An estimation method for intraocular pressure

Abstract

The invention is related to a system and method for measuring intraocular pressure using the image sequences of human eyes, and analyzing the blood flow characteristics within iris and sclera and relating them with intraocular pressure (IOP) value.

Description

[0001] AN ESTIMATION METHOD FOR INTRAOCULAR PRESSURE

[0002] Technical Field

[0003] A system and method for estimation of intraocular pressure using the image sequences of human eyes, and analyzing the blood flow characteristics within the iris and sclera and relating them with intraocular pressure (IOP) and / or an intracranial pressure (ICP) value.

[0004] Prior Art

[0005] Glaucoma is one of the most frequent causes of irreversible blindness. There are many risk factors for Glaucoma disease such as intraocular pressure, neurodegeneration and vascular insufficiency. Controlling the intraocular pressure and lowering the pressure on the optical nerve are crucial during the progression of the disease. Intraocular pressure is one of the most important and controllable risk factors. Aqueous humor that creates intraocular pressure is generated by the ciliary body. Unbalance between the generation of the aqueous humor and drainage to the out of the eye globe causes an increase in intraocular pressure. Long term increase of the intraocular pressure causes damage on the optical nerve and, consequently, glaucomatous vision loss.

[0006] Approximately, 50% of the glaucoma patients are not aware of their disease. The most important reason is that there are no symptoms until the end stage of the disease. After the diagnosis of the glaucoma, patients start to use different drugs with many side effects. During the drug use of a glaucoma patient, it is really important to monitor the variation of the diurnal intraocular pressure. Monitoring the intraocular pressure guides the treatment.

[0007] Today, intraocular pressure monitoring is possible at home or in hospital with methods that require contact with the eye globe. The devices that can be used at home have several problems such as contact with the eye and they are difficult-to-use, requiring long training to get correct measurements.

[0008] Additionally, monitoring intraocular pressure at home without requiring additional equipment is crucial. In the world, it is estimated that there are 80 million glaucoma patients; however almost half of them are not diagnosed. This problem leads to irreversible blindness. Taking these into consideration, developing non-contact, wearable, and easy-to-use intraocular measurement methods / devices are crucial for the early diagnosis and following / tracking / monitoring the disease progression.

[0009] Current systems for measuring IOP use either air-puff or a palpation on the eye globe. These methods have disadvantages such as; difficulty in home use, infection risk, possible cornea damage, limited use for yielding diurnal IOP trends.

[0010] The traditional methods like the Goldmann applanation tonometer (GAT) is still the clinical reference for IOP measurements. However, the measurement procedure is uncomfortable and carries infection risks through direct eye contact. Due to negative effects of the required anaesthetics and the mechanical load on the cornea, the number of daily measurements is limited to about 5. This method is therefore of limited use for yielding diurnal IOP trends which is very important for Glaucoma patients. It also requires a physician for the measurement, which limits the use of GAT.

[0011] These drawbacks also apply for the Schiotz Tonometer (e.g. Rudolf Riester GmbH, Jungingen, Germany), where the measurement uncertainty is larger than with GAT. The Pascal dynamic contour tonometer (Ziemer Ophthalmic Systems AG, Port, Switzerland), which uses a concave contact area that indents the cornea only a few microns and measures the remaining corneal tension with a pressure sensor, is less prone to systematic effects of the corneal thickness, but still requires anaesthetics and a physician for the measurement.

[0012] Ocuton S tonometer (EPS a GmbH, Saalfeld, Germany) can be considered a self-tonometer, which is based on the same measurement principle as the GAT. The disadvantage of the Ocuton S is the high measurement uncertainty of up to 10 mm Hg that does not allow acquiring reliable diurnal IOP trends and the contact to the cornea which brings infection risks.

[0013] In summary, currently available / known methods as a contact tonometer are not suitable for home use and self-measurements. iCare HOME ® tonometer is a device available to patients for measuring intraocular pressure at home. However, it uses rebound tonometry technique (requires contact with the cornea), suffering from the disadvantages of GAT’s disadvantages. Additionally, despite its advantages of home -based tonometry, not everyone may be able to accurately use the device for self- tonometry. Some previous studies have reported 16-25% of patients having difficulties completing training.

[0014] TonoPen® is a handheld device that enables home use for intraocular pressure measurements. However, it requires a contact with the cornea for performing measurement.

[0015] Triggerfish ® Contact Lens Sensor (CLS) was designed for non-invasive, continuous ocular telemetry. The CLS is designed to be worn for a continuous 24-hour period to determine IOP fluctuations in an ambulatory setting. Different adverse events were observed with the use of Triggerfish such as transient blurred vision, conjunctival hyperemia, superficial punctate keratitis which then caused discontinuation of the device use.

[0016] EyeMate® is a permanent, implantable IOP sensor, which is inserted into the ciliary sulcus behind the iris and in front of an artificial lens during cataract surgery. It provides permanent monitoring of intraocular pressure. However, patients do not prefer an invasive procedure for implanting the device. Additionally, some studies show that it causes deterioration of visual fields.

[0017] WO2017132165A1 discloses systems and methods are described for non-invasively monitoring and / or measuring or estimating intraocular pressure. Medical or diagnostic methods embodiments described herein include high resolution imaging of the sclera of one or both of a patient's eyes using digital photography or videography. The hardware employed may be for two-dimensional (2D) or three-dimensional (3D) imaging. In this method, the system observe changes on the eye globe and this method requires a reference value that is measured with standard methods in first place.

[0018] In US6673014B2, non-invasive methods and apparatuses measuring the intraocular pressure (IOP) of the eye using vibratory excitation are disclosed. The method based on assuming that the natural frequency of the eye changes with intraocular pressure. Furthermore, this system uses google to create air pressure on the eyeball.

[0019] US7403805B2 aims that measuring the chemical changes and chemical products, blood components and parameters with a mid-infrared detecting method.

[0020] US10386640B2 teaches a method to create an indent in the eye with an air-puff and then they measure the intraocular pressure using the OCT imaging system. As a result, all of the problem mentioned above has made it necessary to provide a novelty in the related field.

[0021] Objects of the Invention

[0022] The main object of the present invention is to establish a non-contact (and no air-puff) estimation method and system for an intraocular pressure (IOP) and / or an intracranial pressure (ICP).

[0023] Another object of the present invention is to establish a non-contact (and no air-puff) estimation method and system for intraocular pressure (IOP) and / or an intracranial pressure (ICP).

[0024] Thus, this invention prevents infection risks and damages on cornea.

[0025] Another object of the present invention is to establish a method and system without requirement of consumable parts, drops, expert help and enabling home-use and continuous and / or multiple measurements during the day (diurnal intraocular pressure (IOP) and / or an intracranial pressure (ICP) monitoring).

[0026] Another object of the present invention is to establish method and system which patients do not need to fix the device that consists of a camera with respect to their eyes.

[0027] Brief Description of the Invention

[0028] In present method, the changes of an iris and a sclera are observed. This method and system are completely non-contact and non-invasive. It does not include any kind of pressure exposure around the eye globe.

[0029] To estimate an intraocular pressure (IOP) and an intracranial pressure (ICP), a data comprises at least two sequential images of the at least one eye of a patient is received and regions of interest of the eye are determined on these images and color and / or color-density changes of regions of interest of the eye in the sequential images are determined.

[0030] Preferably, the region of interest includes, iris and sclera, and retina of the eye, and a vasculature around the eye, and preferably the red reflex of the eye

[0031] The method starts with capturing sequential images of the eyes of a human. At least one of the left and right eyes or both are detected in received sequential images. In case that the both eyes are detected, the rest of the process continues identical and independently for the left and right eye for estimating intraocular pressure (IOP) and / or an intracranial pressure (ICP) values for the left and right eye.

[0032] Afterwards, regions of interest will be automatically are defined on a single eye. Then, blood pulsation characteristics are derived from an image sequence. Blood detection algorithm can be a state-of-the-art method that is already explained in the literature.

[0033] The color and / or color-density changes can be used to estimate the intraocular pressure (IOP) and / or an intracranial pressure (ICP). The image processing algorithms, learning algorithm such as a machine learning algorithm can be used such a task.

[0034] For training a learning algorithm such as a machine learning algorithm, a specific dataset is constituted by at least two sequential images of the at least one eye of a patient and intraocular pressure (IOP) and / or an intracranial pressure (ICP) a patient and regions of interest of the eye and color change and / or color-density change of regions of interest of the eye in the sequential images.

[0035] After blood pulsation detection and collecting the data from each image, (machine) learning algorithms can be trained with an intraocular pressure (IOP) and an intracranial pressure (ICP) data that is collected with the devices currently being used. After that trained algorithms are used for relating the blood pulsation characteristics with the intraocular pressure (IOP) and / or an intracranial pressure (ICP). In conclusion, the new eye images are used to estimate the current intraocular pressure. The data collected from each image comprises segmented region of interests, post processed image sequences to detect color change or color density change of the regions and extracted blood pulsation values.

[0036] Furthermore, the data set can be used for training learning algorithms, such machine learning or neural network which estimate intraocular pressure (IOP) and / or an intracranial pressure (ICP).

[0037] Preferably, the region of interest includes, iris and sclera, and retina of the eye, and a vasculature around the eye to estimate intraocular pressure of the eye.

[0038] In that case that intracranial pressure data is estimated, the region of interest also includes light that is reflected from the retina of the eye, namely red reflex of the eye and the intracranial pressure data.

[0039] The red reflex refers to the reddish-orange reflection of light that is observed when examining the eye with an ophthalmoscope or other similar medical devices. It is an important clinical sign used to assess the health of the eye, particularly the posterior segment. The red reflex is produced by the reflection of light off the inner surface of the retina, which contains a layer of specialized cells called the retinal pigment epithelium (RPE) and the blood vessels within the retina. When light enters the eye and reaches the retina, it is partially absorbed by the pigmented cells of the RPE and the underlying choroid layer, while the rest of the light is reflected back. Moreover, retina blood pulsation period can be observed by recording the red reflex density difference by time. Since the central retinal artery is exposed to intracranial pressure and intraocular pressure, red reflex pulsation can give an indication when the intraocular pressure is estimated. Thus, above mentioned estimated IOP data and estimation pulsation of iris and sclera can be combined with the pulsation data of the red reflex (or indirectly blood pulsation characteristics of the retina) in order to estimate the intracranial pressure. If the learning algorithm is used to estimation of the intracranial pressure, the dataset which is used to train learning algorithm comprises red reflex data, accordingly.

[0040] This method accepts the fact that iris and sclera are exposed to different pressures. Sclera of the eye is under the effect of the atmospheric pressure and blood pressure of the internal vasculature. On the other hand, IRIS is exposed to atmospheric pressure, blood pressure and - additionally- intraocular pressure. Taking these into account, we can deduct intraocular pressure by just comparing physiological differences in iris and sclera. These physiological parameters for the assessment of differences may include blood intensity, hemoglobin intensity, blood pulsation, color density change and / or color change over time, and periodic difference image under IR (infrared) light or image under daylight.

[0041] Description of the Figures of the Invention

[0042] The figures and related descriptions necessary for the subject matter of the invention to be understood better are given below.

[0043] Figure 1. A schematic view of a human eye.

[0044] Figure la. A schematic view of artery supplies of a human eye.

[0045] Figure lb. A representative view of arteries around eye globe.

[0046] Figure 2. A flow chart of method described for estimation of the intraocular pressure.

[0047] Figure 3. A flow chart of method described for estimation of the intraocular and intracranial pressure. Figure 4. A schematic view of system for estimation of the intraocular and / or intracranial pressure.

[0048] Reference Numbers

[0049] The parts and components are given in the figures are referenced for the subject matter of the invention to be understood better.

[0050] IR. Iris

[0051] S. Sclera

[0052] D. Device

[0053] I. Imaging mean

[0054] C. Communication unit

[0055] P. Processing device

[0056] RS. Remote server

[0057] CA. Carotid artery

[0058] OA. Ophthalmic artery

[0059] MB. Muscular branch

[0060] CAR. Central artery of the retina

[0061] PAC. Posterior ciliary artery

[0062] ACA. Anterior ciliary artery

[0063] LPC. Long posterior ciliary artery

[0064] SPC. short posterior ciliary artery

[0065] ACV. anterior ciliary vein

[0066] VV. vortex vein

[0067] IOV. inferior orbital vein

[0068] Detailed Description of the Invention

[0069] The invention is related to a system and method for measuring intraocular pressure using the image sequences of human eyes, and analyzing the blood flow characteristics within iris and sclera and relating them with intraocular pressure (IOP) value. Firstly, anatomy of the eyes is needed to be clarify to better understanding for the invention.

[0070] Referring to Figure 1, la and lb; Branches of the ophthalmic artery (OA) that supplies the sclera (S) and iris (IR) can be mentioned as follows:

[0071] Long posterior ciliary arteries (LPC): They are branches of the ophthalmic artery (OA) and pass forward within the eye to reach the ciliary body where they ramify and anastomose with the anterior ciliary arteries (ACA), thus forming the major arterial circle of the iris (IR). The long posterior ciliary arteries (LPC) contribute arterial supply to the choroid, ciliary body, and iris (IR). The long posterior ciliary arteries (LPC) first run near the optic nerve before piercing the posterior sclera (S) near the optic nerve. They pass anterior-ward - one along each side of the eyeball - between the sclera (S) and choroid to reach the ciliary muscle where they divide into two branches which go on to form the major arterial circle of the iris.

[0072] Anterior ciliary arteries (ACA): The anterior ciliary arteries (ACA) are seven arteries in each eye-socket that arise from muscular branches (MB) of the ophthalmic artery (OA) and supply the conjunctiva, sclera (S), rectus muscles, and the ciliary body. The arteries end by anastomosing with branches of the long posterior ciliary arteries (LPC) to form the circulus arteriosus major.

[0073] Central artery of the retina (CAR): The central artery of the retina (CAR) branches off the ophthalmic artery (OA), running inferior to the optic nerve within its dural sheath to the eyeball.

[0074] The other arteries that are surrounding eye: the supratrochlear artery, the supraorbital artery, the lacrimal artery and the infraorbital artery.

[0075] The eyeball having two main tissue which are sclera (S)-conjunctiva tissue and iris (IR) tissue. The iris (IR) is the colored, circular portion of the eye located in front of the lens. It surrounds the pupil, which is the opening in the center of the iris that allows light to enter the eye. The sclera, also known as the "white of the eye," is the tough, fibrous outer layer that forms the protective covering of the eyeball. It is a dense, opaque structure that covers almost the entire visible portion of the eyeball. The sclera (S) provides structural support, maintains the shape of the eye, and protects the delicate internal structures of the eye. Sclera (S)-conjunctiva tissue is supplied by the anterior ciliary artery (ACA), and with the pulsation of this artery, there is a rhythmic color change in this tissue over time.

[0076] Iris (IR) tissue is supplied by both the anterior ciliary artery (ACA) and the posterior ciliary artery (PAC), and the pressures that iris (IR) and sclera (S) are exposed to are different.

[0077] The blood vessels that supply the iris (IR) tissue are exposed to IOP, while the blood vessels that supply the sclera (S) do not experience the IOP. By observing the differences in blood pulsation between the iris and the sclera, the IOP can be estimated. Further, the estimation method also be used for ICP estimation by adding red reflex to this method.

[0078] The intraocular pressure (IOP) and / or an intracranial pressure (ICP) of a patient estimation starts with receiving sequential images. These sequential images comprising image of the at least one of the eyes, preferably two eyes.

[0079] The sequential images are images are following one to other. The images can be both still images or frames of the video. Preferably, the images are provided with their time-stamps or sequential orders.

[0080] The mentioned images include at least upper part of face which at least covers at least one eye and vasculature surrounding the eye. The mentioned method can be adjusted or tuned for the people with different eye colors.

[0081] The image may be captured by an imaging mean (I).

[0082] Preferably, the imagining mean (I) can be selected as camera / sensor types can include - but not limited to - RGB cameras, photoplethysmography sensors, lidar sensors and light emitters, infrared light sensitive cameras, and thermal cameras, stereo cameras, ultrasonic sensors, acoustic sensors.

[0083] Furthermore, the imaging mean can be provided as a user device (D). The user device can be portable and / or hand-held device (D). For example, smart phone or tablet having camera, or a dedicated hardware with imaging capability can be used as the user device (D).

[0084] Alternatively, the imaging mean (I) be a security camera or personal computer camera webcam. The mentioned camera can be infrared light (IR) sensitive. The images taken by imaging means (I) is sent to processing device (P). The processing device (P) is configured to detect regions of interest of the eye in sequential images. The detection can be carried out by image processing, recognition algorithms or machine learning models. Preferably, an iris (IR) and a sclera (S) of the eye and a vasculature around the eye is selected as regions of interest.

[0085] Preferably, image processing methods like edge detection, Hough transform, template matching can be used for detecting regions of interest of the eye. However, this shape detection methods shouldn’t be interpreted as limiting because it is obvious that there are too many known shape detection methods on area of image processing.

[0086] The processing device (P) can be standalone unit or is processor of the device (D). Alternatively, device (D) comprises a communication unit (C) to send images to the processing device (P) which configured as a remote server (RS). Preferably, the communication unit (C) is wireless type.

[0087] After determination of the regions of interest of the eye, color and / or color-density difference is determined for each region of interest in sequential images. More clearly, same area of the sequential images is compared to each other, based on to their color and / or color-density. The detection of color and / or color-density difference can be carried out by image processing, recognition algorithms or machine learning models. Preferably the change is analyzed according to sequence of the image or time of the video.

[0088] The blood pulsation data is obtained by color and / or color-density difference. The change between color and / or color-density represent the blood pulsation on the eye which can be used as an indicator for IOP or ICP estimation.

[0089] The detection of color and / or color-density difference can be analyzed by the image processing, recognition algorithms or machine learning models for IOP or ICP estimation.

[0090] A computer implemented method for estimating IOP comprising steps of Receiving at least two sequential images of at least one eye of a patient and Detecting regions of interest of the eye which are an iris and a sclera of the eye, a vasculature around the eye

[0091] Determining color and / or color-density changes of regions of interest of the eye in the sequential images In a preferred embodiment, the regions of interest of the eye further comprises red reflex of the eye and the data further comprises the intracranial pressure data to estimate intraocular pressure. To determine red reflex blood pulsation, similar techniques as in the previous cases, such as photoplethysmography or RGB image analysis, can be used by implementing different image processing algorithms. Calculation of the red reflex pulsation gives an indication of the retina blood flow pulsation, and indirectly it gives an indication of the intracranial pressure when the subject has the intraocular pressure estimation. Collecting data from iris (IR), sclera (S) and red reflex, the method can deduct the intraocular pressure and intracranial pressure, together.

[0092] A computer implemented method for estimating ICP comprising steps of Receiving at least two sequential images of at least one eye of a patient and Detecting regions of interest of the eye which are an iris and a sclera of the eye, a vasculature around the eye and the red reflex,

[0093] Determining color and / or color-density changes of regions of interest of the eye in the sequential images

[0094] The estimation is carried out by machine learning or image processing.

[0095] In the embodiment that uses machine learning method, a dataset required for train such a machine learning model. The dataset is generated a method that begins with receiving images of the eye. The method requires at least two sequential images of the at least one eye, preferably two, because of it is essential that determining color-density changes for the estimation. The images preferably comprise an iris and a sclera of the eye, a vasculature around the eye and further comprises red reflex, if the model want to be trained for intracranial pressure.

[0096] The images and color / color-density difference and blood pulsation data and pre-collected pressure data of subject eye constitute training data together. This dataset comprises data gathered from multiple subjects, human subjects.

[0097] The dataset generated is used to train a learning algorithm that estimates intraocular pressure. Potentially, learning algorithms can be selected between machine learning algorithms such as supervised, unsupervised, semi supervised or reinforcement learning, deep learning, computer vision, algorithms and / or any combination of the different methods. Both data-based estimation or image-based estimation models can be used. For example, image-based models like CNN, object detection, semantic segmentation, anomaly detection or data-based models like decision trees, SVM, linear regression, k-means, neural network can be used for estimation.

[0098] The image of the patient eye and / or regions of interest of the eye and / or data of color and / or color-density changes of regions of interest and / or pulse characteristic derived are fed as input the pre-trained machine learning model which is trained with above mentioned dataset.

[0099] Referring to the Fig 4; The images received from imaging means (I) is transferred to the processing device (P) which is configured to run estimation method, which is image processing or pretrained machine, with the images as inputs to estimate intraocular pressure, preferably together with intracranial pressure.

[0100] For this, region of interests of the received images is determined by the processing device (P) and after that determining color and / or color-density changes between region of interests of the sequential images.

[0101] As can be seen in Fig. 3, red reflex pulsation is determined from the sequential images, if the intracranial pressure determination is carried out, too.

[0102] This user data comprises color and / or color-density changes between region of interests of the sequential images and the image itself inputs to the trained learning algorithm for estimating intraocular pressure, preferably together with intracranial pressure.

Claims

CLAIMS1. A computer implemented method for estimating intraocular pressure (IOP) and / or an intracranial pressure (ICP), characterized by comprising steps of,Receiving at least two sequential images of at least one eye of a patient and Detecting at least two of the regions of interest of the eye,Determining color and / or color-density changes of regions of interest of the eye in the sequential images.

2. A method according to Claim 1, wherein two of the regions of interest of the eye selected from the group; an iris and a sclera of the eye, a vasculature around the eye.

3. A method according to Claim 2, wherein the regions of interest of the eye further comprises a vasculature around the eye.

4. A method according to any of preceding claims, wherein the regions of interest of the eye further comprises red reflex of the eye.

5. A method according to any of preceding claims, characterized by receiving image of both eyes of the patient.

6. A method according any of preceding claims, wherein the sequential images are sequential frames of video of the eye.

7. A method according to Claim 6, wherein color and / or color-density changes of regions of interest of the eye is determined in predetermined time.

8. A method according to any of preceding claims, characterized by deriving blood pulsation from color and / or color-density changes.

9. A method according to Claim 8, characterized by deriving blood pulsation from color and / or color-density changes for each region of interest.

10. A method according to any of preceding claims wherein estimation is carried out by image processing.

11. A method according to any of preceding claims wherein estimation is carried out by pre-trained machine learning algorithm.

12. A data processing device comprising means for carrying out the steps of the method of claim 1 to 11.

13. A computer program comprising instructions which, when the program is executed by a data processing device, cause the data processing device to carry out the method of claim 1 to 11.

14. A computer-readable data carrier having stored thereon the computer program of claim 13.

15. A computer implemented method for generating a dataset for machine learning algorithm of Claim 11 which estimates intraocular pressure (IOP) and / or an intracranial pressure (ICP), characterized by comprising steps of,Receiving at least two sequential images of at least one eye of a patient and precollected intraocular pressure data of the eye, Detecting regions of interest of the eye, Determining color-density changes of regions of interest of the eye in the sequential images.

16. A dataset generated by according to Claim 15.

17. A computer implemented training method for estimating intraocular pressure, characterized by training machine learning algorithm of Claim 11 with a dataset according to Claim 16.

Images