AI auxiliary device special for fundus blood vessel anomaly detection

Abstract

The invention discloses a special AI auxiliary device for fundus blood vessel anomaly detection, which comprises a base, a system terminal, a stand column, a lifting assembly, an anti-blink assembly, a sliding assembly, a cleaning assembly and a pressure assembly, and is characterized in that the system terminal is fixedly mounted on the base. Through an anti-blink assembly, after two curved groove blocks slide to the two ends of a hyperbolic rod respectively, a patient attaches the eyebrow bone to a gap in the hyperbolic rod, wears an elastic strap on the head, attaches an air support to the jaw through a lifting assembly, and then screws a second lead screw to enable a trapezoidal block to slide backwards, at the moment, the cambered surface of the support block descends along the inclined surface of the trapezoidal block, and the air support is fixed to the upper portion of the hyperbolic rod. When the head of a patient descends along with the air support, the double curved bars keep the eyelids at the eyebrow bones still, the downward movement of the head and the fixation of the double curved bars form relative displacement, so that the upper eyelids are stably pulled up, the eyelids are kept at the eye opening amplitude, the upper eyelids are prevented from naturally drooping or reflexively closing, and imaging missing or blurring caused by blinking and eyelid drooping is avoided; and the AI detection data is ensured to be complete and effective.

Description

Technical Field

[0001] This invention relates to the field of fundus vascular detection technology, and in particular to an AI-assisted device for detecting fundus vascular abnormalities. Background Technology

[0002] Fundus vascular examination is a key means of diagnosing various fundus diseases such as diabetic retinopathy, glaucoma, and macular degeneration. By collecting fundus vascular images, the morphology of blood vessels and the characteristics of lesion areas can be observed directly, providing an important basis for the early screening, diagnosis and treatment of diseases. It plays an irreplaceable role in ophthalmic clinical diagnosis and treatment, chronic disease management and public health screening.

[0003] Current fundus vascular examinations mainly involve acquiring fundus images of patients using specialized fundus cameras. During the examination, patients need to align their eyes with the camera lens. Some devices are equipped with head fixation devices or forehead and jaw supports to help patients maintain head stability and ensure accurate image positioning. The acquired fundus images are transmitted to a computer terminal, where ophthalmologists analyze the images to identify morphological changes in blood vessels, the presence of hemorrhage, exudation, microaneurysms, and other lesion characteristics, and then make a diagnostic conclusion.

[0004] However, during the image acquisition stage of fundus vascular detection, patients are prone to reflexive blinking and natural eyelid drooping, making it impossible for them to maintain a continuously open eye state. This can easily lead to missing or blurred fundus images, making it difficult to completely capture the fundus lesion area. The quality of the acquired images is greatly affected by blinking-related factors, which directly impacts the accurate identification and analysis of fundus vascular lesion characteristics in the future.

[0005] Accordingly, this application proposes a dedicated AI-assisted device for detecting abnormalities in retinal blood vessels. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by proposing a dedicated AI-assisted device for detecting abnormal retinal blood vessels.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] An AI-assisted device for detecting abnormal retinal blood vessels includes a base, a system terminal, a column, a lifting component, an anti-blinking component, a sliding component, a cleaning component, and a pressure component. The system terminal is fixedly mounted on the base, the column is fixedly connected to the base, a top block is fixedly connected to one end of the column, an eye mask is fixedly connected to the top block, two glass lenses are fixedly installed inside the eye mask, and an elastic strap is fixedly installed on the eye mask.

[0009] The lifting assembly includes a knob, a first bevel gear, a second bevel gear, a first lead screw, and a bracket; the knob is rotatably connected to the top block, one end of the first lead screw is rotatably connected to the top block, and the other end is rotatably connected to the base, and the first bevel gear is fixedly connected to the knob;

[0010] The lifting assembly is used to adjust the detection height according to the patient's height;

[0011] The anti-blinking component is used to prevent the patient from blinking during the test;

[0012] The cleaning component is used to clean the area around the patient's eyes of oil, foundation, etc. before testing;

[0013] The sliding component is used to drive the cleaning component to work;

[0014] The pressure component is used to alert the patient to excessive pressure on their chin.

[0015] Preferably, the eye mask has airbags on both sides of its edge, a second sliding groove on its surface, a second bevel gear fixedly connected to one end of a first lead screw, the first bevel gear meshing with the second bevel gear, and the bracket slidably connected to the column and the first lead screw.

[0016] Preferably, the anti-blinking component includes a hyperbola, a first slide groove, a support block, a second lead screw, and a trapezoidal block; the hyperbola is fixedly connected inside the eye mask, the support block is slidably connected to the bracket, and the first slide groove is provided on the bracket.

[0017] Preferably, the second lead screw is slidably connected to the bracket, the trapezoidal block is slidably connected to the first slide groove, the trapezoidal block is rotatably connected to the second lead screw, and the lower arc surface of the bracket is in contact with the inclined surface of the trapezoidal block.

[0018] Preferably, the sliding assembly includes an electric telescopic rod, a mounting block, two connecting rods, two short rods, two long rods, and two curved groove blocks; the electric telescopic rod is fixedly mounted on the top block, the mounting block is fixedly connected to the output end of the electric telescopic rod, the two connecting rods are hinged to each other, and the two short rods are slidably connected to the second sliding groove respectively.

[0019] Preferably, the two long rods are fixedly connected to the bottom of the two curved groove blocks, the connecting rod is rotatably connected to the mounting block, the two curved groove blocks are rotatably connected to the double-curved rod, and one end of the two short rods is rotatably connected to the two connecting rods, and the other end is rotatably connected to the two curved groove blocks.

[0020] Preferably, the cleaning assembly includes several springs, two rectangular groove blocks, four top plates, four pull rods, four pressure plates, and several conical spikes; the two long rods are respectively fixedly connected to the bottom of the two curved groove blocks, and the two rectangular groove blocks are respectively fixedly connected to the bottom of the two long rods.

[0021] Preferably, the four top plates are fixedly connected in pairs within the two rectangular slots, the four tie rods are slidably connected in pairs to both sides of the two rectangular slots, and the four pressure plates are fixedly connected to one end of the four tie rods.

[0022] Preferably, a plurality of the cones are fixedly connected to four pressure plates, and a plurality of springs are respectively disposed at both ends of the four pull rods, with one end of the spring fixedly connected to the pressure plate and the other end fixedly connected to the rectangular groove block.

[0023] Preferably, the pressure assembly includes an air support, a pressure cylinder, and a sensor; the air support is fixedly mounted on the support block, the pressure cylinder is connected to the air support, and the sensor is fixedly mounted inside the pressure cylinder.

[0024] The present invention has the following beneficial effects:

[0025] 1. Using the cleaning assembly, pull the lever to place one end of the medical cleaning pad between the cone and the top plate. Then release the lever, and the medical cleaning pad will be clamped between the cone and the top plate. Repeat the above operation on the other end of the medical cleaning pad, so that the top plate lifts the cleaning pad. After the cleaning pads are installed on both rectangular slots, the electric telescopic rod drives the connecting rod to slide forward, so that the short rod drives the two curved slot blocks to slide towards the two ends of the double curved rod along the second slide groove. This allows the cleaning pad to adhere to and wipe the skin below the patient's eyes, avoiding impurities such as oil and foundation around the patient's eyes from affecting the fundus vascular test results.

[0026] Second, using the anti-blinking component, after the two curved blocks slide to the two ends of the hyperboloid, the patient places their brow bone against the gap on the hyperboloid and puts the elastic band on their head. Then, the lifting component moves the air support against the chin. The second screw is then turned to slide the trapezoidal block backward. At this time, the curved surface of the support block descends along the slope of the trapezoidal block. As the patient's head descends with the air support, the hyperboloid keeps the eyelid at the brow bone stationary. The downward movement of the head and the fixation of the hyperboloid create a relative displacement, completing the smooth lifting of the upper eyelid. This maintains the eyelid's open range, preventing the upper eyelid from drooping naturally or reflexively closing, avoiding image loss or blurring due to blinking or eyelid drooping, and ensuring the integrity and effectiveness of the AI ​​detection data.

[0027] Third, through the pressure component, when the patient places their chin on the air support, the chin pressure is transmitted to the air support. The gas inside the air support is pressurized and enters the pressure cylinder. The sensor detects the gas pressure inside the pressure cylinder and transmits the signal to the system terminal, realizing real-time detection of the patient's chin pressure. When the pressure exceeds the set range, the system terminal will prompt the medical staff to adjust the height of the support to avoid discomfort in the patient's jaw due to excessive pressure, which may cause the patient's head to shake. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall structure of an AI-assisted device for detecting fundus vascular abnormalities proposed in this invention.

[0029] Figure 2 This is a schematic diagram of the connection structure of the eye mask, electric telescopic rod, and connecting rod of the AI-assisted device for detecting abnormal retinal blood vessels proposed in this invention.

[0030] Figure 3 This is a schematic diagram of the connection structure of components such as a short rod, a long rod, and a rectangular groove block in an AI-assisted device for detecting fundus vascular abnormalities proposed in this invention.

[0031] Figure 4 for Figure 1 Enlarged diagram of point A in the diagram;

[0032] Figure 5 This is an internal cross-sectional view of the top block of an AI-assisted device for detecting fundus vascular abnormalities proposed in this invention.

[0033] Figure 6 This is an internal cross-sectional view of a bracket for a special AI-assisted device for detecting fundus vascular abnormalities proposed in this invention;

[0034] Figure 7 This is a schematic diagram of the connection structure of the internal components of a rectangular slot in an AI-assisted device for detecting fundus vascular abnormalities proposed in this invention.

[0035] In the diagram: 1. Base; 2. System terminal; 3. Bracket; 4. Knob; 5. Column; 6. Top block; 7. Eye mask; 8. Elastic strap; 9. Support block; 10. Air support; 11. First lead screw; 12. Mounting block; 13. Electric telescopic rod; 14. Connecting rod; 15. Airbag; 16. Second slide groove; 17. Short rod; 18. Hyperboloid rod; 19. Curved groove block; 20. Rectangular groove block; 21. Long rod; 22. Pull rod; 23. Top plate; 24. Spring; 25. Glass mirror; 26. Conical spike; 27. First bevel gear; 28. Second bevel gear; 29. ​​Pressure cylinder; 30. First slide groove; 31. Second lead screw; 32. Trapezoidal block; 33. Sensor; 34. Pressure plate. Detailed Implementation

[0036] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0037] Example 1:

[0038] Reference Figure 1 , Figure 2 , Figure 4 and Figure 5An AI-assisted device for detecting fundus vascular abnormalities includes a base 1, a system terminal 2, a column 5, a lifting component, an anti-blinking component, a sliding component, a cleaning component, and a pressure component. The system terminal 2 is fixedly installed on the base 1 and is the core control and analysis unit of the device. It integrates a deep learning AI algorithm engine based on a Transformer+CNN hybrid model and has a built-in core model trained on a large number of high-quality fundus images and annotated by multiple ophthalmologists. It can realize fully automated processing from image reception and analysis to report generation. The AI ​​module can accurately identify a variety of common fundus diseases such as diabetic retinopathy, glaucoma, and macular degeneration, and complete lesion area detection, feature extraction, disease classification, and severity grading, effectively solving the pain points of low efficiency, high missed diagnosis rate, and poor diagnostic consistency of traditional manual image reading. The system terminal 2 also has data management functions, supports image archiving, historical comparison, and statistical analysis, and can automatically generate structured reports containing lesion descriptions, diagnostic conclusions, and referral suggestions, adapting to multi-terminal access and data interaction.

[0039] The column 5 is fixedly connected to the base 1. One end of the column 5 is fixedly connected to the top block 6. The top block 6 is fixedly connected to the eye shield 7. Two glass mirrors 25 are fixedly installed inside the eye shield 7. A fundus vascular camera is installed behind the glass mirrors 25 inside the eye shield 7. The fundus vascular camera is used to collect images of the patient's fundus blood vessels and transmit them to the system terminal 2. The glass mirrors 25 are used to ensure that the imaging optical path is unobstructed and to avoid foreign objects from interfering with the acquisition of fundus images. An elastic strap 8 is fixedly installed on the eye shield 7. The elastic strap 8 is used to fix the eye shield 7 to the patient's head and ensure that the relative position of the equipment and the eye is stable during the detection process.

[0040] The lifting assembly includes a knob 4, a first bevel gear 27, a second bevel gear 28, a first lead screw 11, and a bracket 3. The knob 4 is rotatably connected to the top block 6, one end of the first lead screw 11 is rotatably connected to the top block 6, and the other end is rotatably connected to the base 1. The first bevel gear 27 is fixedly connected to the knob 4. By turning the knob 4, the first bevel gear 27 is driven to rotate, and the first bevel gear 27 meshes with and drives the second bevel gear 28 to rotate, which in turn drives the first lead screw 11 to rotate, so that the bracket 3 slides up and down along the column 5 and the first lead screw 11 to accommodate patients of different heights, ensure the standardization of fundus image acquisition, and provide standard input materials for AI analysis.

[0041] The lifting component is used to adjust the detection height according to the patient's height, adapting to patients of different heights, ensuring the standardization of fundus image acquisition, and providing standard input materials for AI analysis;

[0042] The anti-blinking component is used to prevent patients from blinking during the test, ensuring that the fundus image is complete and clear, providing a high-quality image foundation for accurate AI recognition, and reducing AI misjudgment caused by missing images;

[0043] The cleaning component is used to clean the oil, foundation, and other substances around the patient's eyes before the test, to prevent debris from blocking the light path and contaminating the lens, thereby improving image quality and ensuring that AI can accurately extract the retinal blood vessels and lesion features.

[0044] The sliding component drives the cleaning component to achieve automatic cleaning around the eyes, improves detection preparation efficiency, and ensures the continuity of the AI ​​analysis process.

[0045] Airbags 15 are provided on both sides of the eye mask 7. The airbags 15 are used to fit the sides of the patient's face to better fit the patient's skin, further reduce the interference of ambient light, ensure image quality, and provide good conditions for AI to accurately identify. The eye mask 7 is provided with a second slide groove 16 to provide a directional sliding track for the sliding component. The second bevel gear 28 is fixedly connected to one end of the first lead screw 11. The first bevel gear 27 is meshed with the second bevel gear 28. The bracket 3 is slidably connected to the column 5 and the first lead screw 11. The first lead screw 11 is rotated through gear transmission, thereby realizing the lifting and lowering of the bracket 3, accurately adjusting the detection height, and ensuring that the fundus image acquisition position meets the standard requirements of AI analysis.

[0046] In this embodiment, the patient wears the elastic band 8, which fits better against the patient's skin via the airbag 15. Then, the detection height is adjusted using the lifting component. Turning the knob 4 drives the first bevel gear 27 to rotate, which in turn drives the second bevel gear 28 to rotate, thereby driving the first lead screw 11 to rotate. This causes the bracket 3 to slide up and down along the column 5 and the first lead screw 11 until it is adapted to the patient's height and the head is positioned. The fundus vascular camera behind the glass lens 25 acquires images of the patient's fundus vessels and transmits the image data to the system terminal 2. The system terminal 2, as the core control and analysis unit, uses its integrated deep learning AI algorithm engine to automatically process the received fundus images, complete lesion area detection, feature extraction, disease classification, and severity grading. Finally, the system terminal 2 automatically generates a structured report containing lesion description, diagnostic conclusion, and referral recommendations, achieving accurate screening and diagnosis of fundus vascular abnormalities.

[0047] Example 2:

[0048] Unlike Example 1, referring to Figures 1 to 4 and Figure 7 This embodiment also has the following further features:

[0049] The sliding assembly includes an electric telescopic rod 13, a mounting block 12, two connecting rods 14, two short rods 17, two long rods 21, and two curved groove blocks 19. The electric telescopic rod 13 is fixedly mounted on the top block 6 to provide power output for the sliding assembly. The mounting block 12 is fixedly connected to the output end of the electric telescopic rod 13 to receive the power of the electric telescopic rod 13 and transmit it to the connecting rods 14. The two connecting rods 14 are hinged to each other and can realize power steering and transmission through angle changes. The two short rods 17 are slidably connected to the second slide groove 16, which provides a directional sliding track for the short rods 17 to ensure stable movement trajectory.

[0050] Two long rods 21 are fixedly connected to the bottom of two curved groove blocks 19 respectively, used to connect the curved groove blocks 19 to the cleaning components and transmit motion. The connecting rod 14 is rotatably connected to the mounting block 12 and can rotate around the mounting block 12 to adapt to the power transmission angle. The two curved groove blocks 19 are rotatably connected to the double curved rod 18 respectively, so that the curved groove blocks 19 can move flexibly with the drive of the connecting rod 14. One end of the two short rods 17 is rotatably connected to the two connecting rods 14 respectively, and the other end is rotatably connected to the two curved groove blocks 19 respectively, realizing the power transmission between the connecting rod 14 and the curved groove blocks 19, and ensuring the linkage and coordination of each component.

[0051] The cleaning assembly includes several springs 24, two rectangular slot blocks 20, four top plates 23, four pull rods 22, four pressure plates 34, and several conical spikes 26; two long rods 21 are fixedly connected to the bottom of two curved slot blocks 19, and two rectangular slot blocks 20 are fixedly connected to the bottom of two long rods 21, which are used to install other parts of the cleaning assembly and drive its overall movement.

[0052] Four top plates 23 are fixedly connected in pairs within two rectangular slots 20, and cooperate with the spikes 26 to clamp the medical cleaning pads. Four pull rods 22 are slidably connected in pairs to both sides of the two rectangular slots 20, and are used to drive the pressure plate 34 and the spikes 26 to move to load and unload the cleaning pads. Four pressure plates 34 are fixedly connected to one end of the four pull rods 22, and are used to install the spikes 26 and transmit the driving force of the pull rods 22.

[0053] Several conical spikes 26 are fixedly connected to four pressure plates 34 to enhance the clamping stability of the medical cleaning pad and prevent it from falling off during the cleaning process. Several springs 24 are respectively located at both ends of the four pull rods 22. One end of the spring 24 is fixedly connected to the pressure plate 34 and the other end is fixedly connected to the rectangular groove block 20 to provide elastic restoring force for the pull rods 22 and ensure that the clamping force of the pressure plate 34 and the conical spikes 26 on the cleaning pad is constant.

[0054] In this embodiment, pulling the lever 22 places one end of the medical cleaning pad between the cone 26 and the top plate 23. Then, releasing the lever 22, the medical cleaning pad is clamped between the cone 26 and the top plate 23. The other end of the medical cleaning pad is also operated in the same way, so that the top plate 23 lifts the cleaning pad. After the cleaning pads are installed on both rectangular slots 20, the electric telescopic rod 13 drives the connecting rod 14 to slide forward, so that the short rod 17 drives the two curved slot blocks 19 to slide along the second sliding groove 16 towards the two ends of the double curved rod 18, thereby making the cleaning pad adhere to and wipe the skin below the patient's eyes, avoiding impurities such as oil and foundation around the patient's eyes from affecting the fundus blood vessel detection results.

[0055] Example 3:

[0056] Reference Figures 1 to 3 , Figure 5 and Figure 6 Compared to Embodiment 1 and Embodiment 2, in this embodiment:

[0057] The anti-blinking component includes a hyperbola 18, a first slide groove 30, a support block 9, a second lead screw 31, and a trapezoidal block 32. The hyperbola 18 is fixedly connected inside the eye mask 7 to fit the patient's brow bone and provide support for the eyelid at the brow bone. The support block 9 is slidably connected to the bracket 3 and can move up and down along the bracket 3 to adapt to different patients' eye positions. The first slide groove 30 is provided on the bracket 3 to provide a directional sliding track for the trapezoidal block 32, ensuring the stability of its movement trajectory.

[0058] The second lead screw 31 is slidably connected to the bracket 3 as an adjustment power transmission component. The trapezoidal block 32 is slidably connected to the first slide groove 30. The trapezoidal block 32 is rotatably connected to the second lead screw 31. The rotation of the second lead screw 31 drives the trapezoidal block 32 to slide along the first slide groove 30. The lower arc surface of the support block 9 is in contact with the inclined surface of the trapezoidal block 32. When the trapezoidal block 32 slides, the inclined surface drives the support block 9 to move up and down, thereby limiting and supporting the lower eyelid and preventing the patient from blinking during the test.

[0059] The pressure assembly includes an air support 10, a pressure cylinder 29, and a sensor 33. The air support 10 is fixedly installed on the support block 9 to fit the patient's chin and provide support. The pressure cylinder 29 is connected to the air support 10 to transmit the pressure borne by the air support 10. The sensor 33 is fixedly installed inside the pressure cylinder 29 to detect pressure changes inside the pressure cylinder 29 in real time. When the pressure exceeds a set threshold, a prompt is issued to prevent the patient's chin from being under too much pressure, which could cause discomfort or head shaking.

[0060] In this embodiment, after the two slot blocks 19 slide to the two ends of the hyperbolic rod 18, the patient places their brow bone against the gap on the hyperbolic rod 18, puts the elastic band 8 on their head, and then uses the lifting assembly to place the air support 10 against their chin. Then, the second lead screw 31 is turned to slide the trapezoidal block 32 backward. At this time, the arc surface of the support block 9 descends along the inclined surface of the trapezoidal block 32. As the patient's head descends with the air support 10, the hyperbolic rod 18 keeps the eyelid at the brow bone stationary. The downward movement of the head and the fixation of the hyperbolic rod 18 create a relative displacement, completing the smooth lifting of the upper eyelid, maintaining the eyelid's open range, and preventing the upper eyelid from drooping naturally or falling. Reflexive closure prevents image loss or blurring due to blinking or drooping eyelids, ensuring complete and effective AI detection data. When the patient places their chin on the air support 10, the chin pressure is transmitted to the air support 10. The gas inside the air support 10 is pressurized and enters the pressure cylinder 29. The sensor 33 detects the gas pressure inside the pressure cylinder 29 and transmits the signal to the system terminal 2, realizing real-time detection of the patient's chin pressure. When the pressure exceeds the set range, the system terminal 2 prompts the medical staff to adjust the height of the support 3 to avoid discomfort in the patient's jaw due to excessive pressure, which could cause the patient's head to shake.

[0061] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A dedicated AI-assisted device for detecting fundus vascular abnormalities, characterized in that, The system includes a base (1), a system terminal (2), a column (5), a lifting assembly, an anti-blink assembly, a sliding assembly, a cleaning assembly, and a pressure assembly. The system terminal (2) is fixedly installed on the base (1), the column (5) is fixedly connected to the base (1), a top block (6) is fixedly connected to one end of the column (5), an eye mask (7) is fixedly connected to the top block (6), two glass mirrors (25) are fixedly installed inside the eye mask (7), and an elastic strap (8) is fixedly installed on the eye mask (7). The lifting assembly includes a knob (4), a first bevel gear (27), a second bevel gear (28), a first lead screw (11), and a bracket (3); the knob (4) is rotatably connected to the top block (6), one end of the first lead screw (11) is rotatably connected to the top block (6), and the other end is rotatably connected to the base (1), and the first bevel gear (27) is fixedly connected to the knob (4); The lifting assembly is used to adjust the detection height according to the patient's height; The anti-blinking component is used to prevent the patient from blinking during the test; The cleaning component is used to clean the area around the patient's eyes of oil, foundation, etc. before testing; The sliding component is used to drive the cleaning component to work; The pressure component is used to alert the patient to excessive pressure on their chin.

2. The AI-assisted device for detecting fundus vascular abnormalities according to claim 1, characterized in that, The eye mask (7) has airbags (15) on both sides of its edge. The eye mask (7) has a second groove (16). The second bevel gear (28) is fixedly connected to one end of the first lead screw (11). The first bevel gear (27) meshes with the second bevel gear (28). The bracket (3) is slidably connected to the column (5) and the first lead screw (11).

3. The AI-assisted device for detecting fundus vascular abnormalities according to claim 1, characterized in that, The anti-blinking component includes a hyperbola (18), a first groove (30), a support block (9), a second lead screw (31), and a trapezoidal block (32); the hyperbola (18) is fixedly connected inside the eye mask (7), the support block (9) is slidably connected to the bracket (3), and the first groove (30) is provided on the bracket (3).

4. The AI-assisted device for detecting fundus vascular abnormalities according to claim 3, characterized in that, The second lead screw (31) is slidably connected to the bracket (3), the trapezoidal block (32) is slidably connected to the first slide groove (30), the trapezoidal block (32) is rotatably connected to the second lead screw (31), and the lower arc surface of the support block (9) is in contact with the inclined surface of the trapezoidal block (32).

5. The AI-assisted device for detecting fundus vascular abnormalities according to claim 2, characterized in that, The sliding assembly includes an electric telescopic rod (13), a mounting block (12), two connecting rods (14), two short rods (17), two long rods (21), and two curved groove blocks (19); the electric telescopic rod (13) is fixedly mounted on the top block (6), the mounting block (12) is fixedly connected to the output end of the electric telescopic rod (13), the two connecting rods (14) are hinged to each other, and the two short rods (17) are slidably connected to the second sliding groove (16).

6. The AI-assisted device for detecting fundus vascular abnormalities according to claim 5, characterized in that, The two long rods (21) are fixedly connected to the bottom of the two curved groove blocks (19), the connecting rod (14) is rotatably connected to the mounting block (12), the two curved groove blocks (19) are rotatably connected to the double curved rod (18), one end of the two short rods (17) is rotatably connected to the two connecting rods (14), and the other end is rotatably connected to the two curved groove blocks (19).

7. The AI-assisted device for detecting fundus vascular abnormalities according to claim 5, characterized in that, The cleaning assembly includes several springs (24), two rectangular slots (20), four top plates (23), four pull rods (22), four pressure plates (34), and several spikes (26); two long rods (21) are fixedly connected to the bottom of two curved slots (19), and two rectangular slots (20) are fixedly connected to the bottom of two long rods (21).

8. The AI-assisted device for detecting fundus vascular abnormalities according to claim 7, characterized in that, The four top plates (23) are fixedly connected in pairs to the two rectangular slots (20), the four tie rods (22) are slidably connected in pairs to both sides of the two rectangular slots (20), and the four pressure plates (34) are fixedly connected to one end of the four tie rods (22).

9. The AI-assisted device for detecting fundus vascular abnormalities according to claim 7, characterized in that, Several of the cones (26) are fixedly connected to four pressure plates (34), and several springs (24) are respectively located at both ends of four pull rods (22). One end of the spring (24) is fixedly connected to the pressure plate (34), and the other end is fixedly connected to the rectangular groove block (20).

10. The AI-assisted device for detecting fundus vascular abnormalities according to claim 3, characterized in that, The pressure assembly includes an air support (10), a pressure cylinder (29), and a sensor (33); the air support (10) is fixedly installed on the support block (9), the pressure cylinder (29) is connected to the air support (10), and the sensor (33) is fixedly installed inside the pressure cylinder (29).

Images