Thyroid associated ophthalmopathy recognition method and system based on facial expression analysis, and terminal

Abstract

The thyroid-associated ophthalmopathy recognition method, system and terminal based on facial expression analysis of the application obtain the normal facial expression recognition result corresponding to the normal human face recognition or the sick facial expression recognition result corresponding to the thyroid-associated ophthalmopathy human face recognition through the thyroid-associated ophthalmopathy recognition model obtained by training, to perform facial expression recognition on the obtained human face image data to be recognized. The application starts from the scientific problems such as the expression face feature description of the thyroid-associated ophthalmopathy patients and the expression analysis based on the neural network, and finally realizes the purpose of the thyroid-associated ophthalmopathy screening, facilitates the user to evaluate the eye health condition of the user, saves a large amount of economic cost and time cost, and can effectively improve the work efficiency of the hospital for the thyroid-associated ophthalmopathy screening, reduces the work cost, and further promotes the popularization and application of the efficient, convenient and low-cost orbital disease screening method in China.

Description

Technical Field

[0001] This invention relates to the field of orbital disease recognition, and in particular to a method, system, and terminal for recognizing thyroid-related eye diseases based on facial expression analysis. Background Technology

[0002] As is well known, the orbit plays a protective and supportive role for vital tissues within the orbit, such as the eyeball, nerves, and blood vessels, and is crucial for maintaining the physiological position and function of the eyeball. Orbital diseases are a group of diseases that seriously affect visual function, have complex clinical manifestations, and are extremely difficult to treat. Orbital inflammation, tumors, congenital malformations, and trauma can all cause damage to the orbital structure, leading to enophthalmos, displacement, movement disorders, diplopia, decreased or lost vision, and even threatening life. Orbital diseases, with thyroid-associated ophthalmopathy (TAO) as a major representative, often cause varying degrees of damage to the eyelids, eyeball, and soft and hard tissues within the orbit, severely impairing the patient's visual function and craniofacial appearance. Patients with many of these diseases have lower quality of life scores than those with diabetes, emphysema, and heart failure, placing a heavy burden on families and society.

[0003] Orbital oculomotor atrophy (TAO) can cause birth defects, blindness, and death. Early diagnosis is crucial for the prognosis of TAO. However, the lack of effective screening methods causes many patients to miss the optimal treatment window, resulting in low TAO diagnosis and cure rates, poor prognosis, and a significant psychological and social burden on patients. Given the characteristics of TAO in terms of anatomy, pathophysiology, diagnostic methods, and current development, screening methods must be able to capture all aspects of the disease while being suitable for widespread implementation at the grassroots level with minimal equipment investment and convenient operation.

[0004] Currently, clinical research on transorbital optic neuropathy (TAO) both domestically and internationally focuses on functional MRI imaging, diagnosis of optic neuropathy, and computer-guided orbital surgery. These studies, through the computation and analysis of multimodal data such as orbital MRI, orbital CT, and medical history, have significantly improved the accuracy of diagnosis and the effectiveness of treatment. However, a common thread among these studies is that they target patients with a high suspicion of or even confirmed TAO, but they do not address the more severe disease burden caused by missed diagnoses, misdiagnoses, and delayed treatment due to a lack of screening methods. Correspondingly, the implementation of screening will directly promote early diagnosis and treatment of TAO, improving the currently high rates of blindness, malformation, and mortality associated with TAO.

[0005] However, TAO screening is rarely reported internationally. The main difficulties in initiating TAO screening are twofold. From the perspective of social resource distribution: because TAO diagnosis involves the application of multidisciplinary knowledge, the learning, diagnosis, and treatment decisions are highly complex. Therefore, effective diagnosis and treatment of this disease can only be effectively carried out in a few national or regional central medical institutions, hindering the implementation of effective hierarchical medical care and impeding the rational allocation of medical and social resources. From a technical perspective: changes in facial appearance are a unique sign of TAO and a potentially important focus for screening this disease. However, for non-ophthalmologists, and even those not specializing in orbital diseases, the level of recognition of eyelid-related signs and acquisition of information related to eye position and movement is very limited. Even for specialized orbital surgeons, relatively accurate differentiation of orbital diseases is difficult without specialized instruments and tools. Furthermore, existing more precise methods (such as MRI) are too costly and technically challenging to be suitable for widespread TAO screening at the grassroots level. Therefore, there is an urgent need for a low-cost and highly accessible TAO screening method to be applied at the grassroots level, serving as a front-end tool in the TAO hierarchical diagnosis and treatment system, and providing support for the early identification of TAO patients and timely referral to higher-level hospitals. Summary of the Invention

[0006] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a method, system and terminal for identifying thyroid-related eye diseases based on facial expression analysis, so as to solve the above-mentioned technical problems in the prior art.

[0007] To achieve the above and other related objectives, the present invention provides a method for identifying thyroid-associated ophthalmopathy based on facial expression analysis. The method includes: acquiring facial image data of a person to be identified; performing facial expression recognition on the facial image data of the person to be identified based on a trained thyroid-associated ophthalmopathy identification model to obtain a normal facial expression recognition result corresponding to normal facial recognition or a diseased facial expression recognition result corresponding to thyroid-associated ophthalmopathy patient facial recognition; wherein, each diseased facial expression recognition result corresponds to a thyroid-associated ophthalmopathy sign.

[0008] In one embodiment of the present invention, the training method of the thyroid-associated ophthalmopathy recognition model includes: training the thyroid-associated ophthalmopathy recognition model using a facial expression training set; wherein, the facial expression training set includes: a normal facial expression recognition training set corresponding to normal facial expression recognition results and a diseased facial expression recognition training set corresponding to various diseased facial expression recognition results.

[0009] In one embodiment of the present invention, the training set for recognizing facial expressions of patients includes: multiple facial image samples of patients with angry expressions corresponding to the recognition results of angry expressions of patients; multiple facial image samples of patients with fear corresponding to the recognition results of fear expressions of patients; multiple facial image samples of patients with sadness corresponding to the recognition results of sad expressions of patients; multiple facial image samples of patients with blank expressions corresponding to the recognition results of blank expressions of patients; and multiple facial image samples of patients with disgust corresponding to the recognition results of disgust expressions of patients.

[0010] In one embodiment of the present invention, the signs of thyroid-related ophthalmopathy include: a first sign of thyroid-related ophthalmopathy corresponding to the recognition result of an angry expression of the patient, a second sign of thyroid-related ophthalmopathy corresponding to the recognition result of a fearful expression of the patient, a third sign of thyroid-related ophthalmopathy corresponding to the recognition result of a sad expression of the patient, a fourth sign of thyroid-related ophthalmopathy corresponding to the recognition result of a blank expression of the patient, and a fifth sign of thyroid-related ophthalmopathy corresponding to the recognition result of an disgusted expression of the patient; wherein, the first sign of thyroid-related ophthalmopathy includes: eyelid swelling, exophthalmos, eyelid retraction, and conjunctival hyperemia; the second sign of thyroid-related ophthalmopathy includes: eyelid swelling, eyelid retraction, and exophthalmos; the third sign of thyroid-related ophthalmopathy includes: conjunctival edema and eyelid swelling; the fourth sign of thyroid-related ophthalmopathy includes: abnormal eye position, eyelid retraction, and exophthalmos; and the fifth sign of thyroid-related ophthalmopathy includes: eyelid swelling and abnormal eye position.

[0011] In one embodiment of the present invention, the disease-related anger feature includes disease-related anger eye features; the disease-related fear feature includes disease-related fear eye features; the disease-related sadness feature includes disease-related sadness eye features; the disease-related sadness feature includes disease-related sadness eye features; the disease-related listlessness feature includes disease-related listlessness eye features; the disease-related sadness feature includes disease-related sadness eye features; and the disease-related aversion feature includes disease-related fear-related aversion features.

[0012] In one embodiment of the present invention, the diseased anger feature further includes the diseased anger forehead and eyebrow features and / or the diseased anger lower face features; the diseased fear feature further includes the diseased fear forehead and eyebrow features and / or the diseased fear lower face features; the diseased sadness feature further includes the diseased sadness forehead and eyebrow features and / or the diseased sadness lower face features; the diseased listlessness feature further includes the diseased listlessness forehead and eyebrow features and / or the diseased listlessness lower face features; the diseased aversion feature further includes the diseased aversion forehead and eyebrow features and / or the diseased aversion lower face features.

[0013] In one embodiment of the present invention, the normal facial expression recognition training set includes: multiple expressionless facial image samples with expressionless features corresponding to the expressionless recognition result; multiple normal angry facial image samples with normal angry features corresponding to the normal angry expression recognition result; multiple normal fear facial image samples with normal fear features corresponding to the normal fear expression recognition result; multiple normal sad facial image samples with normal sad features corresponding to the normal sad expression recognition result; multiple normal blank facial image samples with normal blank features corresponding to the normal dull expression recognition result; and multiple normal disgust facial image samples with normal disgust features corresponding to the normal disgust expression recognition result.

[0014] In one embodiment of the present invention, the thyroid-associated ophthalmopathy identification model adopts a VGG-19 network structure.

[0015] To achieve the above and other related objectives, the present invention provides a thyroid-associated ophthalmopathy (TAO) identification system based on facial expression analysis. The system includes: a data acquisition module for acquiring facial image data of a person to be identified; and an identification module connected to the data acquisition module for performing facial expression recognition on the facial image data of the person to be identified based on a trained TAO identification model, so as to obtain a normal facial expression recognition result corresponding to normal facial recognition or a diseased facial expression recognition result corresponding to TAO patient facial recognition; wherein, each diseased facial expression recognition result corresponds to a TAO sign.

[0016] To achieve the above and other related objectives, the present invention provides a thyroid-associated ophthalmopathy recognition terminal based on facial expression analysis, comprising: one or more memory units and one or more processor units; the one or more memory units are used to store computer programs; the one or more processor units are connected to the memory units and are used to run the computer programs to execute the thyroid-associated ophthalmopathy recognition method based on facial expression analysis.

[0017] As described above, this invention is a method, system, and terminal for identifying thyroid-associated ophthalmopathy (TAO) based on facial expression analysis, which has the following beneficial effects: This invention uses a trained TAO identification model to perform facial expression recognition on acquired facial image data to obtain normal facial expression recognition results for normal faces or diseased facial expression recognition results for faces of people with TAO. This invention addresses scientific issues such as the description of facial expression features in patients with TAO and expression analysis based on neural networks, ultimately achieving the goal of TAO screening. This facilitates users' assessment of their own eye health, saving significant economic and time costs. Simultaneously, it effectively improves the efficiency of hospital TAO screening, reduces operating costs, and promotes the widespread adoption and application of efficient, convenient, and low-cost orbital disease screening methods in my country. Attached Figure Description

[0018] Figure 1 The diagram shown is a flowchart illustrating a method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to an embodiment of the present invention.

[0019] Figure 2 The diagram shows a schematic of the VGG-19 network structure used in the thyroid-associated ophthalmopathy recognition model in one embodiment of the present invention.

[0020] Figure 3 The diagram shown is a flowchart illustrating a method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to an embodiment of the present invention.

[0021] Figure 4 The diagram shown is a structural schematic of a thyroid-associated ophthalmopathy recognition system based on facial expression analysis according to an embodiment of the present invention.

[0022] Figure 5 The diagram shown is a structural schematic of a thyroid-associated eye disease recognition terminal based on facial expression analysis according to an embodiment of the present invention. Detailed Implementation

[0023] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0024] It should be noted that in the following description, reference is made to the accompanying drawings, which illustrate several embodiments of the invention. It should be understood that other embodiments may also be used, and changes in mechanical composition, structure, electrical system, and operation may be made without departing from the spirit and scope of the invention. The following detailed description should not be considered limiting, and the scope of the embodiments of the invention is defined only by the claims of the published patents. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. Spatially related terms, such as “upper,” “lower,” “left,” “right,” “below,” “below,” “lower part,” “above,” “upper part,” etc., may be used herein to illustrate the relationship between one element or feature shown in the figures and another element or feature.

[0025] Throughout this specification, when it is said that a part is "connected" to another part, this includes not only "direct connection" but also "indirect connection" by placing other elements in between. Furthermore, when it is said that a part "includes" a certain constituent element, unless otherwise stated otherwise, this does not exclude other constituent elements, but rather means that other constituent elements may also be included.

[0026] The terms "first," "second," and "third," etc., used herein are for the purpose of describing various parts, components, regions, layers, and / or segments, but are not limiting. These terms are used only to distinguish one part, component, region, layer, or segment from others. Therefore, the "first part," "component," "region," "layer," or "segment" described below may refer to a "second part," "component," "region," "layer," or "segment" without departing from the scope of this invention.

[0027] Furthermore, as used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context indicates otherwise. It should be further understood that the terms “comprising,” “including,” indicate the presence of the stated feature, operation, element, component, item, kind, and / or group, but do not preclude the presence, occurrence, or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms “or” and “and / or” as used herein are interpreted as inclusive, or mean any one or any combination thereof. Thus, “A, B, or C” or “A, B, and / or C” means “any one of: A; B; C; A and B; A and C; B and C; A, B, and C.” Exceptions to this definition arise only when combinations of elements, functions, or operations are inherently mutually exclusive in some manner.

[0028] Because the external characteristic lesions of thyroid-associated ophthalmopathy are concentrated in changes in the orbitofacial morphology, such as protruding eyeballs, sunken eyeballs, changes in orbitofacial structure, eyelid retraction, ptosis, eyelid swelling, and limited eye movement, patients with thyroid-associated ophthalmopathy often exhibit characteristic abnormal expressions such as anger, fear, sadness, blankness, and disgust in a normal, calm state. These expressions have corresponding features in the forehead and eyebrows, eyes, and lower half of the face, respectively.

[0029] Therefore, this invention provides a method, system, and terminal for identifying thyroid-associated ophthalmopathy (TAO) based on facial expression analysis. By training a TAO identification model, facial expression recognition is performed on acquired facial image data to obtain either a normal facial expression recognition result (corresponding to normal face recognition) or a diseased facial expression recognition result (corresponding to TAO patient face recognition). This invention addresses scientific issues such as the description of facial expression features in TAO patients and neural network-based expression analysis, ultimately achieving the goal of TAO screening. This facilitates users' assessment of their eye health, saving significant economic and time costs. Simultaneously, it effectively improves the efficiency of hospital TAO screening, reduces operating costs, and promotes the widespread adoption and application of efficient, convenient, and low-cost orbital disease screening methods in my country.

[0030] The present invention will now be described in detail with reference to the accompanying drawings, so that those skilled in the art can readily implement it. The present invention can be embodied in many different forms and is not limited to the embodiments described herein.

[0031] like Figure 1 This is a flowchart illustrating a method for identifying thyroid-related eye disease based on facial expression analysis, as described in an embodiment of the present invention.

[0032] The method includes:

[0033] Step S11: Obtain the face image data to be identified.

[0034] Optionally, mobile terminals can be used as interfaces to acquire facial image data, such as using a mobile phone camera to capture facial photos of suspected patients. Alternatively, a dedicated facial information collection and built-in screening device can be installed in specific medical institutions to screen patients who come for consultation and are suspected of having TAO.

[0035] Optionally, the face image can be preprocessed to a set spatial resolution; it can also be subsequently processed with data augmentation, such as symmetric reversal or random rotation, to achieve data expansion.

[0036] Step S12: Based on the thyroid-associated eye disease recognition model obtained through training, perform facial expression recognition on the face image data to be recognized, so as to obtain the normal facial expression recognition result corresponding to normal face recognition or the diseased facial expression recognition result corresponding to thyroid-associated eye disease face recognition.

[0037] In detail, the acquired facial image data to be identified is input into the thyroid-associated ophthalmopathy (TAO) recognition model to obtain the corresponding facial expression recognition results. Based on these results, normal facial expression recognition results for normal faces or diseased facial expression recognition results for TAO patients can be obtained. Since each diseased facial expression recognition result corresponds to a TAO sign, the TAO recognition model can not only determine whether someone has TAO through facial expression recognition, but also determine which type of TAO sign it is. By identifying the TAO signs, a recommendation for further medical treatment can be made, thereby achieving the goal of early screening of potential cases and facilitating subsequent treatment.

[0038] Facial expression recognition is used to identify thyroid-associated eye disease. Since facial expression is a specialized research field with many well-established definitions and research tools available, starting with facial expression not only enhances the homogeneity of the evaluation but also broadens the applicability of the tool. Furthermore, it provides a recognition method for physical signs with high evaluation thresholds, thus realizing a more systematic, objective, or automated evaluation method for physical signs.

[0039] Optionally, the training method for the thyroid-associated ophthalmopathy recognition model includes:

[0040] The thyroid-associated ophthalmopathy recognition model is obtained by training a facial expression training set. Specifically, the facial expression training set consists of multiple face image samples, including: a normal facial expression recognition training set corresponding to the normal facial expression recognition results and a diseased facial expression recognition training set corresponding to the diseased facial expression recognition results and a diseased facial expression recognition training set corresponding to the diseased facial expression recognition results.

[0041] Optionally, a large number of facial image samples can be selected and classified according to the facial expression features of each sample, into normal expression facial image samples and disease expression facial image samples. A portion of these samples can be used as the facial expression training set, and the remaining portion as the test set.

[0042] Optionally, the obtained training and test sets are preprocessed to train the model using the preprocessed data; preferably, since the acquired data are three-dimensional images, the resolution of the images in the training and validation sets is first unified; then data augmentation is performed, such as symmetric inversion, random rotation and other data augmentation operations, to achieve the purpose of data expansion.

[0043] Optionally, TAO patients exhibit facial signs resembling facial expressions due to pathological changes, but the current TAO diagnosis and treatment system lacks a description of these facial expressions. Based on the Universal Expressions theory and Facial Motor Coding System (FACS) theory of renowned psychologists Paul Ekman and W.V. Friesen, combined with the facial characteristics of TAO patients and the opinions of several clinical experts deeply involved in TAO diagnosis and treatment, five common expressions of TAO patients have been identified: anger, fear, sadness, blankness, and disgust, each with corresponding expressive characteristics.

[0044] Therefore, the training set for facial expression recognition of patients includes:

[0045] Multiple facial image samples with the characteristics of anger associated with illness, corresponding to the results of anger expression recognition for illness, are used by the thyroid-associated ophthalmopathy recognition model to recognize the characteristics of anger associated with illness and output the corresponding anger expression recognition results.

[0046] Multiple facial image samples with disease fear characteristics corresponding to the disease fear expression recognition results are used to enable the thyroid-associated eye disease recognition model to recognize the disease fear characteristics and output the disease fear expression recognition results corresponding to the disease fear expression.

[0047] Multiple facial image samples with the characteristics of illness and sadness corresponding to the recognition results of illness and sadness expressions are used to enable the thyroid-associated eye disease recognition model to recognize the features of illness and sadness expressions and output the corresponding illness and sadness expression recognition results.

[0048] Multiple facial image samples with symptoms of mental dullness, corresponding to the recognition results of mental dullness expression, are used by the thyroid-associated eye disease recognition model to recognize the features of mental dullness expression and output the corresponding mental dullness expression recognition results.

[0049] Multiple facial image samples with disease aversion features corresponding to the disease aversion expression recognition results are used to enable the thyroid-associated ophthalmopathy recognition model to recognize the disease aversion expression features and output the disease aversion expression recognition results corresponding to the disease aversion expression.

[0050] The thyroid-associated ophthalmopathy (TAO) recognition model, trained using the facial expression recognition training set, can identify facial expressions such as anger, fear, sadness, blankness, and disgust based on the features of the facial image. It can then obtain the corresponding facial expression recognition results. Recognizing any one of these expressions is sufficient to identify thyroid-associated ophthalmopathy. Since each of the three expressions corresponds to a specific thyroid-associated ophthalmopathy sign, the specific type of thyroid-associated ophthalmopathy sign can be determined based on the type of facial expression recognition result.

[0051] Optionally, the signs of thyroid-associated ophthalmopathy include: a first sign of thyroid-associated ophthalmopathy corresponding to the recognition result of the angry expression of the patient, a second sign of thyroid-associated ophthalmopathy corresponding to the recognition result of the fear expression of the patient, a third sign of thyroid-associated ophthalmopathy corresponding to the recognition result of the sad expression of the patient, a fourth sign of thyroid-associated ophthalmopathy corresponding to the recognition result of the dull expression of the patient, and a fifth sign of thyroid-associated ophthalmopathy corresponding to the recognition result of the disgust expression of the patient.

[0052] The first thyroid-associated ophthalmopathy sign includes: eyelid swelling, exophthalmos, eyelid retraction, and conjunctival hyperemia; the second thyroid-associated ophthalmopathy sign includes: eyelid swelling, eyelid retraction, and exophthalmos; the third thyroid-associated ophthalmopathy sign includes: conjunctival edema and eyelid swelling; the fourth thyroid-associated ophthalmopathy sign includes: abnormal eye position, eyelid retraction, and exophthalmos; and the fifth thyroid-associated ophthalmopathy sign includes: eyelid swelling and abnormal eye position.

[0053] Optionally, since the persistent facial features exhibited by TAO patients are due to characteristic orbital pathological changes, such as widening and bulging eyes caused by retrobulbar inflammation leading to protrusion of the eyeball, and redness and swelling of the eyes accompanied by a feeling of moisture caused by conjunctival congestion and edema; while abnormalities such as eyelid swelling and ptosis are attributed to varying degrees of eyelid swelling, different types of eye features can be defined to determine which type of facial feature it belongs to.

[0054] Therefore, the characteristics of illness-related anger include the eye features of illness-related anger; the characteristics of illness-related fear include the eye features of illness-related fear; the characteristics of illness-related sadness include the eye features of illness-related sadness; the characteristics of illness-related sadness include the eye features of illness-related sadness; the characteristics of illness-related listlessness include the eye features of illness-related listlessness; the characteristics of illness-related sadness include the eye features of illness-related sadness; and the characteristics of illness-related aversion include the characteristics of illness-related fear-related aversion.

[0055] In other words, if the eye features identified in the facial image data belong to the characteristics of angry eyes due to illness, then it means that the features of angry eyes due to illness are met; if the eye features identified belong to the characteristics of fearful eyes due to illness, then it means that the features of fearful eyes due to illness are met. The determination methods for the other three features are the same, so they will not be elaborated upon.

[0056] Preferably, the eye characteristics of the diseased angry correspond to the following manifestations: (1) wide eyes, possibly bulging; (2) relatively swollen upper eyelids; the eye characteristics of the diseased fear correspond to the following manifestations: (1) pupils turning downwards; (2) wide eyes, with the whites of the eyes possibly showing below / above the pupils; (3) swollen upper eyelids; the eye characteristics of the diseased sad correspond to the following manifestations: (1) moist eyes, with fluid accumulation; (2) dull eyes, with downward gaze; (3) drooping upper eyelids, giving a feeling of being wrapped around the eyeballs, with drooping outer corners of the eyes; the eye characteristics of the diseased dull eye correspond to the following manifestations: (1) wide eyes, with the whites of the eyes possibly showing below / above the pupils; (2) wide eyes, which may be accompanied by abnormal pupil orientation; (3) thickened upper eyelids; the eye characteristics of the diseased disgust correspond to the following manifestations: (1) eyelid retraction, with the whites of the eyes showing below the pupils; (2) horizontal lines appearing on the lower part of the lower eyelids, with the cheeks pushing them upwards.

[0057] Optionally, to better determine the facial expression characteristics of a patient, it is necessary to combine the features of the forehead, eyebrows, and lower half of the face in addition to relying solely on eye features. Therefore, the anger characteristic of a patient also includes the forehead and eyebrow features and / or the lower half of the face features associated with anger; the fear characteristic of a patient also includes the forehead and eyebrow features and / or the lower half of the face features associated with fear; the sadness characteristic of a patient also includes the forehead and eyebrow features and / or the lower half of the face features associated with sadness; the listlessness characteristic of a patient also includes the forehead and eyebrow features and / or the lower half of the face features associated with listlessness; and the disgust characteristic of a patient also includes the forehead and eyebrow features and / or the lower half of the face features associated with disgust.

[0058] In other words, if the eye features, forehead and eyebrow features and / or lower face features identified in the facial image data belong to the characteristics of angry eyes in patients with illness, and the forehead and eyebrow features and / or lower face features in patients with angry foreheads and eyebrows and / or lower face features in patients with angry foreheads and eyebrows, then it is considered to meet the characteristics of angry patients with illness. The determination methods for the other four features are the same, so they will not be elaborated on.

[0059] Preferably, the facial features of the angry forehead correspond to the following: (1) the eyebrows are furrowed together and the tails of the eyebrows are raised; (2) the side contours of the tails of the eyebrows are deepened; (3) vertical wrinkles appear between the eyebrows on both sides; the facial features of the angry lower part correspond to the following: the lips are tightly closed and the corners of the lips are straightened or turned downwards.

[0060] The forehead and eyebrow features associated with the fear of illness correspond to the following: the center of the eyebrows is raised, and the tail of the eyebrows is lowered; the lower face features associated with the fear of illness correspond to the following: the lips are slightly tense and pulled back; or they are elongated and pulled back at the same time.

[0061] The forehead and eyebrow features of the diseased sad face correspond to the following manifestations: the inner corners of the eyebrows are furrowed together, raised, and move the skin under the eyebrows; the lower face features of the diseased sad face correspond to the following manifestations: the lips are relaxed, can open, protrude slightly forward, and the corners of the mouth do not deviate significantly.

[0062] The characteristics of the affected dull forehead and eyebrows correspond to the following: (1) the eyebrows are horizontal and the distance between the eyebrows is large; (2) the forehead wrinkles are not obvious; the characteristics of the affected dull lower face correspond to the following: the lips are relaxed, can open the mouth, protrude slightly forward, and the corners of the mouth are not obviously deviated.

[0063] The symptoms of aversion to forehead and eyebrows correspond to the following: (1) There may be slight lowering of the brow center and upward raising of the eyebrow tail; (2) There are generally no vertical lines between the eyebrows on both sides; The symptoms of aversion to the lower part of the face correspond to the following: (1) Deepening of the nasolabial folds; (2) Lip raised; (3) Lip and upper lip tightly closed, pushing the upper lip upward, pulling down the corners of the mouth, and slightly protruding lips; (4) Wrinkled nose.

[0064] Optionally, for normal facial expression recognition, different types of facial expressions can also be recognized. Therefore, the normal facial expression recognition training set includes:

[0065] Multiple expressionless face image samples with expressionless features corresponding to the expressionless recognition results are used by the thyroid-associated eye disease recognition model to recognize the expressionless features and output the corresponding expressionless recognition results.

[0066] Multiple normal angry face image samples with normal angry features corresponding to the normal angry expression recognition results are used by the thyroid-associated ophthalmopathy recognition model to recognize the normal angry features and output the normal angry expression recognition results corresponding to the normal angry expressions.

[0067] Multiple normal fear face image samples with normal fear features corresponding to the normal fear expression recognition results are used by the thyroid-associated eye disease recognition model to recognize normal fear features and output the normal fear expression recognition results corresponding to normal fear expressions.

[0068] Multiple normal sadness face image samples with normal sadness features corresponding to the normal sadness expression recognition results are used to enable the thyroid-associated ophthalmopathy recognition model to recognize normal sadness features and output the normal sadness expression recognition results corresponding to normal sadness expressions.

[0069] Multiple normal blank face image samples with normal blank expression characteristics are used to provide the thyroid-related eye disease recognition model to recognize the normal blank expression characteristics and output the normal blank expression recognition results.

[0070] Multiple normal disgust face image samples with normal disgust features corresponding to the normal disgust expression recognition results are used by the thyroid-associated eye disease recognition model to recognize the normal disgust features and output the normal disgust expression recognition results corresponding to the normal disgust expressions.

[0071] The thyroid-associated eye disease recognition model, trained using the normal facial expression recognition training set, can identify facial features such as expressionless, normal angry, normal fearful, normal sad, normal blank, and normal disgusted expressions. It can then obtain the corresponding normal facial expression recognition results. By recognizing any one of these expression results, it can be determined that the face is normal and free from thyroid-associated eye disease.

[0072] It should be noted that the expressions of normal anger, fear, sadness, blankness, and disgust mentioned here are not the same as those of a patient. Therefore, it is possible to avoid situations where a normal person displays a normal angry expression and it is mistaken for a patient's angry expression.

[0073] Preferably, to ensure more accurate recognition, the normal facial expression recognition training set may further include: multiple normal shy face image samples with normal shy features corresponding to the normal shyness recognition result; multiple normal happy face image samples with normal happy features corresponding to the normal happyness recognition result; multiple normal excited face image samples with normal excited features corresponding to the normal excitement expression recognition result; and multiple normal surprised face image samples with normal surprised features corresponding to the normal surprise expression recognition result.

[0074] Optionally, the thyroid-related eye disease recognition model employs a VGG-19 network structure. Specifically, a facial expression training set is fed into the VGG-19 network for training. VGG-19 is a classic convolutional neural network (CNN) model, which can be divided into 19 main network layers (or, in other words, the total number of convolutional and fully connected layers is 19). The subcategories of these layers mainly include convolutional layers, activation layers (using the ReLU function), max pooling layers, and fully connected layers. Figure 2As shown, conv refers to a convolutional layer, relu refers to an activation layer, pool refers to a (max) pooling layer, and fc refers to a fully connected layer. The main process is as follows: first, features are extracted by convolutional and pooling layers, and then classification is performed by fully connected layers.

[0075] Preferably, all convolutional layers use 3x3 convolutions, and pooling layers mainly use max pooling and average pooling, with 2x2 max pooling used. Activation layers use the Rectified Linear Activation Function (ReLU); the first two fully connected layers have 4096 nodes, and the last fully connected layer has 1000 nodes. Finally, softmax classification is performed. To increase the model's robustness, dropout regularization is added to the fully connected layers.

[0076] Optionally, the thyroid-associated eye disease recognition model adopts an improved VGG-19 network structure, that is, the improved part is a fully connected layer. The original multiple fully connected layers are removed and modified to: after dropout, softmax classification is performed in a fully connected layer (if 10 categories are required, the fully connected layer is activated by softmax and outputs a 10x1 vector).

[0077] To better illustrate the above-mentioned method for identifying thyroid-related ophthalmopathy based on facial expression analysis, the present invention provides the following specific embodiments.

[0078] Example 1: A method for identifying thyroid-associated ophthalmopathy based on facial expression analysis. For example... Figure 3 The diagram shows a schematic of thyroid-associated ophthalmopathy identification based on facial expression analysis; the registration methods include:

[0079] 1. Describe the patient's facial expressions;

[0080] Based on the Universal Expressions theory and the Facial Movement Coding System (FACS) theory of renowned psychologists Paul Ekman and WV Friesen, and combined with the facial characteristics of TAO patients and the opinions of many clinical experts who have been deeply involved in the diagnosis and treatment of TAO, five common expressions of TAO patients have been extracted: anger, fear, sadness, blankness, and disgust. The corresponding expression characteristics are described in detail, as shown in Table 1.

[0081] Table 1: Five facial expression characteristics;

[0082]

[0083] The persistent facial expressions exhibited by patients with TAO (Transocular Ophthalmopathy) are due to characteristic pathological changes in the orbit. For example, widened and bulging eyes are caused by retrobulbar inflammation leading to protrusion of the eyeball; redness and swelling of the eyes with a moist appearance are caused by conjunctival congestion and edema; and abnormalities such as eyelid swelling, ptosis, and abnormal eyebrow alignment are attributed to varying degrees of eyelid swelling. Clinical practice shows a specific correlation between persistent abnormal facial expressions and the pathological characteristics of TAO, thus providing a solid scientific basis for screening TAO using facial expressions.

[0084] 2. Image acquisition and facial expression classification;

[0085] The training and test sets were composed of three-dimensional facial images of nearly 600 TAO patients and images from a normal facial database (Asian). The facial images of TAO patients were then classified one by one according to the aforementioned facial expression features.

[0086] 3. Image preprocessing;

[0087] 1) Unify the resolution of images in the training and validation sets.

[0088] 2) Perform data augmentation on the above data; that is, perform data augmentation operations such as symmetric reversal and random rotation to achieve the purpose of data expansion.

[0089] 4. Neural network training;

[0090] Images from the training set are fed into the VGG-19 network for training. The trained VGG-19 network then serves as the network model for facial expression classification.

[0091] 5. Output the discrimination result;

[0092] Using a trained VGG-19 as a classification CNN model, feature discrimination and expression classification are performed on facial images of TAO patients and normal people. If the discrimination result outputs any of the expressions of a TAO patient, then the person is judged to have TAO.

[0093] This solution serves as an auxiliary tool for ophthalmologists to identify tumor-associated eye diseases (TAOs). It requires only a single 3D facial image of a potential patient to generate TAO screening results, facilitating user assessment of their eye health and saving significant economic and time costs. Simultaneously, it effectively improves the efficiency of TAO screening in hospitals, reducing operational costs. Compared to other existing TAO screening methods, this solution effectively combines facial expressions and TAO symptoms for early screening. It is simple and easy to use; the required image acquisition and processing are minimal, requiring no specialized equipment. It overcomes the shortcomings of professional TAO diagnostic systems, such as limited coverage, high operational difficulty, demanding technical requirements, and complex equipment. This reduces the social harm and medical resource consumption caused by TAO, achieving early diagnosis and treatment, which is of great significance.

[0094] Furthermore, compared to the application of facial expression recognition in the screening of mental illnesses such as depression, this patent innovatively focuses on the abnormal facial expressions that patients are forced to exhibit due to pathological signs, and clarifies the correspondence between facial expressions and pathological signs. It incorporates facial expression recognition into the screening system for TAO patients, which not only provides a new approach for the screening, diagnosis and treatment of TAO itself, but also provides a reference for other similar diseases.

[0095] Meanwhile, this solution has a wide range of applications. Firstly, it can use mobile terminals, such as smartphone apps, as an interface to capture facial photos of suspected patients using the phone's camera. After a neural network model performs expression classification and judgment, the screening results are returned, and a recommendation for further hospital treatment is made, thus achieving the goal of early screening of potential cases. Furthermore, for more accurate facial information collection, a dedicated facial information collection and built-in screening device can be installed in specific medical institutions to screen patients who come for consultation and are suspected of having TAO (tonic-oncology) symptoms.

[0096] Similar in principle to the above embodiments, the present invention provides a thyroid-associated eye disease recognition system based on facial expression analysis.

[0097] The following specific embodiments are provided in conjunction with the accompanying drawings:

[0098] like Figure 4 This diagram illustrates the structure of a thyroid-associated eye disease recognition system based on facial expression analysis, as described in an embodiment of the present invention.

[0099] The system includes:

[0100] Data acquisition module 41 is used to acquire face image data of the person to be identified;

[0101] The recognition module 42 is connected to the data acquisition module 41 and is used to perform facial expression recognition on the face image data to be recognized based on the thyroid-associated eye disease recognition model obtained through training, so as to obtain the normal facial expression recognition result corresponding to normal face recognition or the diseased facial expression recognition result corresponding to thyroid-associated eye disease face recognition.

[0102] Each facial expression recognition result corresponds to a thyroid-associated ophthalmopathy sign.

[0103] It should be noted that, as should be understood Figure 4 The division of modules in the system embodiment is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, these units can be implemented entirely in software through processing element calls; they can be implemented entirely in hardware; or some units can be implemented by processing element calls to software, while others are implemented in hardware.

[0104] Since the implementation principle of the thyroid-associated ophthalmopathy recognition system based on facial expression analysis has been described in the foregoing embodiments, it will not be repeated here.

[0105] Optionally, the training method of the thyroid-associated ophthalmopathy recognition model includes: training the thyroid-associated ophthalmopathy recognition model using a facial expression training set; wherein, the facial expression training set includes: a normal facial expression recognition training set corresponding to the normal facial expression recognition results and a diseased facial expression recognition training set corresponding to the facial expression recognition results of various diseases.

[0106] Optionally, the training set for recognizing facial expressions of patients includes: multiple facial image samples of patients with angry expressions corresponding to the recognition results of angry expressions of patients; multiple facial image samples of patients with fear corresponding to the recognition results of fear expressions of patients; multiple facial image samples of patients with sadness corresponding to the recognition results of sad expressions of patients; multiple facial image samples of patients with blank expressions corresponding to the recognition results of blank expressions of patients; and multiple facial image samples of patients with disgust corresponding to the recognition results of disgust expressions of patients.

[0107] Optionally, the signs of thyroid-related ophthalmopathy include: a first sign of thyroid-related ophthalmopathy corresponding to the recognition result of an angry expression, a second sign of thyroid-related ophthalmopathy corresponding to the recognition result of a fearful expression, a third sign of thyroid-related ophthalmopathy corresponding to the recognition result of a sad expression, a fourth sign of thyroid-related ophthalmopathy corresponding to the recognition result of a blank expression, and a fifth sign of thyroid-related ophthalmopathy corresponding to the recognition result of an disgust expression; wherein, the first sign of thyroid-related ophthalmopathy includes: eyelid swelling, exophthalmos, eyelid retraction, and conjunctival hyperemia; the second sign of thyroid-related ophthalmopathy includes: eyelid swelling, eyelid retraction, and exophthalmos; the third sign of thyroid-related ophthalmopathy includes: conjunctival edema and eyelid swelling; the fourth sign of thyroid-related ophthalmopathy includes: abnormal eye position, eyelid retraction, and exophthalmos; and the fifth sign of thyroid-related ophthalmopathy includes: eyelid swelling and abnormal eye position.

[0108] Optionally, the disease-related anger features include disease-related anger eye features; the disease-related fear features include disease-related fear eye features; the disease-related sadness features include disease-related sadness eye features; the disease-related sadness features include disease-related sadness eye features; the disease-related listlessness features include disease-related listlessness eye features; the disease-related sadness features include disease-related sadness eye features; and the disease-related aversion features include disease-related fear aversion features.

[0109] Optionally, the symptoms of anger include forehead and eyebrow features and / or lower facial features; the symptoms of fear include forehead and eyebrow features and / or lower facial features; the symptoms of sadness include forehead and eyebrow features and / or lower facial features; the symptoms of listlessness include forehead and eyebrow features and / or lower facial features; and the symptoms of aversion include forehead and eyebrow features and / or lower facial features.

[0110] Optionally, the normal facial expression recognition training set includes: multiple expressionless facial image samples with expressionless features corresponding to the expressionless expression recognition result; multiple normal angry facial image samples with normal angry features corresponding to the normal angry expression recognition result; multiple normal fear facial image samples with normal fear features corresponding to the normal fear expression recognition result; multiple normal sad facial image samples with normal sad features corresponding to the normal sad expression recognition result; multiple normal blank facial image samples with normal blank features corresponding to the normal dull expression recognition result; and multiple normal disgust facial image samples with normal disgust features corresponding to the normal disgust expression recognition result.

[0111] Optionally, the thyroid-associated ophthalmopathy identification model adopts a VGG-19 network structure.

[0112] like Figure 5 A schematic diagram of the structure of a thyroid-associated eye disease recognition terminal 10 based on facial expression analysis in an embodiment of the present invention is shown.

[0113] The thyroid-associated eye disease recognition terminal 50 based on facial expression analysis includes: a memory 51 and a processor 52. The memory 51 is used to store computer programs; the processor 52 runs the computer programs to implement, for example... Figure 1 The method for identifying thyroid-related eye diseases based on facial expression analysis.

[0114] Optionally, the number of memories 51 can be one or more, and the number of processors 52 can be one or more. Figure 5 Each example is taken as an instance.

[0115] Optionally, the processor 52 in the thyroid-associated ophthalmopathy recognition terminal 50 based on facial expression analysis will perform the following... Figure 1 The steps described involve loading one or more instructions corresponding to the process of an application into memory 51, and having the processor 52 run the application stored in the first memory 51, thereby achieving the following: Figure 1 Various functions in the thyroid-associated ophthalmopathy identification method based on facial expression analysis.

[0116] Optionally, the memory 51 may include, but is not limited to, high-speed random access memory and non-volatile memory. For example, one or more disk storage devices, flash memory devices, or other non-volatile solid-state storage devices; the processor 52 may include, but is not limited to, a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0117] Optionally, the processor 52 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0118] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed, implements as follows: Figure 1 The illustrated method for identifying thyroid-associated eye disease based on facial expression analysis. The computer-readable storage medium may include, but is not limited to, floppy disks, optical disks, CD-ROMs (Read-Only Optical Disk Memory), magneto-optical disks, ROMs (Read-Only Memory), RAMs (Random Access Memory), EPROMs (Erasable Programmable Read-Only Memory), EEPROMs (Electrically Erasable Programmable Read-Only Memory), magnetic cards or optical cards, flash memory, or other types of media / machine-readable media suitable for storing machine-executable instructions. The computer-readable storage medium may be a product not connected to a computer device or a component used with a computer device.

[0119] In summary, the thyroid-associated ophthalmopathy (TAO) identification system based on facial expression analysis of this invention uses a trained TAO identification model to perform facial expression recognition on acquired facial image data to obtain normal facial expression recognition results for normal faces or diseased facial expression recognition results for faces of TAO patients. This invention addresses scientific issues such as the description of facial expression features in TAO patients and expression analysis based on neural networks, ultimately achieving the goal of TAO screening. This facilitates users' assessment of their own eye health, saving significant economic and time costs. Simultaneously, it effectively improves the efficiency of hospital TAO screening, reduces operating costs, and promotes the widespread adoption and application of efficient, convenient, and low-cost orbital disease screening methods in my country. Therefore, this invention effectively overcomes the various shortcomings of existing technologies and possesses high industrial application value.

[0120] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for identifying thyroid-associated ophthalmopathy based on facial expression analysis, characterized in that, The method includes: Acquire the facial image data of the person to be identified; Based on the trained thyroid-associated ophthalmopathy (TAO) recognition model, facial expression recognition is performed on the facial image data to be identified, to obtain normal facial expression recognition results corresponding to normal facial recognition or diseased facial expression recognition results corresponding to TAO patient facial recognition. The training method for the TAO recognition model includes: training the model using a facial expression training set; wherein the facial expression training set includes: a normal facial expression recognition training set corresponding to normal facial expression recognition results and a diseased facial expression recognition training set corresponding to various diseased facial expression recognition results; normal facial expressions include expressionless, normal angry, normal fearful, normal sad, normal blank, and normal disgusted expressions; diseased facial expressions include diseased angry, diseased fearful, diseased sad, diseased blank, and diseased disgusted expressions. Each facial expression recognition result corresponds to a thyroid-associated ophthalmopathy sign; the thyroid-associated ophthalmopathy signs include: the first thyroid-associated ophthalmopathy sign corresponding to the recognition result of the angry expression, the second thyroid-associated ophthalmopathy sign corresponding to the recognition result of the fear expression, the third thyroid-associated ophthalmopathy sign corresponding to the recognition result of the sad expression, the fourth thyroid-associated ophthalmopathy sign corresponding to the recognition result of the dull expression, and the fifth thyroid-associated ophthalmopathy sign corresponding to the recognition result of the disgust expression. The first thyroid-associated ophthalmopathy sign includes: eyelid swelling, proptosis, eyelid retraction, and conjunctival hyperemia; the second thyroid-associated ophthalmopathy sign includes: eyelid swelling, eyelid retraction, and proptosis; the third thyroid-associated ophthalmopathy sign includes: conjunctival edema and eyelid swelling; the fourth thyroid-associated ophthalmopathy sign includes: abnormal eye position, eyelid retraction, and proptosis; and the fifth thyroid-associated ophthalmopathy sign includes: eyelid swelling and eyelid retraction.

2. The method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to claim 1, characterized in that, The training set for recognizing facial expressions of patients includes: multiple facial image samples of patients with angry expressions corresponding to the recognition results of angry expressions of patients; multiple facial image samples of patients with fearful expressions corresponding to the recognition results of fearful expressions of patients; multiple facial image samples of patients with sad expressions corresponding to the recognition results of sad expressions of patients; multiple facial image samples of patients with blank expressions corresponding to the recognition results of blank expressions of patients; and multiple facial image samples of patients with disgusted expressions corresponding to the recognition results of disgusted expressions of patients.

3. The method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to claim 2, characterized in that, The symptoms of illness include angry eye features; the symptoms of illness include fearful eye features; the symptoms of illness include sad eye features; the symptoms of illness include listless eye features; and the symptoms of illness include disgusting eye features.

4. The method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to claim 3, characterized in that, The symptoms of anger include the forehead and eyebrow features of anger and / or the lower part of the face; the symptoms of fear include the forehead and eyebrow features of fear and / or the lower part of the face; the symptoms of sadness include the forehead and eyebrow features of sadness and / or the lower part of the face; the symptoms of listlessness include the forehead and eyebrow features of listlessness and / or the lower part of the face; the symptoms of disgust include the forehead and eyebrow features of disgust and / or the lower part of the face.

5. The method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to claim 2, characterized in that, The normal facial expression recognition training set includes: multiple expressionless face image samples with expressionless features corresponding to the expressionless expression recognition result; multiple normal angry face image samples with normal angry features corresponding to the normal angry expression recognition result; multiple normal fear face image samples with normal fear features corresponding to the normal fear expression recognition result; multiple normal sad face image samples with normal sad features corresponding to the normal sad expression recognition result; multiple normal blank face image samples with normal blank features corresponding to the normal blank expression recognition result; and multiple normal disgust face image samples with normal disgust features corresponding to the normal disgust expression recognition result.

6. The method for identifying thyroid-associated ophthalmopathy based on facial expression analysis according to claim 1, characterized in that, The thyroid-associated ophthalmopathy identification model uses a VGG-19 network structure.

7. A thyroid-associated ophthalmopathy recognition system based on facial expression analysis, characterized in that, The system includes: The data acquisition module is used to acquire the face image data of the person to be identified; The recognition module, connected to the data acquisition module, is used to perform facial expression recognition on the facial image data to be recognized based on the trained thyroid-associated ophthalmopathy (TAO) recognition model, to obtain normal facial expression recognition results corresponding to normal facial recognition or diseased facial expression recognition results corresponding to TAO patient facial recognition. The training method of the TAO recognition model includes: training the TAO recognition model using a facial expression training set; wherein the facial expression training set includes: a normal facial expression recognition training set corresponding to normal facial expression recognition results and a diseased facial expression recognition training set corresponding to various diseased facial expression recognition results; normal facial expressions include expressionless, normal angry, normal fear, normal sad, normal blank, and normal disgusted expressions; diseased facial expressions include diseased angry, diseased fear, diseased sad, diseased blank, and diseased disgusted expressions. Each facial expression recognition result corresponds to a thyroid-associated ophthalmopathy sign; the thyroid-associated ophthalmopathy signs include: the first thyroid-associated ophthalmopathy sign corresponding to the recognition result of the angry expression, the second thyroid-associated ophthalmopathy sign corresponding to the recognition result of the fear expression, the third thyroid-associated ophthalmopathy sign corresponding to the recognition result of the sad expression, the fourth thyroid-associated ophthalmopathy sign corresponding to the recognition result of the dull expression, and the fifth thyroid-associated ophthalmopathy sign corresponding to the recognition result of the disgust expression. The first thyroid-associated ophthalmopathy sign includes: eyelid swelling, proptosis, eyelid retraction, and conjunctival hyperemia; the second thyroid-associated ophthalmopathy sign includes: eyelid swelling, eyelid retraction, and proptosis; the third thyroid-associated ophthalmopathy sign includes: conjunctival edema and eyelid swelling; the fourth thyroid-associated ophthalmopathy sign includes: abnormal eye position, eyelid retraction, and proptosis; and the fifth thyroid-associated ophthalmopathy sign includes: eyelid swelling and eyelid retraction.

8. A terminal for recognizing thyroid-related eye diseases based on facial expression analysis, characterized in that, include: One or more memories and one or more processors; The one or more memories are used to store computer programs; The one or more processors are connected to the memory and are used to run the computer program to perform the method as described in any one of claims 1 to 6.

Images