An automated pupil size recognition determination device

Abstract

The application relates to the technical field of medical detection equipment, and discloses an automatic pupil size recognition and determination device, which comprises a shell, a control unit and a power supply unit arranged in the shell; the shell comprises a detection head and a holding part connected with the detection head, the detection head is internally provided with an AI vision recognition module, a self-adaptive light compensation module, a distance calibration module and a main control circuit board; a light transmission window is arranged on the front end surface of the detection head; the AI vision recognition module comprises a CMOS image sensor and an AI processing chip electrically connected with the CMOS image sensor; the CMOS image sensor comprises a lens, the lens is directly opposite the central region of the light transmission window and is used for collecting a pupil image; the AI processing chip is internally programmed with a pupil feature recognition model and is used for automatically extracting pupil contour and edge features and outputting quantitative data. The application integrates AI vision, self-adaptive infrared light compensation and laser distance calibration, eliminates light interference and perspective deviation, automatically measures and calculates the real diameter of the pupil, and avoids artificial subjective errors.

Description

Technical Field

[0001] This invention relates to the field of medical testing equipment technology, specifically to an automated pupil size recognition and determination device. Background Technology

[0002] In scenarios involving the detection of pupil size in critically ill patients such as those who are unconscious, poisoned, undergoing emergency treatment, or suffering from traumatic brain injury, pupil size is a core objective indicator for clinically assessing the severity of brain injury, poisoning, and central nervous system function. Accurate and stable pupil size detection results directly impact the scientific validity and safety of clinical treatment plans. Currently, clinical methods for identifying and determining patient pupil size mainly fall into two categories: manual visual observation and detection using simple pupil size testing equipment.

[0003] Traditional testing relies entirely on medical staff's direct visual observation of the pupils, allowing only rough estimations of pupil size based on individual clinical experience. There is no standardized, objective, or quantitative basis for judgment. Differences in years of experience, visual perception, and clinical expertise among medical personnel can easily lead to vastly different assessments of the same patient's pupils. Subjective bias can interfere with a doctor's accurate judgment of the patient's condition, posing a risk of misdiagnosis.

[0004] Existing detection equipment is highly susceptible to the influence of ambient light intensity and the angle of incidence of the light source. In strong light, the light will form a bright reflective spot in the patient's pupil area, obscuring the pupil's edge outline; in low light, the overall image of the eye is blurry, and the boundary between the pupil and iris is not clearly distinguished. Whether by manual observation or imaging with ordinary detection equipment, the true boundary of the pupil cannot be accurately identified due to imaging defects, ultimately resulting in a large deviation in the measured pupil size and poor detection stability.

[0005] Conventional automated pupil detection devices on the market rely solely on image pixel ratios to calculate pupil size, lacking distance calibration components and thus unable to acquire the actual object distance between the device and the patient's eye in real time. Lens imaging suffers from perspective distortion, where objects appear larger when closer and smaller when farther away; relying solely on pixel calculations only yields relative dimensions, failing to output the true absolute diameter of the pupil. Furthermore, existing devices have poor adaptability. When patients exhibit special eye conditions such as orbital edema, eye trauma, or eyelid swelling, contour recognition is easily interfered with by extraneous eye tissue. They lack adaptability and adjustment capabilities for severe scenarios such as head injuries and poisoning, and their anti-interference ability is weak. Summary of the Invention

[0006] To address the aforementioned shortcomings in existing technologies, this invention provides an automated pupil size recognition and determination device that solves the problems of large subjective bias in manual observation, inaccurate pupil boundary recognition due to the susceptibility of ordinary detection equipment to ambient light interference, and the inability to output the true absolute diameter of the pupil due to the lack of real-time object distance calibration in existing pupil size detection methods.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows: an automated pupil size recognition and determination device, comprising a housing, a control unit disposed within the housing, and a power supply unit;

[0008] The housing includes a detection head and a gripping part connected to the detection head. The detection head contains an AI visual recognition module, an adaptive lighting compensation module, a distance calibration module, and a main control circuit board. A light-transmitting window is provided on the front end of the detection head.

[0009] The AI ​​visual recognition module includes a CMOS image sensor and an AI processing chip electrically connected to the CMOS image sensor. The CMOS image sensor includes a lens that faces the center area of ​​the infrared window to acquire pupil images. The AI ​​processing chip contains a pupil feature recognition model, which is used to automatically extract pupil contours and edge features and output quantified data.

[0010] The adaptive illumination compensation module includes a ring infrared light source substrate, several sets of infrared LED beads and a light sensor; the ring infrared light source substrate is arranged around the CMOS image sensor, several sets of infrared LED beads are evenly arranged on the ring infrared light source substrate and face the light-transmitting window, and the light sensor is set at the edge of the front end of the detection head and is electrically connected to the main control circuit board.

[0011] The distance calibration module includes a laser rangefinder sensor, which is located inside the detection head and its emitting end is parallel to the axis of the light-transmitting window. It is used to measure the absolute object distance from the device to the eye being measured in real time.

[0012] The AI ​​processing chip is electrically connected to the CMOS image sensor, the ring infrared light source substrate, the light sensor, and the laser rangefinder. The AI ​​processing chip calculates and outputs the absolute diameter of the pupil based on the absolute object distance and the pixel size of the pupil in the pupil image.

[0013] Furthermore, in the aforementioned automated pupil size recognition device, each group of infrared LED beads has a ball joint support at its bottom, which is fixed to the annular infrared light source substrate. The adaptive illumination compensation module also includes a micro-drive component, which includes a micro-stepping motor and a push rod corresponding to the number of infrared LED beads. The micro-stepping motor is fixed to the back of the annular infrared light source substrate, one end of the push rod is connected to the output shaft of the micro-stepping motor, and the other end passes through the annular infrared light source substrate and is hinged to the ball joint support. The micro-stepping motor is electrically connected to the AI ​​processing chip. The AI ​​processing chip controls the micro-stepping motor to move according to the ambient light intensity signal fed back by the light sensor, so as to adjust the illumination angle of each group of infrared LED beads.

[0014] Furthermore, the aforementioned automated pupil size recognition and determination device also has a flexible light-blocking goggle fitted on the front end of the detection head. The flexible light-blocking goggle is truncated conical in shape. The large-diameter end of the flexible light-blocking goggle is detachably snapped onto the outer peripheral wall of the detection head through a buckle structure, and the small-diameter end of the flexible light-blocking goggle is an arc-shaped edge that fits the contour of the human eye socket.

[0015] Furthermore, in the aforementioned automated pupil size recognition and determination device, the light-transmitting window is made of infrared-transmitting quartz glass, and the edge of the infrared-transmitting quartz glass is embedded in the detection head by a sealing ring.

[0016] Furthermore, the aforementioned automated pupil size recognition and determination device also includes a data intelligent management module, which includes a touch screen, a memory, and an alarm buzzer; the touch screen is embedded on the outer side of the grip, and the memory and alarm buzzer are both located inside the grip and electrically connected to the main control circuit board.

[0017] Furthermore, the aforementioned automated pupil size recognition and determination device also has a heat-conducting copper pipe and heat dissipation fins on the inner wall of the detection head. One end of the heat-conducting copper pipe is attached to the back of the AI ​​processing chip and the ring infrared light source substrate, and the other end extends to connect to the heat dissipation fins. The heat dissipation fins are exposed on the outer wall of the detection head to dissipate internal heat to the outside.

[0018] Furthermore, in the aforementioned automated pupil size recognition and determination device, the surface of the grip part is provided with an anti-slip texture, and the end face of the grip part away from the detection head is provided with a measurement trigger button that is electrically connected to the main control circuit board.

[0019] Furthermore, in the aforementioned automated pupil size recognition and determination device, the AI ​​processing chip calculates the absolute diameter of the pupil using the following formula: D = (d pixel ×L×s) / f; where D is the absolute diameter of the pupil; d pixel is the pupil pixel diameter extracted by the AI ​​visual recognition module; L is the absolute object distance measured by the laser rangefinder; s is the physical size of a single pixel in the CMOS image sensor, which is a factory-calibrated fixed parameter; f is the lens focal length of the CMOS image sensor.

[0020] Furthermore, in the aforementioned automated pupil size recognition device, a narrow-band infrared filter is attached to the front end of the lens of the CMOS image sensor. The narrow-band infrared filter only allows infrared light of a specific wavelength to pass through and blocks visible light to eliminate the interference of ambient stray light on pupil imaging; the specific wavelength matches the emission wavelength of the infrared LED beads.

[0021] Furthermore, in the aforementioned automated pupil size recognition and determination device, the AI ​​processing chip has a preset effective object distance threshold range. When the absolute object distance measured by the laser rangefinder exceeds the effective object distance threshold range, the AI ​​processing chip locks the calculation and outputs a distance abnormality prompt signal; when the absolute object distance is within the effective object distance threshold range, the AI ​​processing chip triggers the calculation and output of the absolute pupil diameter value.

[0022] The beneficial effects of this invention are as follows: Addressing the shortcomings of existing pupil detection equipment, such as large subjective errors from manual observation, interference from ambient light, distortion in size calculations due to lack of real-time object distance calibration, poor adaptability to critical ocular scenarios, poor heat dissipation, and lack of data management, this invention integrates an AI visual recognition module, an adaptive illumination compensation module, a laser ranging distance calibration module, an adjustable infrared illumination structure, a light-shielding adaptation structure, a heat dissipation structure, and a data intelligent management module working collaboratively. Through real-time dynamic object distance correction of imaging perspective deviation, adaptive multi-angle infrared illumination to eliminate ambient stray light interference, precise segmentation of the pupil edge using an AI model, and a matching light-shielding and heat dissipation structure to ensure stable long-term operation, this invention can automatically output the true absolute diameter value of the pupil, improving pupil detection accuracy, detection stability, and clinical scenario adaptability.

[0023] Unlike traditional devices that statically convert pixel size to fixed object distance, this device dynamically collects the actual object distance in real time and uses it for calculation. This eliminates dimensional deviations caused by perspective distortion ("near objects appear larger, far objects appear smaller"), significantly improving the accuracy of pupil diameter measurement and preventing numerical distortion caused by relative pixel size. An effective object distance judgment logic is added, locking the calculation and outputting a prompt when the object distance exceeds a reasonable range. This avoids invalid detection data generated by improper hand-held distances by medical staff, and avoids misleading judgments of illness from an algorithmic perspective, reducing the risk of misdiagnosis in severe cases such as traumatic brain injury and poisoning.

[0024] A light sensor collects ambient light intensity in real time, driving the components to automatically adjust the infrared illumination angle. In strong light, it prevents reflective spots from obscuring the pupil boundary, while in weak light, it provides a uniform infrared light source to ensure that the pupil and iris outlines remain clearly distinguishable. A narrow-band infrared filter blocks stray visible light, retaining only matching wavelength infrared imaging. This completely eliminates interference from stray light such as ward lights and natural light from outside windows, significantly reducing the probability of pupil outline segmentation failure in AI models. The ring-shaped infrared light source layout provides uniform light coverage of the eye area, eliminating blind spots and adapting to imaging asymmetrical eye conditions such as unilateral eyelid swelling and orbital edema.

[0025] The device replaces the traditional manual visual estimation method, eliminating the need for medical staff to rely on subjective judgment based on clinical experience. The device automatically completes the entire process of image acquisition, contour segmentation, and size conversion, significantly reducing the time required for single pupil detection and improving the efficiency of detecting batches of critically ill patients in the emergency department and ICU. The AI ​​model unifies the recognition standard, and different medical staff can output consistent quantitative diameter values, eliminating subjective judgment biases caused by differences in personnel experience and vision, and achieving standardized and objective pupil detection results, which facilitates cross-medical and cross-departmental comparison and evaluation of patients' conditions. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the internal structure of the device;

[0027] Figure 2 This is a schematic diagram of the structure of the device;

[0028] Figure 3 A schematic diagram of the isometric top structure of the AI ​​visual recognition module;

[0029] Figure 4 This is an isometric bottom structure diagram of the AI ​​visual recognition module;

[0030] Figure 5 This is a schematic diagram of the structure of an infrared LED bead and a micro-driving component.

[0031] The components include: 1. Housing, 2. Detection head, 3. Grip, 4. Light-transmitting window, 5. CMOS image sensor, 6. AI processing chip, 7. Lens, 8. Ring infrared light source substrate, 9. Infrared LED beads, 10. Light sensor, 11. Laser rangefinder, 12. Ball joint support, 13. Micro stepper motor, 14. Push rod, 15. Flexible light-shielding eye mask, 16. Buckle structure, 17. Touch screen, 18. Heat-conducting copper pipe, 19. Heat sink fins, 20. Anti-slip texture, and 21. Measurement trigger button. Detailed Implementation

[0032] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0033] like Figures 1-5As shown, this embodiment provides an automated pupil size recognition and determination device. The device includes a housing 1, a control unit, and a power supply unit. The housing 1 is divided into two parts: a detection head 2 and a gripping part 3, which are rigidly connected. The detection head 2 integrates four core functional components: an AI visual recognition module, an adaptive illumination compensation module, a distance calibration module, and a main control circuit board. A light-transmitting window 4 is opened on the front surface of the detection head 2. The gripping part 3 integrates a data intelligent management module, a whole-machine power supply circuit, and a trigger control structure. The detection head 2 is equipped with a heat-conducting copper pipe 18 and a heat dissipation fin 19 heat dissipation structure. All electrical components of the whole machine are uniformly connected to the main control circuit board and are uniformly powered by the power supply unit. The AI ​​processing chip 6 serves as the core computing unit, coordinating the entire process of image acquisition, illumination adjustment, distance measurement, pupil diameter conversion, anomaly judgment, data storage, and prompt output. It can automatically complete the entire process of human eye pupil image acquisition, contour feature extraction, ambient light adaptive supplementary lighting, real-time object distance calibration, absolute pupil diameter quantification calculation, abnormal status alarm, and detection data storage and display. It is suitable for standardized pupil detection scenarios in clinical critical care such as ICU, emergency room, neurology department, and poisoning emergency.

[0034] The power supply unit is integrated inside the housing 1 and the grip part 3. It includes a rechargeable lithium battery, a voltage regulator circuit, and an overcharge and over-discharge protection chip. It provides a stable DC operating voltage for all electrical components of the device. The power supply unit is electrically connected to the main control circuit board to realize the power supply on / off, power monitoring and low power reminder. The power supply unit provides power support for the continuous operation of the device and is a basic supporting structure of the device.

[0035] The housing 1 is divided into a detection head 2 and a gripping part 3, which are integrally formed and fixedly connected. The detection head 2 is located at the front end of the housing 1, and the gripping part 3 is the rear hand-held operation section.

[0036] The detection head 2 is a cylindrical hollow cavity with a light-transmitting window 4 installed in the center of the front face. The cavity is used to house the AI ​​visual recognition module, adaptive lighting compensation module, laser rangefinder 11, heat-conducting copper pipe 18, and heat dissipation fins 19. The outer wall of the detection head 2 is provided with an annular buckle groove for the detachable assembly of a flexible light-shielding eye mask 15. The outer wall of the detection head 2 has reserved slots for the heat dissipation fins 19 to be exposed, so as to conduct internal heat outward.

[0037] The grip part 3 is a columnar structure adapted for one-handed gripping. The outer surface is integrally molded with anti-slip texture 20 to increase the grip friction and prevent the device from slipping during medical staff operation. The measurement trigger button 21 is embedded in the rear end face of the grip part 3 away from the detection head 2. The measurement trigger button 21 is electrically connected to the main control circuit board. When the medical staff presses the measurement trigger button 21, the whole machine starts a pupil detection process. The grip part 3 has a touch screen display 17 embedded slot on the side. The internal cavity accommodates the memory, alarm buzzer, and power supply unit. All component circuits are uniformly connected to the main control circuit board.

[0038] The light-transmitting window 4 is made of infrared-transmitting quartz glass. The edge of the infrared-transmitting quartz glass is fitted with a sealing ring, and the entire seal is embedded in the window position on the front face of the detection head 2. The sealing ring fills the gap between the glass and the shell 1, preventing dust and moisture from the ward from entering the equipment. At the same time, it only allows infrared light to pass through, blocking ambient visible light from directly entering the cavity and reducing stray light interference.

[0039] The front end of the detection head 2 is detachably fitted with a flexible light-shielding goggle 15, which has a truncated conical structure. The inner wall of the large-diameter end is provided with a buckle structure 16, which matches the buckle groove on the outer peripheral wall of the detection head 2 to achieve quick assembly, disassembly and fixation of the goggle. The small-diameter end of the flexible light-shielding goggle 15 has an arc-shaped edge that perfectly matches the physiological curve of the human eye socket. During the test, it fits the skin around the patient's eyes, blocking natural light and artificial light from the ward environment from directly hitting the eye imaging area, further eliminating interference from external stray light on pupil image acquisition. The flexible light-shielding goggle 15 is made of medical-grade silicone, which is soft and suitable for patients with orbital edema, eyelid trauma, and facial swelling, without causing pressure or discomfort.

[0040] The AI ​​visual recognition module is fixedly installed at the center of the detection head cavity 2. It consists of two main components: a CMOS image sensor 5 and an AI processing chip 6. The CMOS image sensor 5 and the AI ​​processing chip 6 are electrically connected via a ribbon cable.

[0041] The CMOS image sensor 5 integrates an imaging chip and a lens 7. The lens 7 is axially aligned with the center area of ​​the light-transmitting window 4, and a narrow-band infrared filter is fixedly attached to the front end of the lens 7. The narrow-band infrared filter only allows specific wavelengths of infrared light to pass through unidirectionally, completely blocking visible light. This specific wavelength value is perfectly matched with the emission wavelength of the infrared LED bead 9, filtering out stray visible light such as ward lights and sunlight, and preventing stray light from forming reflective spots in the eye that obscure the edge of the pupil. The CMOS image sensor 5 is used to acquire raw images of the pupil of the subject's eyes in real time and transmit the image electrical signals to the AI ​​processing chip 6. The CMOS image sensor 5 has a built-in factory-calibrated fixed parameter s, where s is the physical size of a single pixel of the sensor. The value is stored in the register of the AI ​​processing chip 6, providing a fixed constant for pupil diameter conversion.

[0042] The AI ​​processing chip 6 is soldered and fixed to the main control circuit board. A dedicated pupil feature recognition model is burned into the chip's internal non-volatile storage medium. This model can automatically identify the iris and pupil boundary within an image, accurately extract the complete pupil outline and edge features, and output a standardized pupil pixel diameter d. pixelThe AI ​​processing chip 6 is electrically connected to the CMOS image sensor 5, the ring infrared light source substrate 8, the light sensor 10, the laser rangefinder 11, the touch screen 17, the memory, the alarm buzzer, and the measurement trigger button 21, and undertakes the core functions of data processing, logic control, and signal output of the whole machine.

[0043] AI processing chip 6 has a built-in fixed conversion formula: D=(d pixel ×L×s) / f;

[0044] Where D is the absolute diameter of the pupil (in mm); d pixel L is the pupil pixel diameter extracted by the AI ​​visual recognition module; L is the absolute object distance from the device to the measured eye measured in real time by the laser rangefinder 11; s is the physical size of a single pixel of the CMOS image sensor 5 (fixed parameter calibrated at the factory); f is the focal length of the lens 7 of the CMOS image sensor 5; all parameters are entered into the formula in real time for automatic calculation, and the final output is the true absolute diameter value of the pupil without perspective distortion.

[0045] The AI ​​processing chip 6 has an internally pre-programmed effective object distance threshold range, which is calibrated according to clinical handheld detection standards; the laser range sensor 11 continuously acquires the absolute object distance L in real time and transmits it to the AI ​​processing chip 6.

[0046] (1) When the measured absolute object distance L exceeds the preset effective object distance threshold range, the AI ​​processing chip 6 locks the diameter calculation program and does not perform the pupil diameter conversion calculation. At the same time, it sends a distance abnormality prompt signal to the main control circuit board. The main control circuit board drives the alarm buzzer to emit intermittent prompt sound. The touch screen 17 simultaneously pops up a text distance abnormality reminder to prompt medical staff to adjust the distance between the equipment and the eyes.

[0047] (2) When the absolute object distance L is within the preset effective object distance threshold range, the AI ​​processing chip 6 automatically starts the pupil diameter calculation formula calculation, and after the calculation is completed, outputs the absolute pupil diameter value to the touch screen 17 and transmits the detection data to the memory for storage.

[0048] The adaptive illumination compensation module is arranged around the CMOS image sensor 5 and includes four parts: a ring infrared light source substrate 8, multiple sets of infrared LED beads 9, illumination sensor 10, and micro drive components. All components are electrically connected to the main control circuit board, and the main control circuit board uniformly regulates the working status.

[0049] The ring-shaped infrared light source substrate 8 is a ring-shaped PCB board with a through hole in the center. The CMOS image sensor 5 and lens 7 are arranged in the center through the through hole to realize the ring-shaped layout of the light source surrounding the lens 7. Multiple sets of infrared LED beads 9 are evenly arranged on the front side of the substrate, and all micro stepper motors 13 are fixed on the back side of the substrate. A heat-conducting copper pipe 18 is attached to the back of the substrate, and the working heat is quickly discharged through the heat-conducting copper pipe 18.

[0050] Each set of infrared LED beads 9 is equipped with a ball joint bracket 12 at the bottom, which is rigidly fixed to the front of the annular infrared light source substrate 8. The ball joint bracket 12 has a omnidirectional rotational freedom, supporting the infrared LED beads 9 to deflect at multiple angles. The light-emitting surface of the beads is uniformly facing the light-transmitting window 4, providing a uniform infrared supplementary light source. Multiple sets of beads are evenly arranged in a ring, eliminating blind spots of light and dark on one side of the eye, and are suitable for imaging patients with eyelid swelling and unilateral orbital edema.

[0051] The light sensor 10 is embedded at the edge of the front end face of the detection head 2, without obstructing the imaging area of ​​the light-transmitting window 4; the light sensor 10 collects the simulated light intensity signal of the ward environment in real time, and transmits the light intensity value to the main control circuit board, providing the control basis for the main control circuit board to adjust the supplementary light angle and the brightness of the light source.

[0052] The micro-drive assembly includes micro-stepping motors 13 and push rods 14, each corresponding to a specific number of infrared LED beads 9. All micro-stepping motors 13 are fixed to the back of the annular infrared light source substrate 8, and the output shaft of each micro-stepping motor 13 is connected to a corresponding push rod 14. The push rod 14 vertically penetrates the annular infrared light source substrate 8, and its front end is hinged to the ball joint support 12 at the bottom of the corresponding infrared LED bead 9. After receiving the ambient light intensity signal from the light sensor 10, the main control circuit board outputs control pulses to drive the micro-stepping motors 13 to rotate forward / reverse. The stepping motors drive the push rods 14 to push and pull the ball joint support 12, simultaneously adjusting the illumination angle of one or more groups of infrared LED beads 9: In strong light conditions, the stepping motors drive the beads to deflect outwards, expanding the infrared light illumination range and reducing the local infrared light intensity in the eye, avoiding high-brightness reflections in the pupil area; in weak light conditions, the stepping motors drive the beads to converge inwards, concentrating the beam to illuminate the eye, supplementing the uniform infrared light source, and ensuring clear distinction between the pupil and iris boundaries.

[0053] The distance calibration module is independently composed of a laser rangefinder 11, which is fixed inside the cavity of the detection head 2. The optical path of the laser emitting end of the sensor is parallel to the central axis of the light-transmitting window 4. After the device starts detection, the laser rangefinder 11 continuously emits a range-measuring laser to collect the absolute object distance L from the device lens 7 to the eye of the person being measured in real time, and transmits the digital object distance signal to the AI ​​processing chip 6 in real time. The AI ​​processing chip 6 substitutes the real-time object distance L into the pupil diameter conversion formula, dynamically corrects the perspective distortion of the lens 7 (near is larger than far), eliminates the size error caused by fixed pixel conversion, and outputs the true physical diameter of the pupil.

[0054] The main control circuit board is horizontally fixed at the connection position between the rear end of the detection head 2 and the gripping part 3, serving as the electrical hub of the entire machine. The AI ​​processing chip 6 is soldered onto the surface of the main control circuit board. All modules and component circuits are uniformly connected to the main control circuit board to realize signal relay, power distribution, and command issuance. The main control circuit board receives signals collected by various sensors and issues drive commands to control the micro stepper motor 13, infrared LED beads 9, alarm buzzer, and touch display screen 17 to work together.

[0055] A heat-conducting copper pipe 18 and a heat dissipation fin 19 are arranged on the inner wall of the detection head 2. One end of the heat-conducting copper pipe 18 is tightly attached to the back of the AI ​​processing chip 6 and the back of the ring infrared light source substrate 8, completely covering the two major heat-generating core components and quickly absorbing the heat generated by the chip and the light source substrate. The other end of the heat-conducting copper pipe 18 extends to the outer wall of the detection head 2 and is tightly attached to the heat dissipation fin 19. The heat dissipation fin 19 is partially exposed on the outer wall of the detection head 2, exchanging the internal heat conducted by the copper pipe with the external air through convection, realizing active heat dissipation, avoiding chip processing lag and LED brightness decay caused by high internal temperature after long-term continuous detection, and ensuring stable operation of the equipment under continuous multi-case detection conditions in emergency and ICU.

[0056] The data intelligence management module is integrated and installed inside and outside the grip part 3, including a touch screen 17, a memory, and an alarm buzzer, all of which are electrically connected to the main control circuit board.

[0057] The touch screen 17 is embedded in the outer wall of the grip part 3 and has the functions of real-time image preview, numerical display and touch operation. It can display the real-time eye image captured by the CMOS image sensor 5 and intuitively display the absolute diameter value of the pupil after the test is completed. It can receive various prompt signals such as abnormal distance, low battery, and test completion, and display text prompts simultaneously. Medical staff can retrieve historical test records in the memory through the touch screen.

[0058] The memory has a built-in high-capacity storage chip that stores the pupil diameter value, detection time, and device serial number for each test. It also supports local storage of historical test data, making it convenient for medical staff to continuously track and compare pupil changes in critically ill patients.

[0059] The alarm buzzer is built into the cavity of the grip part 3. It receives abnormal signals from the main control circuit board and issues a graded buzzer prompt when the object distance exceeds the effective threshold, the device has low battery, or the imaging recognition fails. This promptly reminds medical staff of abnormal operation and avoids the generation of invalid detection data.

[0060] In use, medical staff hold the grip unit 3 with one hand. The curved edge of the flexible light-shielding eye mask 15 fits the patient's eye socket, blocking stray light from the external environment. Pressing the measurement trigger button 21 at the end of the grip unit 3 initiates a complete testing process: the power supply unit powers on the main control circuit board and all modules; the light sensor 10 immediately collects the current ambient light intensity and transmits it to the AI ​​processing chip; the AI ​​processing chip issues instructions to the main control circuit board to control the operation of the micro stepper motor 13; the push rod 14 drives each group of infrared LED beads 9 to adjust to the appropriate illumination angle; the ring infrared light source substrate 8 drives the infrared LED beads 9 to emit infrared supplementary light matching the wavelength of the narrow-band infrared filter; the laser range sensor 11 synchronously emits laser light to measure the absolute distance L between the device and the eye in real time and transmits it to the AI ​​processing chip 6; the AI ​​processing chip 6 reads the preset effective distance threshold and determines whether the current L is within the effective range.

[0061] (1) If L exceeds the threshold: AI processing chip 6 locks the calculation program, the main control circuit board drives the alarm buzzer to sound, the touch screen 17 pops up a distance abnormality prompt, the process is paused, and the measurement is retried after the medical staff adjusts the hand-held distance;

[0062] (2) If L is within the effective range: the CMOS image sensor 5 lens 7 collects infrared images of the eye through the narrow-band infrared filter and transmits the image signal to the AI ​​processing chip 6.

[0063] AI processing chip 6 calls upon its built-in pupil feature recognition model to automatically segment the pupil and iris boundaries and extract the pupil pixel diameter d. pixel The AI ​​processing chip 6 retrieves preset fixed parameters s (physical size of a single pixel) and f (focal length of the lens 7), and combines them with the real-time object distance L and pixel diameter d. pixel Substituting into the formula D=(d pixel The calculation is completed by (×L×s) / f, and the absolute pupil diameter D is output. After the calculation is completed, the AI ​​processing chip 6 transmits the pupil diameter value to the touch screen 17 for real-time display, and at the same time stores the detection time and pupil diameter data into the memory. If the imaging recognition fails, the alarm buzzer will sound an abnormal recognition prompt. The heat-conducting copper pipe 18 continuously absorbs the heat from the AI ​​processing chip 6 and the ring infrared light source substrate 8 throughout the detection process, and dissipates the heat outward through the exposed heat dissipation fins 19 to ensure the stability of continuous detection. After a single detection process is completed, the device stands by and waits for the next measurement to trigger the button 21 command.

[0064] This embodiment is equipped with a laser ranging distance calibration module, which collects the real object distance between the device and the eye in real time to participate in the pupil diameter conversion. It is equipped with a dedicated correction algorithm formula, which completely eliminates the perspective distortion caused by traditional devices that rely solely on pixel conversion. It outputs the true absolute diameter of the pupil, solving the problems of distortion caused by manual estimation and ordinary device size measurement, and greatly improving the detection quantification accuracy. At the same time, it sets an effective object distance threshold judgment logic. When the object distance is abnormal, the calculation is locked and an audio-visual prompt is given. From the algorithm level, it eliminates the possibility of incorrect detection data caused by improper hand-held distance, reducing the risk of misdiagnosis in the treatment of traumatic brain injury and severe poisoning.

[0065] The adaptive illumination compensation module is equipped with a triple anti-stray light structure consisting of an infrared LED bead 9 with an adjustable angle, a front-end narrow-band infrared filter, and an outer flexible light-shielding goggle 15. The light sensor 10 provides real-time feedback on ambient light intensity, and the micro stepper motor 13 automatically adjusts the infrared illumination angle. Strong light eliminates glare spots around the eyes, and weak light provides a uniform light source. The narrow-band infrared filter blocks all visible stray light, and the flexible light-shielding goggle 15 physically isolates external ambient light. The ring-shaped infrared light source eliminates blind spots in imaging, and even in special eye conditions such as orbital edema and eyelid swelling, it can still clearly distinguish the boundary between the pupil and the iris, significantly reducing the probability of AI model contour recognition failure and improving the imaging stability in complex clinical scenarios.

[0066] The entire machine is equipped with an AI visual recognition module to automatically complete the entire process of image acquisition, pupil contour extraction, and numerical conversion. It eliminates the need for medical staff to rely on personal experience and visual estimation, unifies AI recognition standards, eliminates subjective judgment bias among different medical staff, and achieves standardized and objective quantification of pupil detection. The automated operation of a single detection process shortens the detection time for batches of critically ill patients in the ICU and emergency departments, and improves the efficiency of clinical testing.

[0067] The detection head 2 is equipped with a heat-conducting copper pipe 18 and an exposed heat dissipation fin 19, which can quickly dissipate the working heat of the AI ​​processing chip 6 and the infrared light source substrate. The device can conduct continuous and uninterrupted detection for a long time without imaging lag or light source attenuation failure due to high temperature. The grip part 3 has an anti-slip texture 20 and a rear trigger button that are ergonomically designed for easy one-handed operation. The detachable flexible light-shielding eye mask 15 is suitable for patients with various eye injuries and diseases, and is easy to disassemble and clean. The intelligent data management module realizes real-time display of detection values, storage of historical data, and abnormal sound and light alarms, which facilitates medical staff to continuously track the dynamic changes of the patient's pupils and improve the clinical data retention capability.

[0068] The housing 1 is an integrated design consisting of the detection head 2 and the gripping part 3. The power supply unit and the main control circuit board coordinate all electrical components to work together, taking into account multiple needs such as detection accuracy, environmental anti-interference ability, long-term working stability, adaptability to multiple clinical scenarios, and standardized data management. It makes up for the technical shortcomings of existing pupil detection equipment, such as light interference, lack of object distance calibration, poor heat dissipation, weak adaptability, and lack of data storage management.

Claims

1. An automated pupil size recognition and determination device, characterized in that, Includes a housing (1), a control unit and a power supply unit disposed within the housing (1); The housing (1) includes a detection head (2) and a gripping part (3) connected to the detection head (2). The detection head (2) is equipped with an AI visual recognition module, an adaptive illumination compensation module, a distance calibration module and a main control circuit board. A light-transmitting window (4) is provided on the front end face of the detection head (2). The AI ​​visual recognition module includes a CMOS image sensor (5) and an AI processing chip (6) electrically connected to the CMOS image sensor (5). The CMOS image sensor (5) includes a lens (7), which faces the central area of ​​the infrared window and is used to acquire pupil images. The AI ​​processing chip (6) has a pupil feature recognition model burned into it, which is used to automatically extract pupil contours and edge features and output quantitative data. The adaptive illumination compensation module includes an annular infrared light source substrate (8), several sets of infrared LED beads (9) and an illumination sensor (10); the annular infrared light source substrate (8) is arranged around the CMOS image sensor (5), several sets of infrared LED beads (9) are evenly arranged on the annular infrared light source substrate (8) and facing the light-transmitting window (4), the illumination sensor (10) is arranged at the edge of the front end face of the detection head (2), and the illumination sensor (10) is electrically connected to the main control circuit board; The distance calibration module includes a laser rangefinder (11), which is located inside the detection head (2) and its emitting end is parallel to the axis of the light-transmitting window (4) for real-time measurement of the absolute distance between the device and the eye being measured. The AI ​​processing chip (6) is electrically connected to the CMOS image sensor (5), the ring infrared light source substrate (8), the light sensor (10) and the laser rangefinder (11), respectively. The AI ​​processing chip (6) calculates and outputs the absolute diameter value of the pupil based on the absolute object distance and the pixel size of the pupil on the pupil image.

2. The automated pupil size recognition and determination device according to claim 1, characterized in that, Each group of infrared LED beads (9) has a ball joint support (12) at its bottom end, which is fixed on the annular infrared light source substrate (8). The adaptive illumination compensation module also includes a micro drive component, which includes a micro stepper motor (13) and a push rod (14) corresponding to the number of infrared LED beads (9). The micro stepper motor (13) is fixed on the back of the annular infrared light source substrate (8). One end of the push rod (14) is connected to the output shaft of the micro stepper motor (13), and the other end passes through the annular infrared light source substrate (8) and is hinged to the ball joint support (12). The micro stepper motor (13) is electrically connected to the AI ​​processing chip (6). The AI ​​processing chip (6) controls the micro stepper motor (13) to move according to the ambient light intensity signal fed back by the light sensor (10) in order to adjust the illumination angle of each group of infrared LED beads (9).

3. The automated pupil size recognition and determination device according to claim 1, characterized in that, The front end of the detection head (2) is also fitted with a flexible light-shielding goggle (15), which is truncated conical in shape. The large-diameter end of the flexible light-shielding goggle (15) is detachably snapped onto the outer peripheral wall of the detection head (2) through a snap-fit ​​structure (16), and the small-diameter end of the flexible light-shielding goggle (15) is an arc-shaped edge that fits the contour of the human eye socket.

4. The automated pupil size recognition and determination device according to claim 1, characterized in that, The light-transmitting window (4) is made of infrared-transmitting quartz glass, and the edge of the infrared-transmitting quartz glass is embedded in the detection head (2) by a sealing ring.

5. The automated pupil size recognition and determination device according to claim 1, characterized in that, It also includes a data intelligence management module, which includes a touch screen (17), a memory and an alarm buzzer; the touch screen (17) is embedded on the outer side of the grip (3), and the memory and the alarm buzzer are both located inside the grip (3) and electrically connected to the main control circuit board.

6. The automated pupil size recognition and determination device according to claim 1, characterized in that, The inner wall of the detection head (2) is also provided with a heat-conducting copper pipe (18) and a heat dissipation fin (19). One end of the heat-conducting copper pipe (18) is attached to the back of the AI ​​processing chip (6) and the ring infrared light source substrate (8), and the other end extends to the heat dissipation fin (19). The heat dissipation fin (19) is exposed on the outer wall of the detection head (2) to dissipate internal heat to the outside.

7. The automated pupil size recognition and determination device according to claim 1, characterized in that, The surface of the grip (3) is provided with anti-slip texture (20), and the end face of the grip (3) away from the detection head (2) is provided with a measurement trigger button (21) that is electrically connected to the main control circuit board.

8. The automated pupil size recognition and determination device according to any one of claims 1 to 7, characterized in that, The formula used by the AI ​​processing chip (6) to calculate the absolute diameter of the pupil is: D=(d pixel ×L×s) / f; where D is the absolute diameter of the pupil; d pixel L is the pupil pixel diameter extracted by the AI ​​visual recognition module; L is the absolute object distance measured by the laser rangefinder (11); s is the physical size of a single pixel of the CMOS image sensor (5), which is a factory-calibrated fixed parameter; f is the focal length of the lens (7) of the CMOS image sensor (5).

9. The automated pupil size recognition and determination device according to claim 8, characterized in that, The front end of the lens (7) of the CMOS image sensor (5) is also attached with a narrow-band infrared filter. The narrow-band infrared filter only allows infrared light of a specific wavelength to pass through and blocks visible light to eliminate the interference of ambient stray light on pupil imaging. The specific wavelength is matched with the emission wavelength of the infrared LED beads (9).

10. The automated pupil size recognition and determination device according to claim 8, characterized in that, The AI ​​processing chip (6) has a preset effective object distance threshold range. When the absolute object distance measured by the laser rangefinder (11) exceeds the effective object distance threshold range, the AI ​​processing chip (6) locks the calculation and outputs a distance abnormality prompt signal. When the absolute object distance is within the effective object distance threshold range, the AI ​​processing chip (6) triggers the calculation and output of the absolute pupil diameter value.

Images