Computer program

Abstract

The computer program comprises instructions that cause an electronic processor to perform the steps of acquiring at least one infrared image of a user's eye and selecting an area of the image corresponding to an inner surface of the user's eyelid. The program also provides for correcting the luminosity of said image and applying appropriate filters to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue.

Description

[0001] Description

[0002] COMPUTER PROGRAM

[0003] Technical field

[0004]

[0001] The present invention relates to a computer program and a related system, in particular for evaluating the state of the Meibomian glands.

[0005] Prior art

[0006]

[0002] The Meibomian glands are glands of tubulo-acinar conformation located in the tarsal part of the upper and lower eyelid. They are positioned vertically, next to each other, and their excretory ducts open along the palpebral margin.

[0007]

[0003] They are holocrine sebaceous glands without internal vascularization, which are supplied by the blood vessels that cross the palpebral tissues. The blood vessels that supply the palpebral tissues surrounding the Meibomian glands, therefore, provide the nutrients and oxygen necessary for the correct function of the glandular tissues, supporting the production of the sebaceous secretion of these glands, which constitutes the outer layer of the tear film, suitable for protecting the aqueous layer of the tear film from evaporation, maintaining the level of hydration of the corneal epithelium.

[0008]

[0004] Excessive evaporation of the tear film is the basis of dry eye syndrome, an ocular pathology that causes symptoms such as redness, burning, photophobia, a foreign body sensation in the eye and, in more severe cases, ocular pain and blurred vision.

[0009]

[0005] One of the main causes of dry eye syndrome is the dysfunction of the Meibomian glands which are located inside the eyelids and are responsible for the production of the lipid component of the tear film.

[0010]

[0006] In particular, Meibomian Gland Dysfunction (MGD) is a pathology affecting the Meibomian glands that leads to a malfunction of the glands caused by inflammatory states, obstruction of the excretory ducts, and modifications of the glandular structure.

[0011]

[0007] Therefore, the need to perform an evaluation of the functionality of the Meibomian glands is known, in order to be able to act promptly to restore the tissue through appropriate therapies, where necessary.

[0012]

[0008] A widespread diagnostic examination of the Meibomian glands is meibography, which has the purpose of evaluating the state of the glands and, in particular, estimating the percentage of gland loss within the palpebral tissue. This examination involves acquiring an image of the area where the glands are located by means of an infrared camera and appropriately processing the image to perform the aforementioned evaluation. In particular, the processing of the image exploits the different thermal responses of the glands and the surrounding tissue. It should be considered that the level of Meibomian gland dysfunction is generally classified according to the international scale of Dr. Heiko Pult. According to the Pult scale, the severity levels vary from level 0, in which all glands are present, i.e., the lack of glands is 0%, up to level 4 in which the lack of glands is between 75% and 100%.

[0013]

[0009] A known type of apparatus capable of performing meibography comprises a device for the detection and acquisition of images of the type of a camera or a video camera suitable for acquiring an image of the tarsal palpebral area subjected to the emission of infrared light. The apparatus also comprises a control unit suitable for processing the image and obtaining data related to the patient's health status. The processing of the image by the control unit provides for selecting the pixels corresponding to gray levels having an intensity greater than a preset threshold value to estimate the area occupied by the glands.

[0014]

[0010] A lamented problem is that there is a substantial overlap between the area occupied by the glands and the area related to the surrounding tissue, due to the nature of the surrounding tissue and the exposure to light which does not allow for the automatic and accurate recognition of the Meibomian glands. Therefore, the intervention of the operator is often required to make corrections to the obtained results, for example by means of a visual examination of the image.

[0015]

[0011] To improve the precision in the automatic identification of Meibomian glands, methods have been developed for the analysis of the presence of Meibomian glands based on "Deep Neural Networks", i.e., on deep neural networks, which serve to simulate the decisionmaking power of the human brain. The methods allow, for example, to automatically segment the Meibomian glands along the temporo-occipital direction to acquire data on the variation of the glandular tissue in that direction. However, these methods are still in the development phase and do not currently offer sufficient explanations of their functioning. Moreover, such models require the acquisition of a large number of data to train the network.

[0016]

[0012] Therefore, there is a felt need to identify solutions that guarantee greater accuracy in the automatic evaluation of the Meibomian glands within the palpebral tissue, also considering that the quantity of Meibomian glands is a fundamental indicator for the precise determination of the degree of Meibomian Gland Dysfunction (MGD).

[0017] Disclosure

[0018]

[0013] The object of the present invention is to solve the aforementioned problems by devising a computer program that allows for an optimal evaluation of the state of the Meibomian glands.

[0019]

[0014] Within the scope of this object, a further purpose of the present invention is to devise a program that allows for the accurate and automatic determination of the presence of Meibomian glands and, therefore, to estimate their eventual percentage of loss.

[0020]

[0015] A further purpose of the invention is to provide a computer program that allows for determining the position and dimensions of the Meibomian glands present.

[0021]

[0016] Another purpose of the invention is to provide a program that allows for estimating the severity of the Meibomian gland dysfunction.

[0017] A further purpose of the invention is to provide a system that implements the phases of the computer program of simple constructive and functional design, equipped with surely reliable operation, versatile use, as well as being relatively economical.

[0022]

[0018] The cited purposes are achieved, according to the present invention, by the computer program according to claim 1 and by the system according to claim 9.

[0023]

[0019] The computer program comprises instructions that cause an electronic processor to perform the steps of acquiring at least one infrared image of a user's eye and selecting an area of said image corresponding to an inner surface of a user's eyelid.

[0024]

[0020] The program then provides for correcting the luminosity of said image and applying appropriate filters to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue.

[0025]

[0021] The program also provides for calculating at least two lines and tracing said lines on said image, spaced apart from each other, each said line joining points arranged at opposite edges of the image, at a predetermined height of said image.

[0026]

[0022] Preferably, each said line extends along a first initial direction that is tangent to a curved trajectory.

[0027]

[0023] Preferably, each line is substantially curved.

[0028]

[0024] Subsequently, the program comprises the steps of visualizing, for each line, a signal representing the variation of the gray values along said line and detecting, for each line, the local maximum values, each local maximum value being indicative of the presence of a Meibomian gland.

[0029]

[0025] The program thus allows an evaluation of the state of the Meibomian glands to be performed by tracing at least two lines on the image, at different heights, and detecting the local maximum values along these lines, thereby estimating the number of glands. It is observed, in fact, that the number of Meibomian glands is an important indicator of gland dysfunction.

[0030]

[0026] Preferably, said lines are traced at respective increasing height values of the image.

[0031]

[0027] Preferably, the step of acquiring at least one infrared image of a user's eye can be performed by extracting the infrared image from a memory of the electronic processor or by receiving the infrared image from a device for the detection and acquisition of images or from a further device that has one or more infrared images in a relative internal memory.

[0032]

[0028] Advantageously, the program comprises the further step of creating a graph using the local maximum values identified for each line, wherein the nodes correspond to the local maximum values identified for said lines and each arc, which joins a pair of nodes, represents the distance between said local maximum values represented by said nodes.

[0033]

[0029] Advantageously, the program further comprises the step of extracting the most probable connecting lines that join the local maximum values belonging to said lines, based on the distance between the local maximum values detected along said lines, each said connecting line representing a Meibomian gland.

[0034]

[0030] Preferably, the program provides for extracting from said connecting lines information on the length and shape of said Meibomian glands. Furthermore, the most probable connecting lines also provide more accurate information on the position of the Meibomian glands.

[0035]

[0031] It is observed that the construction of a graph allows for a more accurate determination of the presence of the Meibomian glands as the most probable connecting lines of the local maximum values are extracted. The identification of the lines that connect the local maximum values also provides information on the length and shape of the Meibomian glands, as well as on their position.

[0036]

[0032] Preferably, in the case where more than two lines are traced on said image, said step of creating a graph using the local maximum values identified for each line is performed by connecting the nodes associated with one line only with the nodes of the subsequent line.

[0037]

[0033] Preferably, said step of extracting the most probable connecting lines that join the local maximum values is performed using Dijkstra's algorithm, which allows for identifying the closest local maximum values of said lines.

[0038]

[0034] Preferably, the step of extracting the most probable connecting lines also provides for extracting connecting lines that join a number of local maximum values lower than the number of lines traced on the image. In this way, it is possible to evaluate the loss of Meibomian glands and thus perform an optimal evaluation of the gland dysfunction.

[0039]

[0035] Preferably, the program also provides for the step of classifying the identified Meibomian glands into long, medium, and short glands, based on their length, and calculating an atrophy index correlated to the severity of the Meibomian gland dysfunction, said atrophy index being defined by the following equation:

[0040] A L- NM- NS) Ntot

[0041] Wherein:

[0042] A = atrophy index

[0043] Ntot = total glands

[0044] NL = long glands

[0045] NM = medium glands

[0046] Ns = short glands

[0047]

[0036] Preferably, the classification of the glands into the three mentioned classes is performed based on the number of graph layers involved; therefore, the length is discretized. The term layer is intended as a group of nodes belonging to the same line and, therefore, not connected to each other.

[0048]

[0037] Preferably, the step of calculating at least two lines and tracing said lines on said image provides, for each line, the steps of identifying, at a pair of opposite edges of the image, a point for each edge located at a preset height of the image and connecting the points of said opposite edges with a line identified by the following equation: y = m x x + q in which the coefficients m and q are determined, based on the coordinates of the points (xi, X2, yi, y?), by the following equations:

[0049]

[0038] Preferably, the program comprises the step of estimating, for each signal related to the gray values along a relative line, the amplitude of the peaks corresponding to the local maximum values, said information being useful for excluding local maximum values that are not associated with respective glands and for estimating the area occupied by each gland.

[0050]

[0039] The present invention also relates to a system comprising a control unit equipped with a user interface, said control unit comprising an electronic processor configured to perform the steps of acquiring at least one infrared image of a user's eye and selecting an area of the image corresponding to an inner surface of the user's eyelid.

[0051]

[0040] The subsequent step provides for correcting the luminosity of said image and applying appropriate filters to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue.

[0052]

[0041] Subsequently, the processor performs the steps of calculating at least two lines and tracing said lines on said image, spaced apart from each other, each said line joining points arranged at opposite edges of the image, at a predetermined height of said image, and visualizing, for each line, a signal representing the variation of the gray values along said line.

[0053]

[0042] Preferably, each said line extends along a first initial direction that is tangent to a curved trajectory.

[0054]

[0043] Preferably, each said line is substantially curved.

[0055]

[0044] The subsequent step is to detect, for each line, the local maximum values, each local maximum value being indicative of the presence of a Meibomian gland.

[0056]

[0045] Preferably, said control unit comprises a memory readable by said electronic processor comprising instructions that, when executed by said electronic processor, cause said electronic processor to perform the above-mentioned steps.

[0057]

[0046] Preferably, said system comprises at least one device for the detection and acquisition of images, for example of the type of a camera or a video camera, said control unit being configured to receive and process the images acquired from said device for the detection and acquisition of images.

[0058]

[0047] Preferably, the system is used to perform diagnostic procedures, for example in the ophthalmological field.

[0059]

[0048] Preferably, the system implements the computer program described above.

[0060] Description of the drawings

[0061]

[0049] The particulars of the invention will become more evident from the detailed description of a preferred embodiment of the computer program according to the invention, illustrated by way of non-limiting example in the accompanying drawings, in which: figure 1 shows an image of a user's eye; figure 2 shows an image of the eye illustrated in figure 1 following the application of filters; figure 3 shows an image of an area of a user's eye corresponding to an inner surface of the eyelid; figure 4 shows an image of an area of a user's eye, corresponding to an inner surface of the eyelid, in which a series of lines are traced to divide the image into corresponding regions of space; figure 5 shows a graph representing the signals of the gray level variation detected along three lines traced on an image of an inner surface of the eyelid in the case of a healthy subject; figure 6 shows an image of a portion of an inner surface of the eyelid in the case of a healthy subject; figure 7 shows a graph representing the signals of the gray level variation detected along three lines traced on the image of an inner surface of the eyelid in the case of a subject who has a pathology; figure 8 shows an image of a portion of an inner surface of the eyelid in the case of a subject who has a pathology; figure 9 shows a graph representing the signals of the gray level variation detected along three lines traced on the image of an inner surface of the eyelid in the space defined by the Cartesian axes; figure 10 shows an image of a user's eye in which the local maximum values of the gray levels along the lines traced in the area of an inner surface of the eyelid are represented by a series of points, wherein the local maximum values are visible in figure 9; figure 11 shows a graph in which the nodes are the local maximum values of the gray levels detected along the three lines; figure 12 shows an image of a user's eye in which the identified Meibomian glands are traced with connecting lines; figures 13 and 14 respectively show an image of a user's eye in which the identified Meibomian glands are traced with lines and the same image after the application of a filter; figure 15 shows a graph representing the signals of the gray level variation detected along three lines traced on the image of an inner surface of the eyelid.

[0062] Description of embodiments of the invention

[0063]

[0050] With particular reference to these figures, images and graphs are shown relating to the steps implemented by the computer program, in particular for evaluating the state of the Meibomian glands, according to the present invention.

[0064]

[0051] The computer program comprises instructions that cause an electronic processor to perform the steps described below.

[0065]

[0052] First, the program provides for acquiring at least one infrared image of a user's eye (see figure 1 ). The retrieved image is, therefore, a grayscale image.

[0066]

[0053] The image of an eye shows at least one everted eyelid so as to expose the inner surface of the eyelid that is to be examined.

[0067]

[0054] This step can be performed by extracting the infrared image from a memory of the electronic processor or by receiving the infrared image from a device for the detection and acquisition of images or from a further device that has one or more infrared images in a relative internal memory.

[0068]

[0055] Subsequently, the program provides for selecting an area of the image corresponding to an inner surface of the user's eyelid, as indicated, for example, in figures 1 and 2 by the dashed line.

[0069]

[0056] The area has a substantially crescent shape and covers, substantially in its entirety, the inner surface of an eyelid where the Meibomian glands are located.

[0070]

[0057] The subsequent step provides for correcting the luminosity of the image and applying appropriate filters to the image to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue. In this way, the contrast between the Meibomian glands and the surrounding tissue is increased (see figure 2).

[0071]

[0058] It is possible that a binary mask is applied to isolate the selected area, corresponding to the inner surface of a user's eyelid (see figure 3).

[0072]

[0059] The program then provides for calculating at least two lines and tracing the lines on the image, spaced apart from each other. Each line joins points of the image arranged at opposite ends, at a predetermined height of the image. The variation of gray levels along each line is used to obtain information on the Meibomian glands by exploiting the fact that the glands alternate with the surrounding tissue in a rather regular manner.

[0073]

[0060] In essence, the lines are traced at respective increasing percentages of the image height. By way of example, it is possible to define three lines at 25%, 50%, and 75% of the image height respectively, or four lines at 10%, 30%, 55%, and 75% of the image height respectively.

[0074]

[0061] Preferably, the number of traced lines is between two and four, but it is possible to calculate and trace a number of lines greater than four.

[0075]

[0062] The number of lines can obviously vary as it is a modifiable parameter to improve the identification capability of the Meibomian glands. In figure 4, for example, three lines are traced on the image. The step of calculating the lines provides for identifying, at a pair of opposite edges of the image, at least two points for each edge located at respective height values of the image. The rounded corners at the opposite edges of the image are excluded so as to have a length along a longitudinal direction y equal to the length in further regions of the image.

[0076]

[0063] Preferably, each line extends along a first initial direction that is tangent to a curved trajectory.

[0077]

[0064] Preferably, each line is substantially a curved line.

[0078]

[0065] For each line, one then proceeds to connect the points identified at the opposite edges, at a preset height of the image, with a line identified by the following equation: y = m x x + q

[0079]

[0066] The coefficients m and q are determined, based on the coordinates of the points (xi, X2, yi, y2), by the following equations:

[0080]

[0067] The above equations allow obtaining the coordinates (x, y) of the pixels underlying each line.

[0081]

[0068] The program provides for visualizing, for each line, a signal that represents the variation of the gray values along the line. Each pixel of the image intercepted by a line and previously identified in space by a pair of coordinates x, y, provides information on the luminous intensity, and this information is shown by the signal.

[0082]

[0069] In figure 9, the signals relating to the variation of the gray levels along three lines are shown, in which the lines are identified with the letters B, C, and D.

[0083]

[0070] Advantageously, it is possible to apply a filter, preferably a moving average filter, to the signals to reduce noise. Preferably, the moving average filter has a window length between 40 and 120 pixels, even more preferably a window having a length of 80 pixels.

[0084]

[0071] Subsequently, one proceeds to detect, for each line, the local maximum values wherein each local maximum value is indicative of the presence of a Meibomian gland. The step of detecting the local maximum values provides for setting the derivative to zero and comparing with the minima to the right and left.

[0085]

[0072] The detection of the local maximum values, corresponding to the peaks of the signals (see figure 9) allows estimating the number of glands at a given height.

[0073] The program also comprises the step of estimating, for each signal related to the gray values, the amplitude of the peaks and comparing it with a first preset reference value to evaluate whether there are local maximum values that are not associated with respective Meibomian glands. More in detail, if the amplitude of the peak, corresponding to a local maximum value, is higher than the first reference value, it means that the peak does not correspond to a gland and is, therefore, excluded from the count. Furthermore, the step of estimating the amplitude of the peaks provides useful information for estimating the area occupied by each gland where the line crosses the gland and, in particular, the amplitude of each gland.

[0086]

[0074] Furthermore, adjacent peaks that are closer than a second reference value are equally excluded from the count in order to avoid counting the same gland twice. The second reference value can be, for example, 10 pixels.

[0087]

[0075] By way of example, the first reference value mentioned for evaluating whether local maximum values not associated with respective glands exist can be between 80 and 100 pixels, preferably it can be equal to 90 pixels. This reference value is determined by calculating a moving average of the amplitude of the peaks of the signal related to the variation of the gray levels.

[0088]

[0076] It is, therefore, possible to apply to the image a filter based on the amplitude of the peaks, which functions as already mentioned, to obtain more accurate information on the presence of the glands. Figures 13 and 14 show the result of applying the filter on the amplitude of the peaks.

[0089]

[0077] It is observed that the application of a filter on the amplitude of the peaks allows for a reduction in the size of the graph, which is explained below. The graph is created and used more quickly.

[0090]

[0078] Subsequently, the program comprises, in fact, the step of creating a graph using the local maximum values identified for each line. The nodes of the graph correspond to the local maximum values identified for the lines and each arc, which joins a pair of nodes, represents the distance between the local maximum values.

[0091]

[0079] In the case where more than two lines are traced on the image, the step of creating a graph is performed by connecting the nodes associated with one line only with the nodes of the subsequent line.

[0092]

[0080] Figure 11 shows an example of a graph in which the lines of belonging of the local maximum values are indicated with the letters B, C, D respectively. These local maximum values are traced on an image of a surface of the eyelid in figure 10.

[0093]

[0081] From the graph, the most probable connecting lines are extracted that join the local maximum values belonging to the lines, based on the distance between the local maximum values detected along the lines. The graph allows, therefore, considering the local maximum values along the lines, to identify which is the most probable connecting line that joins the local maximum values of the lines along a direction substantially orthogonal to the lines themselves.

[0094]

[0082] Each connecting line represents a Meibomian gland, as shown in figure 12.

[0095]

[0083] From the connecting lines, it is possible to obtain information on the shape and length of the Meibomian glands. Furthermore, the connecting lines provide a rather accurate indication of the position of the Meibomian glands.

[0096]

[0084] The step of extracting the most probable connecting lines uses Dijkstra's algorithm, which allows identifying the closest local maximum values of the lines.

[0097]

[0085] The aforementioned step of extracting the most probable connecting lines also provides for extracting connecting lines that join a number of local maximum values lower than the number of lines traced on the image. For example, if three lines are traced on the image, the connecting lines are also extracted and, therefore, the corresponding Meibomian glands, which might have been identified by only two local maximum values. This type of information is useful as it allows for evaluating the loss of Meibomian glands and, in general, contributes to an evaluation of the gland dysfunction.

[0098]

[0086] In particular, the step of extracting the most probable connecting lines initially provides for using Dijkstra's algorithm for all the layers of the graph, wherein a layer of the graph is intended as a group of nodes belonging to the same line and, therefore, not connected to each other. The layers are arranged in order in the graph according to the corresponding lines traced on the image and two special nodes are added: a node, called "super source", which is connected to each node of the first layer by a series of arcs having a weight of 0, and a node, called "super sink", which is connected to each node of the last layer by a series of arcs having a weight of 0, wherein the weight indicates the distance between the nodes. These special nodes respectively represent the starting point and the arrival point used by Dijkstra's algorithm to identify the shortest path, i.e., the connection between nodes of different layers that are closest.

[0099]

[0087] Subsequently, to identify the glands defined by a number of local maximum values lower than the number of lines traced on the image, one proceeds to repeat the application of the algorithm by removing the first layer or the last layer and the nodes that have already been associated with the glands. The "super sink" or "super source" node that has remained without connection following the removal of a layer is connected again to the new available layer.

[0100]

[0088] The program also comprises the step of classifying the identified Meibomian glands into long NL, medium NM, and short Ns glands based on their length and calculating an atrophy index A, which is correlated to the severity of the Meibomian gland dysfunction. The classification of the glands into the three mentioned classes is performed based on the number of layers involved; therefore, the length is discretized.

[0101]

[0089] The atrophy index is calculated using the following equation:

[0102] A L- NM- NS)

[0103] Ntot

[0104] Wherein:

[0105] Ntot = total glands

[0106] NL = long glands

[0107] NM = medium glands

[0108] Ns = short glands

[0109]

[0090] The atrophy index is comprised between the values -1 and 1 , where the value -1 relates to a user who has all shortened Meibomian glands, while the maximum value 1 relates to a user who has all healthy glands covering the entire length of the eyelid.

[0110]

[0091] Experimentally, the described program has been implemented using an image of an eye of a healthy subject and an image of an eye of a pathological subject (see figures 5-8). In the case of a healthy subject, the gland, identified by the star symbol, is present on all lines, whereas for a subject with a dysfunction of the Meibomian glands, the gland is absent on one of the lines. The information relating to a gland present on only two lines indicates a loss of at least 25% of the gland.

[0111]

[0092] To evaluate the robustness of the program, a validation was performed by comparing the identification of the local maximum values, i.e., the peaks, by the program with the identification of the peaks by visual inspection by an operator. This operation was carried out by 10 different operators and provided an average deviation of +1-2 glands on a group of 10 subjects. The results of the program's application are illustrated in figure 15, which shows a graph representing the signals of the gray level variation detected along three lines (lines E, F, G) traced on the image of an inner surface of the eyelid, in which the peaks are indicated with triangle-shaped symbols.

[0112]

[0093] The steps of the program can be implemented by a system comprising a control unit comprising an electronic processor.

[0113]

[0094] The system, not represented in the figures, comprises at least one device for detecting and acquiring images, for example of the type of a camera or a video camera. The control unit also comprises a user interface, which allows an operator to interact with the system, and is configured to receive and process the images acquired from the device.

[0114]

[0095] The system can be used to perform diagnostic procedures, preferably in the ophthalmological field.

[0115]

[0096] According to an embodiment, the system also comprises a support member configured to support in a removable manner and in respective positions the control unit and the device for the detection and acquisition of images.

[0097] The support member allows the system to be attached to a base body or can be manually supported by an operator assigned to perform the diagnostic procedures.

[0116]

[0098] The computer program allows for an optimal evaluation of the state of the Meibomian glands as it provides for tracing a series of lines on the image, at different heights of the image, and detecting the local maximum values of the gray levels along these lines, thus estimating the number of glands. Extracting information on the number of glands is a significant evaluation of the state of the Meibomian glands.

[0117]

[0099] Furthermore, the method allows for accurately determining the presence of Meibomian glands by creating a graph that allows for extracting the most probable connecting lines of the local maximum values. The identification of the lines that connect the local maximum values also provides information on the length and shape of the glands, as well as on their position.

[0118]

[0100] A further aspect of the invention is that the program can provide an indication of the severity of the gland dysfunction by calculating the atrophy index, using the information derived from the graph on the length of the individual glands.

[0119]

[0101] Furthermore, it should be emphasized that the program provides information on the Meibomian glands automatically starting from an infrared image of a user's eye.

[0120]

[0102] In the practical implementation of the invention, the materials used, as well as the shape and dimensions, can be any according to the requirements.

[0121]

[0103] Where the technical features mentioned in any claim are followed by reference signs, those reference signs have been included for the sole purpose of increasing the intelligibility of the claims and accordingly, such reference signs have no limiting effect on the scope of each element identified by way of example by such reference signs.

Claims

Claims1 . A computer program comprising instructions that cause an electronic processor to perform the following steps: a) acquiring at least one infrared image of a user's eye; b) selecting an area of said image corresponding to an inner surface of a user's eyelid; c) correcting the luminosity of said image and applying appropriate filters to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue; d) calculating at least two lines and tracing said lines on said image, spaced apart from each other, each said line joining points arranged at opposite edges of the image, at a predetermined height of said image; e) visualizing, for each line, a signal representing the variation of the gray values along said line; f) detecting, for each line, the local maximum values, each local maximum value being indicative of the presence of a Meibomian gland; g) creating a graph using the local maximum values identified for each line, wherein the nodes correspond to the local maximum values identified for said lines and each arc, which joins a pair of nodes, represents the distance between said local maximum values represented by said nodes; h) extracting the most probable connecting lines that join the local maximum values belonging to said lines, based on the distance between the local maximum values detected along said lines, each said connecting line representing a Meibomian gland.

2. The program of claim 1 , wherein it provides for extracting from said connecting lines information on the length and shape of said Meibomian glands.

3. The program of claim 1 or 2, wherein, in the case where more than two lines are traced on said image, said step g. is performed by connecting the nodes associated with one line only with the nodes of the subsequent line.

4. The program of claim 1 , 2 or 3, wherein said step h. is performed using Dijkstra's algorithm, which allows for identifying the closest local maximum values of said lines.

5. The program of any one of claims 1-4, wherein the step d. of calculating at least two lines and tracing said lines on said image provides, for each line, the steps of: d1 ) identifying, at a pair of opposite edges of the image, a point for each edge located at a preset height of the image; d2) connecting the points of said opposite edges with a line identified by the following equation: y = m x x + qin which the coefficients m and q are determined, based on the coordinates of the points (xi, X2, yi, y?), by the following equations: yi- - y m = -X2~ X y2~ yi q = x2X — - — + y2x2X16. The program of any one of claims 1-5, wherein it comprises the step of estimating, for each signal related to the gray values along a relative line, the amplitude of the peaks corresponding to the local maximum values, said information being useful for excluding local maximum values that are not associated with respective glands and for estimating the area occupied by each gland.

7. The program of any one of claims 1-6, wherein it comprises the step of classifying the identified Meibomian glands into long (NL), medium (NM), and short (Ns) glands, based on their length, and calculating an atrophy index (A) correlated to the severity of the Meibomian gland dysfunction, said atrophy index (A) being defined by the following equation:WhereinNtot = total glandsNL = long glandsNM = medium glandsNs = short glands8. The program of any one of claims 1-7, wherein said lines are traced at respective increasing height values of the image.

9. A system comprising a control unit equipped with a user interface, said control unit comprising an electronic processor configured to perform the following steps: a) acquiring at least one infrared image of a user's eye; b) selecting an area of the image corresponding to an inner surface of a user's eyelid; c) correcting the luminosity of said image and applying appropriate filters to highlight the difference in luminous intensity between the Meibomian glands and the surrounding tissue; d) calculating at least two lines and tracing said lines on said image, spaced apart from each other, each said line joining points arranged at opposite edges of the image, at a predetermined height of said image; e) visualizing, for each line, a signal representing the variation of the gray values along said line;f) detecting, for each line, the local maximum values, each local maximum value being indicative of the presence of a Meibomian gland; g) creating a graph using the local maximum values identified for each line, wherein the nodes correspond to the local maximum values identified for said lines and each arc, which joins a pair of nodes, represents the distance between said local maximum values represented by said nodes; h) extracting the most probable connecting lines that join the local maximum values belonging to said lines, based on the distance between the local maximum values detected along said lines, each said connecting line representing a Meibomian gland.

10. The system of claim 9, wherein it comprises at least one device for the detection and acquisition of images, for example of the type of a camera or a video camera, said control unit being configured to receive and process the images acquired from said device for the detection and acquisition of images.

Images