifact-free aqueous flare measurements.
12. The method as claimed in claim 11, wherein said preprocessing is performed to remove noise in the blink data and is attenuated by applying a 4th - order band-pass filter at a frequency range of 0.5 - 19Hz.

13. The method as claimed in claim 11, wherein said pretraining of machine learning tool (P2) comprises of applying uveitis ILS-WB data (U-ILS-WB) and ILS data of healthy subjects (i.e., N-ILS-WB) to train machine learning tool (P2) for detecting and removing blinks in aqueous flare measurements.

14. The system (S) as claimed in claim 1, wherein said system (S) enables the health care providers in accurate grading of uveitis objectively and for assessing other ocular diseases and the effectiveness of their respective treatments.
Dated this the 20th day of December 2024
_____________________
Daisy Sharma
IN/PA-3879
of SKS Law Associates
Attorney for the Applicant

To
The Controller of Patents,
The Patent Office, Chennai