Intelligent medical gauze counting and storing system based on modular image recognition

Abstract

The invention discloses an intelligent medical gauze counting and storing system based on modular image recognition, and belongs to the technical field of medical instruments. The system comprises transparent box body units which can be spliced, a folding camera and a control module which are integrated in a fixed frame, and an image identification and AI analysis system which is deployed in the control module. The transparent box body units are freely combined into an array through the splicing structures of the side walls to form standardized units for independently storing single pieces of gauze; the folding camera can be unfolded and performs global image acquisition on the box body array; the AI analysis system processes the collected images in real time, automatically identifies the empty / full state of the gauze in each box body so as to accurately count, and analyzes the blood suction amount percentage of each piece of gauze based on the image color depth, so that the problems that traditional manual counting is low in efficiency and prone to making mistakes, and the bleeding state cannot be quantitatively evaluated are solved, and the working efficiency is improved. And the operation safety and the management efficiency are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of medical devices, and more specifically to an intelligent inventory and storage system for medical gauze based on modular image recognition. Background Technology

[0002] In medical settings such as surgery, trauma emergency care, and interventional treatments, medical gauze is a fundamental and high-consumption consumable. Accurate counting and real-time monitoring of its usage are crucial for preventing medical waste, assessing patient bleeding, and ensuring surgical safety. Traditional management models heavily rely on manual counting and recording by a circulating nurse and scrub nurse. This process is not only cumbersome and consumes valuable intraoperative time, but it is also prone to errors due to staff fatigue, distraction, or emergency situations. Furthermore, it completely fails to quantify the amount of blood absorbed by the used gauze.

[0003] To improve management efficiency, some electronic improvements have emerged in existing technologies, such as gauze counters using infrared sensors or simple weighing devices. However, these solutions still have significant limitations: First, they are mostly centralized counting designs, unable to manage each gauze in an independent, identifiable physical unit, resulting in the inability to achieve closed-loop tracking and status binding throughout the entire process, and the risk of confusion still exists during postoperative verification. Second, their design is fixed, their capacity is variable, and they lack modular expansion capabilities, making it difficult to flexibly adapt to different scales and consumable needs, from minor outpatient surgeries to major surgical procedures. Third, their human-computer interaction interfaces mostly rely on traditional capacitive touchscreens, which are extremely inconvenient or even malfunction when medical staff are wearing sterile gloves (especially when the gloves are stained with saline or blood), affecting the reliability and efficiency of use. Crucially, current technologies for detecting the blood absorption volume of gauze, an important clinical indicator, largely rely on postoperative centralized weighing (which cannot distinguish individual gauze pieces) or complex chemical reagent testing. These processes are time-consuming, pose a risk of contamination, and cannot provide real-time intraoperative feedback, failing to meet the real-time data requirements of modern precision surgery. Therefore, there is an urgent clinical need for an intelligent medical gauze counting and storage system based on modular image recognition to solve these problems. Summary of the Invention

[0004] The purpose of this invention is to solve the problems in the background art and provide a smart counting and storage system for medical gauze based on modular image recognition.

[0005] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A modular image recognition-based intelligent inventory and storage system for medical gauze includes a modular transparent box unit, a foldable camera, a control module, an image recognition and AI algorithm-based inventory and blood absorption analysis system, and a fixed frame. The fixed frame is detachably installed on the outer edge of the array formed by splicing the transparent box units. The foldable camera and control module are located inside the fixed frame. The image recognition and AI algorithm-based inventory and blood absorption analysis system is deployed in the control module, receives images captured by the foldable camera, identifies and analyzes the state of the gauze in each box unit, and realizes the counting of gauze quantity and the assessment of blood absorption.

[0006] This invention introduces modular, transparent box units to create an independent, identifiable physical storage space for each gauze pad. In conjunction with a foldable camera and control module integrated into a fixed frame, it achieves global, non-contact image acquisition of the box array. Finally, through a built-in image recognition and AI algorithm-based analysis system, the acquired images are processed in real time. This not only automatically completes the counting and status tracking of gauze pads down to the individual pad, completely avoiding the risks of omissions and confusion caused by manual verification, but also innovatively achieves rapid and quantitative analysis of the blood absorption of each gauze pad after use. This fundamentally solves the core problems of low efficiency, lack of traceability, and lack of real-time intraoperative status feedback in traditional gauze management models, elevating gauze management from passive manual verification to proactive intelligent monitoring and early warning, significantly improving surgical safety and the accuracy of postoperative risk assessment.

[0007] Preferably, the folding camera includes a folding column, a high-definition camera, and an adjustable-angle fill light. The bottom end of the folding column is located within a fixed frame, and the high-definition camera is located at the top end of the folding column. The adjustable-angle fill light consists of at least two independent light sources, symmetrically arranged on both sides of the high-definition camera.

[0008] This invention achieves the storage and unfolding of the camera through its folding column. By placing the high-definition camera at the top of the column, a top-down view covering the box array is obtained. At least two independent light sources are symmetrically arranged on both sides of the high-definition camera as adjustable-angle supplementary lights, so that the supplementary lighting angle can be specifically adjusted according to the layout of the box array and the camera height. This provides uniform, sufficient and shadow-free illumination inside each transparent box in the complex lighting environment of the operating room, ultimately solving the problem of unstable image quality caused by uneven ambient light or reflection from transparent materials. This ensures the accuracy and reliability of subsequent image recognition and AI algorithms for gauze status judgment and blood absorption analysis.

[0009] The control module is a flat-panel controller with a built-in battery and a buckle on the bottom that matches the box body; The user interface integrates a touchscreen, physical power buttons, and navigation keys; Foldable column mechanism: When stored, it is flush with the controller.

[0010] Software features: Image recognition counting: The gauze inside the box is photographed by a camera, and the AI ​​algorithm identifies the empty / full status and counts the items; Bloodsucking analysis: Calculates the percentage of bloodsucking based on image color depth and a preset AI model; Data Management: Displays inventory results and blood loss alarms, and supports Bluetooth / Wi-Fi transmission to the hospital information system.

[0011] Preferably, the folding column includes a column body, a column connecting cylinder, a camera connecting cylinder, a camera rotation positioner, and a column rotation positioner. The column connecting cylinder is fixed to the side wall of the fixed frame, and the camera connecting cylinder is located at the top of the column body. The column connecting cylinder and the column body are connected by the column rotation positioner. The camera connecting cylinder and the high-definition camera are connected by the camera rotation positioner, which includes a stator housing, a rotating shaft, rotor blades, a fixed gear, and a pressing locking element. The stator housing is fixed to the inner side wall of the column connecting cylinder, the fixed gear is located on the back of the stator housing, and the pressing locking element is located on the fixed gear. The rotating shaft is rotatably located in the middle of the stator housing and extends to the stator housing. The fixed gear is fixedly connected to the rotating shaft. The rotor blades are located on the side wall of the rotating shaft. A viscous fluid, which can be silicone, is filled between the rotor blades and the stator housing. The structure of the camera rotation positioner is the same as that of the column rotation positioner.

[0012] This invention fills the space between the stator housing and the rotor blades on the rotating shaft with a viscous fluid, utilizing fluid resistance to provide smooth damping and stepless hovering capability for the rotation of the column. At the same time, a pressing locking device on the fixed gear achieves mechanical locking of the rotation angle. This design allows the column to be smoothly adjusted to any angle and held stably during operation, and to be firmly locked after the position is determined. Ultimately, it solves the problems of shaking, sliding, or unstable positioning that may occur during manual adjustment of folding cameras, ensuring the absolute stability and anti-interference capability of the image acquisition device in working condition.

[0013] This invention integrates a damping unit filled with viscous fluid with a fixed gear and a pressing locking element, enabling the positioner to achieve smooth, stepless, arbitrary-angle hovering during rotational adjustment using fluid damping. This overcomes the shortcomings of pure gear mechanisms, such as stiff adjustment, stepped adjustments, and the inability to fine-tune. Once the angle is determined, the pressing locking element provides a rigid mechanical lock using gear meshing, completely avoiding the slow displacement or drift that may occur when a pure damper structure is subjected to force. Ultimately, this invention combines the flexibility of stepless adjustment with the absolute stability of mechanical locking in a single, compact structure, resolving the contradiction between smooth adjustment and reliable locking inherent in traditional single-mechanism systems.

[0014] Preferably, a Hall sensor is provided on the inner wall of the column connecting cylinder, and a magnet that cooperates with the Hall sensor is provided on the fixed gear. The adjustable angle supplementary light includes a support plate, a rotary motor, a support plate, and a lamp plate. Two support plates are fixed on the folding column, and the lamp plate is located between the support plates. One end of the lamp plate is rotatably connected to the support plate through a bearing, and the other end is fixedly connected to the output end of the rotary motor. The motor end of the rotary motor is fixed on the folding column.

[0015] This invention constructs a high-precision angle sensing unit by setting a Hall sensor on the inner wall of the column connecting cylinder and correspondingly setting a matching magnet on the fixed gear. This unit acquires the angular position of the rotating shaft in real time. The angle data is fed back to the control system and compared with the preset lighting model. The system then dynamically adjusts the illumination angle and brightness of the adjustable angle supplementary light, ultimately achieving adaptive optimization of the supplementary lighting effect based on the actual pitch angle of the camera. This solves the problem of local overexposure or insufficient lighting caused by changes in the shooting angle, ensuring that uniformly illuminated and clearly detailed recognition images can be obtained under different working postures.

[0016] Preferably, the rotor blade includes a main blade and an auxiliary blade extending radially. A first radial gap is formed between the tip of the main blade and the inner wall of the stator housing, and a second radial gap is formed between the tip of the auxiliary blade and the inner wall of the stator housing. The first radial gap is smaller than the second radial gap, and the main blade is asymmetrically arranged. The included angle between the main blades is between 60° and 80°.

[0017] This invention employs an asymmetrical double-blade design. By setting unequal first and second radial gaps between the tips of the main blade and the inner wall of the stator housing, and arranging the two blades asymmetrically at a specific included angle of 60°-80°, the damping force generated by the fluid flowing through the different gaps when the rotating shaft rotates in both directions is different. This asymmetrical damping characteristic makes the column smoother and less strenuous to operate when unfolding, while providing a stronger sense of resistance and positioning when folding and storing. Ultimately, a bidirectional differentiated operating feel is achieved within a single damping unit, which optimizes the smoothness of commonly used unfolding actions and enhances the stability and anti-accidental movement during the storage process.

[0018] Preferably, the transparent box unit includes a box with a top opening, a sealing cap connected to the top of the box, and a splicing structure on at least one side wall of the box for detachable mechanical splicing with at least one other identical transparent box unit.

[0019] This invention achieves airtight storage and retrieval of gauze through the cooperation of its top-opening box and sealing cover. At the same time, by setting a splicing structure on the side wall of the box, multiple identical units can be quickly and stably combined in a horizontal modular manner. This design not only provides each piece of gauze with an independent, pollution-proof storage space, but more importantly, it enables the storage capacity to be freely expanded and flexibly configured according to surgical needs. Ultimately, while ensuring precise management at the individual item level, it completely solves the problem that traditional fixed storage devices cannot adapt to different surgical scales and cannot be assembled on demand.

[0020] Preferably, the splicing structure includes convex strips formed on the two side walls of the box body in a first direction, and concave grooves formed on the two side walls of the box body in a second direction perpendicular to the first direction. The outline of the convex strips matches the outline of the concave grooves, so that multiple box body units can be spliced ​​together in an array by the lateral interlocking of the convex strips and concave grooves.

[0021] This invention achieves array-style splicing of box units simply by lateral interlocking by setting convex locking strips on the sidewall in the first direction and concave locking grooves on the sidewall in the second direction perpendicular to it, with their contours adapting to each other. This bidirectional complementary interlocking design enables the boxes to achieve precise positioning and firm connection in both the horizontal and vertical directions, allowing for the rapid assembly of rectangular arrays of different sizes without additional tools or connectors. Ultimately, while ensuring splicing stability and alignment accuracy, it greatly simplifies the assembly process of storage units in preoperative preparation and makes the overall structure compact and regular, perfectly adapting to the placement requirements of limited space at the edge of the operating table.

[0022] Preferably, the convex retaining strip has a T-shaped groove along its length. A pressing and fixing component is provided on the side wall of the fixing frame opposite to the T-shaped groove. The pressing and fixing component includes a base, a pressing rod, and an adjusting screw. The base is fixed to the side wall of the fixing frame and has a receiving cavity along its vertical direction. The adjusting screw passes through the receiving cavity and forms a threaded engagement with the base. A partially penetrated guide groove is provided on the side wall of the adjusting screw. An inclined guide rail is fixedly installed in the guide groove. The pressing rod slides through the receiving cavity of the base and is arranged perpendicularly to the adjusting screw in space, with the pressing rod located in front of the adjusting screw. A guide notch is provided on the pressing rod, and the guide notch matches the shape of the inclined guide rail. When the adjusting screw is rotated to move left and right, the interaction between the inclined guide rail and the guide notch drives the clamping rod to move up and down in the vertical direction. The bottom of the clamping rod is provided with a retaining plate, which is slidably disposed in the T-shaped groove. The two ends of the adjusting screw are provided with limiting protrusions to prevent the clamping rod from disengaging from the left and right ends of the inclined guide rail during the movement.

[0023] This invention utilizes the naturally formed array of convex locking strips during the assembly of the box units to combine T-shaped sliding grooves into a continuous guide track. By rotating the adjusting screw on the side wall of the fixed frame, the internal inclined mechanism is driven to convert the horizontal rotation into the vertical movement of the clamping rod, allowing the bottom locking plate to slide in along the combined track and finally lock in place. This design enables the control module and the box array to form a rigid connection with both lateral positioning and longitudinal clamping through a single operation, and can be quickly separated by rotating in the opposite direction during disassembly. Ultimately, it achieves a reliable plug-and-play connection and aseptic separation between the control module and the disposable consumable storage unit, which not only meets the requirements for stable operation during surgery but also adapts to the postoperative classification and processing procedures.

[0024] Preferably, the box body and the sealing cover are connected and closed. The box body is provided with a locking connector and a miniature magnetic induction switch. The locking connector includes a shell, a compression spring, a claw, a connecting rod, and a compression spring. The shell is disposed on the box body. The claw is slidably embedded in the receiving cavity of the shell. The upper end of the claw is provided with an elastic hook-shaped part. One end of the connecting rod is rotatably disposed on the claw, and the other end is a free end. The shell is provided with a guide groove that cooperates with the connecting rod. The inner wall of the guide groove is formed with a guide block. The claw is provided with a cavity. The compression spring is disposed in the cavity and its top is fixed to the bottom of the shell. The sealing cover is provided with a locking block that cooperates with the elastic hook-shaped part.

[0025] When the sealing cover is closed, the locking block pushes the elastic hook-shaped part of the claw to compress the spring and slide along the inside of the outer shell until the locking block passes the hook-shaped part. Then, the spring rebounds and drives the claw to reset, so that the hook-shaped part and the locking block automatically engage. This design utilizes the preload of the spring and the sliding cooperation of the connecting rod in the guide groove to achieve automatic locking when the sealing cover is closed and stable unlocking when it is opened. Its structure is compact and its operation is smooth. Ultimately, while ensuring the sealing of the box, it provides a reliable opening and closing experience that can be easily completed with one hand even when wearing gloves, perfectly meeting the efficient and sterile operation requirements of the operating room.

[0026] In summary, the beneficial effects of this invention are as follows: 1. This invention introduces a modular transparent box unit to create an independent and identifiable physical storage space for each gauze pad. In collaboration with a folding camera and control module integrated into a fixed frame, it achieves global non-contact image acquisition of the box array. Finally, through a built-in image recognition and AI algorithm-based analysis system, the acquired images are processed in real time. This not only automatically completes the counting and status tracking of gauze pads down to the individual pad, completely avoiding the risks of omissions and confusion caused by manual verification, but also innovatively achieves rapid and quantitative analysis of the blood absorption of each gauze pad after use. This fundamentally solves the core problems of low efficiency, lack of traceability, and lack of real-time intraoperative status feedback in traditional gauze management, transforming gauze management from passive manual verification to proactive intelligent monitoring and early warning, significantly improving surgical safety and the accuracy of postoperative risk assessment. 2. This invention achieves the storage and unfolding of the camera through its folding column. By placing the high-definition camera at the top of the column, a top-down view covering the box array is obtained. At least two independent light sources are symmetrically arranged on both sides of the high-definition camera as adjustable angle supplementary lights, so that the supplementary light angle can be adjusted according to the layout of the box array and the camera height. This provides uniform, sufficient and shadow-free lighting for the interior of each transparent box in the complex lighting environment of the operating room. Ultimately, it solves the problem of unstable image quality caused by uneven ambient light or reflection of transparent materials, ensuring the accuracy and reliability of subsequent image recognition and AI algorithm for gauze status judgment and blood absorption analysis. 3. This invention utilizes the naturally formed array of convex locking strips during the assembly of the box units to combine T-shaped sliding grooves into a continuous guide track. By rotating the adjusting screw on the side wall of the fixed frame, the internal inclined mechanism is driven to convert the horizontal rotation into the vertical movement of the clamping rod, allowing the bottom locking plate of the rod to slide in along the combined track and finally lock in place. This design allows the control module and the box array to form a rigid connection with both lateral positioning and longitudinal clamping through a single operation. Moreover, it can be quickly separated by rotating in the opposite direction during disassembly. Ultimately, it achieves a reliable plug-and-play connection and aseptic separation between the control module and the disposable consumable storage unit, which not only meets the requirements for stable operation during surgery but also adapts to the postoperative classification and processing procedures. 4. This invention fills the space between the stator housing and the rotor blades on the rotating shaft with a viscous fluid, utilizing fluid resistance to provide smooth damping and stepless hovering capability for the rotation of the column. At the same time, a pressing locking device on the fixed gear achieves mechanical locking of the rotation angle. This design allows the column to be smoothly adjusted to any angle and held stably during operation, and to be firmly locked after the position is determined. Ultimately, it solves the problems of shaking, sliding, or unstable positioning that may occur during manual adjustment of folding cameras, ensuring the absolute stability and anti-interference capability of the image acquisition device in working condition. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the overall machine of the present invention; Figure 2 This is an overall schematic diagram of the foldable camera of the present invention; Figure 3 This is a schematic diagram of the folding column of the present invention; Figure 4 This is a cross-sectional schematic diagram of the column rotation positioner of the present invention; Figure 5 This is a cross-sectional schematic diagram of the pressing and locking component of the present invention; Figure 6 This is a schematic diagram of the present invention, showing the removal of the column connecting cylinder through column rotation positioning; Figure 7 This is a schematic diagram of the transparent box unit of the present invention; Figure 8 This is a schematic diagram of the transparent box unit of the present invention after it has been opened; Figure 9 This is a cross-sectional schematic diagram of the pressing and fixing component of the present invention; Figure 10 This is a cross-sectional schematic diagram of the pressing and fixing component of the present invention being snapped into the T-shaped sliding groove; Figure 11 This is an overall schematic diagram of the locking connector of the present invention; Figure 12 This is a schematic diagram of the locking connector of the present invention after it has been pressed down; Figure 13 This is a schematic diagram of the chuck claw of the present invention; Figure 14 This is a schematic diagram of the compression spring of the present invention inside the chuck; Figure 15 This is a schematic diagram of the back of the pressing and fixing component of the present invention; Figure 16 This is a schematic diagram of the interior of the outer casing of the present invention; 1. Interlocking transparent box unit; 2. Foldable camera; 3. Control module; 4. Counting and blood absorption analysis system based on image recognition and AI algorithm; 5. Fixing frame; 21. Folding column; 21. High-definition camera; 22. Adjustable angle supplement light; 211. Column body; 212. Column connecting tube; 213. Camera connecting tube; 23. Camera rotation positioner; 24. Column rotation positioner; 241. Stator housing; 242. Rotating shaft; 243. Rotor blade; 244. Fixed gear; 245. Press locking element; 210. Hall sensor; 220. Magnet; 2431. Main blade; 2434. Secondary blade; 2435. First radial gap; 2436. Second radial gap; 11. Box with top opening; 12. Sealing cover; 13. Interlocking structure; 131. Convex locking strip; 132. Concave locking groove; 1311. T 5. Type-shaped slide; 5. Pressing fastener; 51. Base; 52. Pressing rod; 53. Adjusting screw; 511. Receiving cavity; 531. Guide groove; 532. Inclined guide rail; Guide notch; 54. Clamping plate; 533. Limiting protrusion; 6. Locking connector; 60. Miniature magnetic induction switch; 61. Housing; 62. Compression spring; 63. Claw; 64. Connecting rod; 65. Compression spring; 631. Elastic hook-shaped part; 611. Guide groove; 612. Guide block; 632. Cavity; 121. Clamping block; 221. Support plate; 222. Rotary motor; 224. Light panel. Detailed Implementation

[0028] The following specific embodiments are merely illustrative of the present invention and are not intended to limit the invention. After reading this specification, those skilled in the art can make modifications to these embodiments without contributing any inventive step, but such modifications are protected by patent law as long as they fall within the scope of the claims of the present invention.

[0029] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] Example 1

[0031] like Figure 1As shown, a modular image recognition-based intelligent counting and storage system for medical gauze includes a modular transparent box unit 1, a foldable camera 2, a control module 3, a counting and blood absorption analysis system based on image recognition and AI algorithms, and a fixed frame 4. The fixed frame 4 is detachably installed on the outer edge of the array formed by splicing the transparent box units 1. The foldable camera 2 and the control module 3 are located inside the fixed frame 4. The counting and blood absorption analysis system based on image recognition and AI algorithms is deployed in the control module 3, which receives images collected by the foldable camera 2, identifies and analyzes the state of the gauze in each box unit, and realizes the counting of gauze quantity and the assessment of blood absorption.

[0032] Transparent plastic box (medical grade), with a sealing cap on top (buckle / flip-top / rotating / pull-out type), and the interior can independently hold a single piece of gauze; The sides of the box are equipped with standardized convex and concave grooves (similar to Lego structure), which support the horizontal / vertical splicing of multiple boxes to form a multi-compartment integrated box; Software features: Image recognition counting: The gauze inside the box is photographed by a camera, and the AI ​​algorithm identifies the empty / full status and counts the items; Blood absorption analysis: Based on image color depth and a preset AI model (the training dataset contains RGB data of gauze with different blood absorption amounts), the percentage of blood absorption is calculated. Data Management: Displays inventory results and blood loss alarms (threshold adjustable), and supports Bluetooth / Wi-Fi transmission to the hospital information system.

[0033] Modular expansion: Storage units can be freely combined through standardized card slots to adapt to different surgical scales; Non-contact monitoring: Combining foldable cameras with AI algorithms to avoid contamination and improve detection efficiency; Human-computer interaction optimization: The physical buttons and touch screen are designed in a coordinated manner to adapt to the special operating environment of the operating room; Dual-effect data tracking: Simultaneously realizes quantity counting and blood suction analysis, improving the accuracy of postoperative risk assessment.

[0034] like Figure 2-6As shown, the folding camera 2 includes a folding column 21, a high-definition camera 25, and an adjustable-angle fill light 22. The bottom end of the folding column 21 is located within the fixed frame 4, and the high-definition camera 25 is located at the top of the folding column 21. The adjustable-angle fill light 22 consists of at least two independent light sources, symmetrically arranged on both sides of the high-definition camera 25. The folding column 21 includes a column body 211, a column connecting tube 212, a camera connecting tube 213, a camera rotation positioner 23, and a column rotation positioner 24. The column connecting tube 212 is fixed to the fixed frame 4. The camera connecting tube 213 is located on the top of the column body 211. The column connecting tube 212 and the column body 211 are connected by a column rotation positioner 24. The camera connecting tube 213 and the high-definition camera 25 are connected by a camera rotation positioner 23. The column rotation positioner 24 includes a stator housing 241, a rotating shaft 242, a rotor blade 243, a fixed gear 244, and a pressing locking member 245. The stator housing 241 is fixed on the inner side wall of the column connecting tube 212, and the fixed gear 244 is located on the outer side wall of the stator. On the back of the housing 241, a pressing locking member 245 is provided on the fixed gear 244. A rotating shaft 242 is rotatably located in the middle of the stator housing 241 and extends to the stator housing 241. The fixed gear 244 is fixedly connected to the rotating shaft 242. Rotor blades 243 are provided on the side wall of the rotating shaft 242. A viscous fluid is filled between the rotor blades 243 and the stator housing 241. A Hall sensor 210 is provided on the inner side wall of the column connecting cylinder 212. A magnet 220 that cooperates with the Hall sensor 210 is provided on the fixed gear 244. The blade 243 includes a main blade 2431 and a secondary blade 2434 extending radially. A first radial gap 2435 is formed between the tip of the main blade 2431 and the inner wall of the stator housing 241, and a second radial gap 2436 is formed between the tip of the secondary blade 2434 and the inner wall of the stator housing 241. The first radial gap 2435 is smaller than the second radial gap 2436, and the main blade 2431 and the secondary blade 2434 are arranged asymmetrically. The included angle between the main blade 2431 and the secondary blade 2434 is between 60° and 80°.

[0035] like Figure 5As shown, the pressing locking component 245 includes a support base 2451, a return spring 2452, a movable slide 2453, an operation button 2454, a linkage arm 2455, and a locking pin 2456. The support base 2451 is fixed to the side wall of the column connecting cylinder 212. A receiving cavity 2457 is provided on the support base 2451. The movable slide 2453 is disposed in the receiving cavity 2457. Two return springs 2452 are disposed at the bottom of the movable slide 2453. One end of the return spring 2452 is fixedly connected to the movable slide 2453, and the other end is fixedly connected to the bottom of the receiving cavity 2457. A heart-shaped automatic guide groove 2458 is provided on the side wall of the movable slide 2453. One end of the linkage arm 2455 is disposed in the heart-shaped automatic guide groove, and the other end is rotatably connected to the support base 2451. The operation button 2454 is disposed in the movable slide 2455. At the top of the slide body 2453, a locking pin 2456 is located at the bottom of the movable slide body. A limit block 2459 is provided at the bottom of the locking pin 2456. By pressing the operation button 2454, the movable slide body 2453 is driven to move downward, causing the return spring 2452 to retract. At the same time, the linkage arm 2455 slides to its top trajectory point in the heart-shaped automatic guide groove 2458, thereby locking the fixed gear with the bottom locking pin 2456. This prevents the fixed gear from moving unexpectedly without external force, which would affect the surgical operation. When the telescopic function needs to be activated, the operation button 2454 is pressed again. The linkage arm 2455 slides to its bottom trajectory point in the heart-shaped automatic guide groove 2458. Under the push of the return spring 2452, the movable slide body 2453 returns to its original position, and the locking pin leaves the fixed gear, thereby releasing the lock on the fixed gear.

[0036] like Figures 7-10As shown, the transparent box unit 1 includes a box body 11 with a top opening. A sealing cap 12 is connected to the top of the box body 11. At least one side wall of the box body 11 is provided with a splicing structure 13 for detachable mechanical splicing with at least one other identical transparent box unit 1. The splicing structure 13 includes convex retaining strips 131 formed on the two side walls of the box body 11 in a first direction, and concave retaining grooves 132 formed on the two side walls of the box body 11 in a second direction perpendicular to the first direction. The contours of the convex retaining strips 131 and the concave retaining grooves 132 are adapted to each other, so that multiple box units 1 can be spliced ​​in an array by the lateral interlocking of the convex retaining strips 131 and the concave retaining grooves 132. The convex retaining strips 131 have T-shaped grooves 1311 along their length. A pressing and fixing component 5 is provided on the side wall opposite to the T-shaped slide groove 1311 of the fixed frame 4. The pressing and fixing component 5 includes a base 51, a pressing rod 52 and an adjusting screw 53. The base 51 is fixed to the side wall of the fixed frame 4. The base 51 has a receiving cavity 511 along the vertical direction. The adjusting screw 53 passes through the receiving cavity 511 and forms a threaded engagement with the base 51. The side wall of the adjusting screw 53 has a non-through guide groove 531. An inclined guide rail 532 is fixedly installed in the guide groove 531. The pressing rod 52 slides through the receiving cavity 511 of the base and is arranged perpendicularly to the adjusting screw 53 in space. The pressing rod 52 is located in front of the adjusting screw 53. The pressing rod 52 has a guide notch, which matches the shape of the inclined guide rail 532. When the adjusting screw is rotated to move it left and right, the interaction between the inclined guide rail and the guide notch drives the clamping rod to move up and down in the vertical direction. The bottom of the clamping rod 52 is provided with a retaining plate 54, which can be slidably disposed in the T-shaped slide groove 1311. The two ends of the adjusting screw 53 are provided with limiting protrusions 533 to prevent the clamping rod from disengaging from the left and right ends of the inclined guide rail during the movement.

[0037] like Figures 11-16As shown, the box body 11 and the sealing cover 12 are connected and closed. The box body 11 is provided with a locking connector 6, which includes a housing 61, a compression spring 62, a claw 63, a connecting rod 64, and a compression spring 65. The claw 63 is slidably embedded in the receiving cavity of the housing 61. The upper end of the claw 63 is provided with an elastic hook-shaped part 631. One end of the connecting rod 64 is rotatably disposed on the claw 63, and the other end is a free end. The housing 61 is provided with a guide groove 611 that cooperates with the connecting rod 64. The inner wall of the guide groove 611 is formed with a guide block 612. The claw 63 is provided with a cavity 632. The compression spring 65 is disposed in the cavity 632 and its top is fixed to the housing. The bottom of the shell 61 has a sealing cover 12 with a locking block 121 that cooperates with the elastic hook-shaped part 631. The claw 63 has a limiting protrusion 633. The side wall of the shell 61 has a limiting groove 634 that cooperates with the limiting protrusion 633. The adjustable angle supplementary light 22 includes a support plate 221, a rotary motor 222, a support plate 223, and a light plate 224. The two support plates 221 are fixed on the folding column 21. The light plate 224 is located between the support plates 223. One end of the light plate 224 is rotatably connected to the support plate 221 through a bearing, and the other end is fixedly connected to the output end of the rotary motor 222. The motor end of the rotary motor 222 is fixed on the folding column 21.

[0038] The counting and blood-sucking volume analysis system based on image recognition and AI algorithms is the core intelligent processing unit of this invention, deployed in the control module (3). This system achieves the following core functions through the collaborative work of software algorithms and hardware: Workflow and Functions of a Counting and Blood Suction Analysis System Based on Image Recognition and AI Algorithms Image acquisition and preprocessing Triggering mechanism: Receives status signal from miniature magnetic induction switch (60) on box unit (1) and automatically controls folding camera (2) to take pictures of box array.

[0039] Image optimization: At the moment of shooting, the system automatically adjusts the illumination angle and brightness of the adjustable angle fill light (22) based on the camera angle data fed back by the Hall sensor (210) to ensure that a clear in-box image with uniform illumination, no shadows, and low reflection is obtained.

[0040] Region segmentation: Using computer vision technology, based on the known physical layout of the box array, the acquired panoramic image is automatically segmented, and the independent sub-images inside each transparent box unit (1) are accurately located and extracted.

[0041] Gauze count (target detection and state classification) Core algorithm: Employs deep learning models (such as convolutional neural networks CNN).

[0042] Functionality implementation: State recognition: The model analyzes the sub-images of each box and classifies the state inside the box as "full" (with gauze) or "empty" (without gauze).

[0043] Automatic counting: The system counts the number of boxes that are "full", which is the current total number of gauze remaining. Every time the box lid is opened or closed, the system re-analyzes and updates the count, achieving real-time and automatic counting.

[0044] Change tracking: By comparing the analysis results before and after, the system can accurately record which specific location of the box has changed its status (taken or returned), realizing single-item-level traceability.

[0045] Analysis of blood absorption by gauze (image feature quantification and regression prediction) Core algorithm: Employs image processing and machine learning regression models.

[0046] Functionality implementation: Feature extraction: For boxes identified as "full" (especially those with gauze replaced after surgery), the system further extracts color features from sub-images. It focuses on blood-related channels in the RGB or HSV color space (such as R channel intensity, saturation S, etc.) and calculates their statistical characteristics (such as mean, variance, and histogram distribution).

[0047] Model calculation: The extracted color features are input into a pre-trained blood absorption analysis model. This model is trained based on a large number of gauze sample images with known blood absorption (calibrated by weighing) and their color feature data, and can establish a mapping relationship between image color depth and blood absorption weight (or volume).

[0048] Quantitative output: The model outputs the estimated percentage (or absolute amount) of blood absorption for this gauze. The system binds and stores this data with the unique identifier of the gauze (box location number).

[0049] Data fusion, decision-making, and output Threshold alarm: Users can preset a safe threshold for blood absorption (e.g., 30%). When the system analyzes that the blood absorption of a piece of gauze exceeds the threshold, a visual alarm will be immediately triggered on the interactive interface of the control module (3) (e.g., highlighting the corresponding box position box and flashing red) and an audio reminder will be triggered.

[0050] Data Management: The system continuously records all inventory events (time, box location, status changes) and all data absorption analysis results, generating a complete audit log.

[0051] Information transmission: The counting results, blood absorption data and alarm information are transmitted to the hospital information system (HIS) in real time via Bluetooth or Wi-Fi and integrated with the electronic medical record.

[0052] Working principle: such as Figures 1-16 As shown, during the preoperative preparation stage, medical staff take out the corresponding number of transparent box units 1 according to the estimated amount of gauze to be used in the surgery. These box units 1 are quickly spliced ​​together horizontally and vertically by the convex strips 131 and concave slots 132 that fit each other on their side walls. They can be assembled into a tight and regular rectangular storage array without any tools. Then, the pressing fasteners 5 on the side wall of the fixing frame 4 are aligned with the T-shaped slide groove 1311 track formed by the array edge combination. The plate 54 slides into the T-shaped slide groove 1311. The rotating adjusting screw 53 drives the internal inclined guide rail 532 to interact with the guide notch on the pressing rod 52, so that the plate 54 at the bottom of the pressing rod 52 presses against the inner top of the T-shaped slide groove 1311 to lock it. Thus, the fixing frame 4, which integrates the folding camera 2 and the control module 3, is firmly installed on one side of the box array, completing the physical assembly of the system. After the controller completes its self-test upon power-on, the camera is turned on, and the operator manually unfolds the folding column 21. By holding the column body 211 and swinging it outward, the column rotates around the column connecting cylinder 212. This rotation is controlled by the column rotation positioner 24. Inside the rotation positioner, the asymmetric rotor blades 243, fixed to the rotation shaft 242, rotate within the stator housing 241 filled with viscous fluid. The fluid resistance provides a smooth, stepless damping feel during the unfolding process, allowing the column to easily stop at any intermediate angle. When the column swings to a suitable tilt angle that roughly covers the box array, the operator presses the button set on the fixed gear 244. The locking component 245 engages with the fixed gear on the rotating shaft 242 to achieve mechanical locking of the horizontal rotation angle, preventing the column from shaking or returning to its original position during subsequent operations. Subsequently, the operator fine-tunes the pitch angle of the high-definition camera 25. The camera is connected to the top of the column body 211 through the camera connecting tube 213 and is adjusted by the camera rotation positioner 23 with the same structure. It is also manually adjusted and uses fluid damping to achieve stepless hovering. Finally, it is locked by pressing the locking component on it. Medical staff can judge whether the field of view completely covers the interior of all the spliced ​​transparent box units 1 by observing the camera. Once the camera angle is fixed, the system software prompts for initialization. After operator confirmation, control module 3 triggers the high-definition camera 25 to take its first shot. At this time, the adjustable-angle supplementary lights 22, symmetrically arranged on both sides of the camera, automatically illuminate. The light angle is determined based on the current camera tilt angle. This angle information can be indirectly calculated from the signal measured by the Hall sensor and magnet in the camera rotation positioner 23, or preliminarily determined by the image analysis algorithm for adaptive adjustment, to ensure that the light evenly illuminates the interior of each box, avoiding reflections or shadows. After acquiring this clear panoramic image of the array, the built-in image recognition algorithm immediately analyzes it: identifying each individual box area and determining whether gauze is present inside (status: "full"). The system records this status as the initial baseline, stores it in memory, and displays the corresponding total number of gauze on the human-machine interface, thus completing the pre-operation preparation and calibration. Once the surgical procedure begins, whenever medical staff need to retrieve gauze, they must open the corresponding box's sealed lid 12. The lid interacts with the elastic hook-shaped part 631 of the locking connector 6 on the box body 11 via its latch 121. When closed, it locks in place, and when opened, it can be easily unlocked by pressing. This allows for convenient one-handed operation even while wearing gloves. After the gauze is removed, the lid is closed. This physical change is captured by the system's miniature magnetic induction switch 60. The control module 3 then drives the folding camera 2 to quickly acquire images of the array. The built-in image recognition and AI algorithms immediately analyze the image, accurately identify empty box units 1, and update and display the remaining gauze quantity in real time on the interactive interface, completing an automatic count. This process requires no manual counting or input, and the count results can be synchronized to the hospital information system in real time via wireless network. During surgery or post-operative inventory, once a bloodstained gauze is placed back into its designated container and the lid is closed, the system triggers image acquisition again. The AI ​​algorithm not only confirms the container's status changes from "empty" to "full," completing the recycling registration, but more importantly, it quantifies the color depth of the gauze image. The algorithm calls a pre-trained blood absorption assessment model, comparing the gauze's current RGB characteristics with the model database to calculate the percentage of blood absorption. If the calculation result exceeds a preset safety threshold (e.g., 30%), the control module 3's screen immediately highlights (e.g., in red) the container location as a warning and issues an alert, while simultaneously marking and uploading the abnormal data. In this way, the system achieves intelligent management of the entire process—from precise quantity tracking and full traceability of individual items to real-time assessment of usage status (blood absorption)—without human intervention. Post-operatively, the control module 3 can be removed simply by rotating the adjusting screw 53 on the fixing frame 4 in the reverse direction. The used container array can be disposed of as medical waste or separated from the container and subjected to sterilization, preparing for the next surgery.

[0053] Example 2

[0054] Five storage boxes were selected and assembled, and the controller was installed before powering on. The folding column was unfolded to a 60° tilt angle, and the camera captured images of the box array. The algorithm identified box number 3 as empty (count -1), and the RGB values ​​of the gauze in box number 4 were calculated by the model to indicate a blood absorption rate of 45%. When this exceeded the threshold (30%), a red warning was displayed on the screen. This method is suitable for mass production, and the modular design reduces customization costs. It is applicable to medical institutions at all levels, reducing the rate of gauze residue accidents by over 70%, demonstrating significant clinical value.

Claims

1. A smart counting and storage system for medical gauze based on modular image recognition, characterized in that, The system includes a modular transparent box unit (1), a foldable camera (2), a control module (3), an image recognition and AI algorithm-based counting and blood absorption analysis system, and a fixed frame (4). The fixed frame (4) is detachably installed on the outer edge of the array formed by splicing the transparent box units (1). The foldable camera (2) and the control module (3) are located inside the fixed frame (4). The image recognition and AI algorithm-based counting and blood absorption analysis system is deployed in the control module (3), which receives images collected by the foldable camera (2), identifies and analyzes the gauze status in each box unit, and realizes the counting of gauze and the assessment of blood absorption.

2. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 1, characterized in that, The folding camera (2) includes a folding column (21), a high-definition camera (25), and an adjustable angle fill light (22). The bottom of the folding column (21) is located in a fixed frame (4), and the top of the folding column (21) is provided with the high-definition camera (25). The adjustable angle fill light (22) consists of at least two independent light sources, which are symmetrically arranged on both sides of the high-definition camera (25).

3. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 2, characterized in that, The folding column (21) includes a column body (211), a column connecting tube (212), a camera connecting tube (213), a camera rotation locator (23), and a column rotation locator (24). The column connecting tube (212) is fixed to the side wall of the fixed frame (4). The camera connecting tube (213) is located at the top of the column body (211). The column connecting tube (212) and the column body (211) are connected by the column rotation locator (24). The camera connecting tube (213) and the high-definition camera (25) are connected by the camera rotation locator (23).

4. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 3, characterized in that, The column rotation positioner (24) includes a stator housing (241), a rotating shaft (242), rotor blades (243), a fixed gear (244), and a pressing locking element (245). The stator housing (241) is fixed to the inner wall of the column connecting cylinder (212). The fixed gear (244) is located on the back of the stator housing (241). The pressing locking element (245) is located on the fixed gear (244). The rotating shaft (242) is rotatably mounted on the stator housing (241). The fixed gear (244) is fixedly connected to the rotating shaft (242) in the middle and extends to the stator housing (241). The rotor blade (243) is provided on the side wall of the rotating shaft (242). The rotor blade (243) and the stator housing (241) are filled with viscous fluid. The inner side wall of the column connecting cylinder (212) is provided with a Hall sensor (210). The fixed gear (244) is provided with a magnet (220) that cooperates with the Hall sensor (210).

5. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 4, characterized in that, The rotor blade (243) includes a main blade (2431) and a secondary blade (2434) extending radially. A first radial gap (2435) is formed between the top tip of the main blade (2431) and the inner wall of the stator housing (241). A second radial gap (2436) is formed between the top tip of the secondary blade (2434) and the inner wall of the stator housing (241). The first radial gap (2435) is smaller than the second radial gap (2436). The main blade (2431) and the secondary blade (2434) are arranged asymmetrically. The included angle between the main blade (2431) and the secondary blade (2434) is between 60° and 80°.

6. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 1, characterized in that, The transparent box unit (1) includes a box (11) with a top opening, a sealing cap (12) connected to the top of the box (11), and a splicing structure (13) provided on at least one side wall of the box (11) for detachable mechanical splicing with at least one other identical transparent box unit (1).

7. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 6, characterized in that, The splicing structure (13) includes a convex strip (131) formed on the two side walls of the box body (11) in the first direction, and a concave groove (132) formed on the two side walls of the box body (11) in the second direction perpendicular to the first direction. The outline of the convex strip (131) is adapted to the outline of the concave groove (132), so that multiple box units (1) can be spliced ​​in a horizontal interlocking array by the convex strip (131) and the concave groove (132).

8. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 7, characterized in that, The convex retaining strip (131) has a T-shaped groove (1311) along its length. A pressing and fixing member (5) is provided on the side wall of the fixing frame (4) opposite to the T-shaped groove (1311). The pressing and fixing member (5) includes a base (51), a pressing rod (52), and an adjusting screw (53). The base (51) is fixed to the side wall of the fixing frame (4). The base (51) has a receiving cavity (511) along its vertical direction. The adjusting screw (53) passes through the receiving cavity (511) and forms a threaded connection with the base (51). A non-through guide is provided on the side wall of the adjusting screw (53). The guide groove (531) is fixedly provided with a sloping guide rail (532). The clamping rod (52) slides through the receiving cavity (511) of the base and is arranged perpendicularly to the adjusting screw (53) in space. The clamping rod (52) is located in front of the adjusting screw (53). The clamping rod (52) is provided with a guide notch. The guide notch matches the shape of the sloping guide rail (532). The bottom of the clamping rod (52) is provided with a retaining plate (54). The retaining plate (54) can be slidably provided in the T-shaped slide groove (1311). The two ends of the adjusting screw (53) are provided with limiting protrusions (533).

9. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 7, characterized in that, The box body (11) and the sealing cover (12) are connected together. The box body (11) is provided with a locking connector (6) and a miniature magnetic induction switch (60). The locking connector (6) includes a shell (61), a compression spring (62), a claw (63), a connecting rod (64), and a compression spring (65). The shell (61) is located on the box body (11). The claw (63) is slidably embedded in the receiving cavity of the shell (61). The upper end of the claw (63) is provided with an elastic hook-shaped part (631). The connecting rod (64) One end of the connecting rod (64) is rotatably mounted on the claw (63), and the other end is a free end. The outer shell (61) is provided with a guide groove (611) that cooperates with the connecting rod (64). The inner wall of the guide groove (611) is formed with a guide block (612). The claw (63) is provided with a cavity (632). The compression spring (65) is located in the cavity (632) and its top is fixed to the bottom of the outer shell (61). The sealing cover (12) is provided with a locking block (121) that cooperates with the elastic hook-shaped part (631).

10. The intelligent medical gauze counting and storage system based on modular image recognition according to claim 7, characterized in that, The adjustable angle supplement light (22) includes a support plate (221), a rotary motor (222), a support plate (223), and a lamp plate (224). The two support plates (221) are fixed on the folding column (21), and the lamp plate (224) is located between the support plates (221). One end of the lamp plate (224) is rotatably connected to the support plate (221) through a bearing, and the other end is fixedly connected to the output end of the rotary motor (222). The motor end of the rotary motor (222) is fixed on the folding column (21).

Images