Medical instrument package inspection quality control system and method based on video monitoring and AI identification

Abstract

The invention relates to the technical field of medical instrument quality control, in particular to a medical instrument package inspection quality control system and method based on video monitoring and AI identification, and the technical scheme provided by the invention has the following beneficial effects: the whole operation process can be traced; a complete quality tracing chain is formed, whether the operation process is standard or not can be traced afterwards, and responsibility attribution is clear; based on a specially trained AI recognition algorithm, the placement condition of the indicator card is automatically verified, invalid sterilization caused by manual omission is avoided, and medical safety is guaranteed; by means of a real-time early warning and process blocking mechanism, problems are found and corrected in time in the inspection process, unqualified instrument sets are prevented from entering the subsequent sterilization process, and the reworking cost and time delay are reduced.

Description

Technical Field

[0001] This invention relates to the field of medical device quality control technology, specifically to a medical device package inspection quality control system and method based on video monitoring and AI recognition. Background Technology

[0002] The initial inspection (first inspection) of medical device packages before sterilization by the hospital's disinfection supply center is a key step in ensuring medical safety. Its core process includes counting the number of devices, checking the integrity of the devices, and placing chemical indicator cards and biological monitoring cards to verify the sterilization effect.

[0003] The current medical device pack inspection process mainly relies on manual operation. The specific steps are: the operator manually counts the instruments in the device pack, visually inspects the cleanliness and integrity of the instruments, manually places chemical indicator cards or biological monitoring cards, signs the paper record sheet for confirmation, and then sends the device pack to the sterilization area. The corresponding equipment mainly includes a workbench for placing the inspection device packs, lighting equipment to provide a light source, a paper record sheet for recording the inspection results, a barcode scanner to identify the device pack number, and, in some hospitals, a separate surveillance camera for security purposes.

[0004] However, existing technologies have the following significant drawbacks: Operational standardization cannot be traced: Reliance on manual operation and paper records, lack of real-time monitoring and effective recording of the inspection process, makes it impossible to trace and verify whether the operation is standardized or whether key steps are omitted afterward. Insufficient reliability of key indicator card placement: The placement of key indicators such as chemical indicator cards and biological monitoring cards relies on manual methods, which can easily lead to omissions, and there is a lack of automated verification methods; Quality control is delayed: existing methods often only discover problems such as missing indicator cards after sterilization, which leads to the need to reprocess the instrument packs, increasing time costs and wasting resources; Data fragmentation is a problem: Inspection records, operator information, instrument package information, and other data are stored in a scattered manner, making it difficult to form a complete quality traceability chain. Training and quality control are challenging: there is a lack of standardized operation video recordings, and new employee training and quality review lack intuitive reference materials.

[0005] The aforementioned deficiencies have a clear causal relationship: reliance on manual visual inspection and recording easily leads to subjective judgment discrepancies and recording errors, resulting in unstable inspection quality; the lack of process monitoring makes it impossible to verify whether operations are standardized, making it difficult to trace responsibility when problems occur; the manual placement of indicator cards carries the risk of forgetting them, potentially leading to ineffectively sterilized medical device packs entering the usage stage; and the lack of intelligent early warning systems means errors can only be detected in later stages, increasing rework costs and delaying use. Therefore, there is an urgent need for a quality control technology solution for medical device pack inspection that can address these issues. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a quality control method and system for medical device package inspection based on video monitoring and AI recognition, so as to establish a fully traceable video archive system for medical device package inspection; realize automated verification of indicator card placement through AI technology; build a real-time early warning mechanism to prevent unqualified medical device packages from entering the sterilization process; and realize digital integrated management of inspection data.

[0007] To achieve the above objectives, the present invention provides a medical device package inspection quality control system based on video surveillance and AI recognition, comprising a hardware device layer, a core function layer, an application service layer, and a user interface layer, with each layer working collaboratively. Hardware equipment layer: includes a high-definition camera, barcode scanner, workstation terminal, and card reader; the high-definition camera has a resolution of ≥1080P, is installed above the inspection workbench, and has automatic focus and supplementary lighting functions; the barcode scanner is used to obtain the unique identification information of the instrument pack; the card reader is used for operator authentication. The core functional layer includes a video acquisition module, an AI recognition module, a data management module, and an early warning module. The video acquisition module connects to a high-definition camera to achieve real-time video stream acquisition, encoding, compression, and storage management. The AI ​​recognition module, based on the YOLOv5 target detection model, is specifically designed for feature recognition of chemical indicator cards and biological monitoring cards. The data management module performs correlation modeling and persistent storage of video data, recognition results, operator information, and instrument pack information. The early warning module triggers real-time alerts based on AI recognition results and has a process interruption function. Application service layer: includes quality control traceability service, report statistics service, access control service and system integration service; User interface layer: includes web management terminal, operation workstation, mobile APP and large screen display.

[0008] Preferably, the recognition algorithm of the AI ​​recognition module includes color feature extraction, shape matching, position determination, and confidence calculation.

[0009] Preferably, the real-time alerts from the warning module include voice prompts and on-screen pop-up warnings, and the process blocking function can lock subsequent operations when an anomaly is detected.

[0010] Preferably, it also includes an indicator card recognition module, which adopts an indicator card recognition algorithm based on the YOLOv5 target detection model of deep learning. The model is specifically trained for the color features, shape features and position features of chemical indicator cards and biological monitoring cards to realize automatic recognition and confidence calculation of indicator cards.

[0011] Preferably, it also includes an associated storage module, which structurally associates video data, AI recognition results, operator information, and instrument package information to form a quality traceability database containing multi-dimensional data.

[0012] Preferably, it also includes an inspection process control module, which triggers an early warning prompt based on the AI ​​recognition result, blocks the process in case of abnormal recognition, and releases the blockage after the problem is corrected, thereby ensuring the compliance of the inspection process.

[0013] This invention also provides a quality control method for medical device package inspection based on video surveillance and AI recognition, characterized by the following steps based on the above-mentioned system implementation: S1: Operators verify their identity using a card reader, and the system records the operator's information; S2: Scan the instrument pack barcode; the system obtains the unique identifier of the instrument pack and associates it with the operator's information. S3: The system automatically turns on the high-definition camera, and the video capture module begins full-process video recording; S4: The operator performs a manual inspection of the instrument pack, including counting the number of instruments, checking cleanliness and integrity, and placing the instruction card; S5: The AI ​​recognition module analyzes the video stream in real time, automatically identifies the placement of the indicator card, and calculates the confidence level; S6: If the recognition is successful (confidence level ≥ preset threshold), the operator confirms that the recognition is successful; if the recognition fails, the warning module triggers a real-time warning and blocks the process until the recognition is successful. S7: After the operation is completed, the system will associate and store the video file, recognition results, and related information, and generate a quality control report.

[0014] Preferably, in step S5, the AI ​​recognition module uses a specially trained YOLOv5 target detection model to identify the morphological features of the chemical indicator card and the biological monitoring card.

[0015] Preferably, in step S6, the preset threshold is 0.8, which can be adjusted according to the actual application scenario.

[0016] Preferably, the video file naming rules include the instrument package barcode, operator ID, and inspection date and time information, which facilitates traceability and query.

[0017] Compared with the prior art, the technical solution proposed in this application has the following beneficial effects: achieving full traceability of the operation: by recording the entire process video and storing it in association with business data, a complete quality traceability chain is formed, which can trace whether the operation process is standardized afterward and clarify the attribution of responsibility; Improve the reliability of indicator card placement: Based on a specially trained AI recognition algorithm, the system automatically verifies the placement of indicator cards, avoiding ineffective sterilization caused by human oversight and ensuring medical safety; Eliminate quality control lag: Through real-time early warning and process interruption mechanisms, problems can be detected and corrected in a timely manner during the inspection process, preventing unqualified instrument packs from entering the subsequent sterilization process, reducing rework costs and time delays; Achieve integrated data management: Structure and store multi-dimensional data such as video data, recognition results, operator information, and instrument pack information in a linked manner to facilitate querying, statistics, and analysis, and provide data support for quality control management; Facilitating Training and Quality Control Review: Standardized operation video recordings can serve as intuitive teaching materials for new employee training and provide reliable references for quality review, reducing the difficulty of training and quality control. Attached Figure Description

[0018] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described and discussed below with reference to the accompanying drawings. Obviously, what is described here is only a part of the examples of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0020] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1 See Figure 1 This embodiment provides a medical device package inspection quality control system based on video surveillance and AI recognition. Its hardware equipment layer includes: a 1080P high-definition camera, installed 0.8m directly above the inspection workbench, with autofocus and LED fill light functions; an LS2208 barcode scanner, connected to the workstation terminal via a USB interface; a workstation terminal configured with an i7 processor, 16GB of memory, and a 1TB hard drive; and an MFRC522 card reader, supporting contactless IC card authentication.

[0023] In the core functional layer, the video acquisition module uses the H.265 encoding standard to compress the video stream, stores it in MP4 format, and names the video files according to the rule of "instrument package barcode - operator ID - inspection date and time"; the AI ​​recognition module is based on the YOLOv5s model, trained using 5,000 labeled chemical indicator cards and biological monitoring cards image samples, with 300 training iterations and a confidence threshold set to 0.8; the voice prompts of the warning module use TTS speech synthesis technology, the screen pop-up is a red warning box, and the process is blocked by locking the next operation button.

[0024] The application service layer is developed using the Spring Boot framework. The quality control traceability service supports multi-dimensional queries based on conditions such as instrument package barcode, operator ID, and inspection date. The report statistics service can generate reports such as daily / weekly / monthly inspection pass rates and abnormality type distribution. The permission management service sets up three roles: administrator, quality control personnel, and operator, each with different operation permissions. The system integration service connects with the hospital's HIS system through the HL7 interface.

[0025] The web management interface layer is developed using Vue.js and supports browser access; the workstation is running Windows 10 and equipped with a 27-inch high-definition display; the mobile app supports Android and iOS systems; and the large screen display uses a 55-inch LCD video wall to show the inspection progress and anomaly warning information in real time.

[0026] Example 2 See Figure 2 A quality control method for medical device package inspection based on video surveillance and AI recognition, applying the above system, includes the following specific steps: The operator holds the employee IC card close to the card reader, which reads the card information and transmits it to the system. After verifying the identity, the system records the operator's ID "OP001" and name. The operator uses a barcode scanner to scan the barcode "MED20240520001" on the instrument pack, and the system obtains the instrument pack information and associates it with the operator's information. The system automatically starts the high-definition camera, and the video capture module begins recording video. At the same time, the automatic fill light function is turned on to ensure that the brightness of the work surface is ≥500 lux. The operator opened the instrument bag, manually counted the number of instruments as 12, visually inspected all instruments for stains and damage, and then took out the chemical indicator card and placed it in the instrument bag. The AI ​​recognition module receives the video stream in real time, extracts color features to identify the chemical indicator card (red circular card) in the picture, confirms the placement is compliant through shape matching and position judgment, and calculates the confidence level to be 0.89, which is higher than the preset threshold of 0.8. The system displays a "Successfully recognized indicator card" message on the workstation screen, and the operator clicks the "Confirm Pass" button. The operator completes the instrument pack sealing and clicks the "Operation Complete" button on the operation interface; The data management module associates and stores the recorded video file (named "MED20240520001-OP001-20240520103025.mp4"), AI recognition results (confidence 0.89, recognition successful), operator information, and instrument pack information in the database; The system automatically generates a quality control report, displaying information such as the instrument package barcode, operator, inspection time, and inspection result (qualified). This concludes the quality control process for this inspection.

[0027] Example 3 When an operator forgets to place the indicator card, the following procedure applies: After completing the instrument count and cleanliness check, the operator prepared to package the instrument without placing the chemical indicator card. The AI ​​recognition module, after analyzing the video stream, did not detect the indicator card and calculated a confidence level of 0.05, lower than the preset threshold of 0.8. The warning module immediately triggered an alert: issuing a voice prompt "Please place the chemical indicator card," and a red warning box popped up on the workstation screen displaying "No chemical indicator card detected, please place and try again," while simultaneously locking the "Confirm Pass" and "Operation Complete" buttons. Upon hearing the reminder, the operator removed the chemical indicator card and placed it in the instrument package. The AI ​​recognition module re-analyzed the video stream, recognized the indicator card, calculated a confidence level of 0.91, cleared the warning, and unlocked the buttons. The operator continued with subsequent operations, and the process was completed normally.

[0028] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, nor to combinations thereof. Those skilled in the art can make various changes, modifications, or combinations within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A medical device package inspection quality control system based on video surveillance and AI recognition, characterized in that, It includes a hardware device layer, a core function layer, an application service layer, and a user interface layer, with each layer working together. Hardware equipment layer: includes a high-definition camera, barcode scanner, workstation terminal, and card reader; the high-definition camera has a resolution of ≥1080P, is installed above the inspection workbench, and has automatic focus and supplementary lighting functions; the barcode scanner is used to obtain the unique identification information of the instrument pack; the card reader is used for operator authentication. The core functional layer includes a video acquisition module, an AI recognition module, a data management module, and an early warning module. The video acquisition module connects to a high-definition camera to achieve real-time video stream acquisition, encoding, compression, and storage management. The AI ​​recognition module, based on the YOLOv5 target detection model, is specifically designed for feature recognition of chemical indicator cards and biological monitoring cards. The data management module performs correlation modeling and persistent storage of video data, recognition results, operator information, and instrument pack information. The early warning module triggers real-time alerts based on AI recognition results and has a process interruption function. Application service layer: includes quality control traceability service, report statistics service, access control service and system integration service; User interface layer: includes web management terminal, operation workstation, mobile APP and large screen display.

2. The medical device package inspection quality control system based on video surveillance and AI recognition according to claim 1, characterized in that, The AI ​​recognition module's recognition algorithm includes color feature extraction, shape matching, position determination, and confidence calculation.

3. The medical device package inspection quality control system based on video surveillance and AI recognition according to claim 1, characterized in that, The real-time alerts from the warning module include voice prompts and on-screen pop-up warnings, and the process blocking function can lock subsequent operations when an anomaly is detected.

4. A medical device package inspection quality control system based on video surveillance and AI recognition according to claim 1, characterized in that, It also includes an indicator card recognition module, which uses an indicator card recognition algorithm based on the YOLOv5 target detection model of deep learning. The model is specifically trained for the color features, shape features and position features of chemical indicator cards and biological monitoring cards to realize automatic recognition and confidence calculation of indicator cards.

5. A medical device package inspection quality control system based on video surveillance and AI recognition according to claim 1, characterized in that, It also includes an associated storage module, which structurally associates video data, AI recognition results, operator information, and instrument package information to form a quality traceability database containing multi-dimensional data.

6. A medical device package inspection quality control system based on video surveillance and AI recognition according to claim 1, characterized in that, It also includes an inspection process control module, which triggers early warning prompts based on AI recognition results, blocks the process in cases of abnormal recognition, and lifts the blockage after the problem is corrected, thus ensuring the compliance of the inspection process.

7. A quality control method for medical device package inspection based on video surveillance and AI recognition, characterized in that, The system implementation based on any one of claims 1-6 includes the following steps: S1: Operators verify their identity using a card reader, and the system records the operator's information; S2: Scan the instrument pack barcode; the system obtains the unique identifier of the instrument pack and associates it with the operator's information. S3: The system automatically turns on the high-definition camera, and the video capture module begins full-process video recording; S4: The operator performs a manual inspection of the instrument pack, including counting the number of instruments, checking cleanliness and integrity, and placing the instruction card; S5: The AI ​​recognition module analyzes the video stream in real time, automatically identifies the placement of the indicator card, and calculates the confidence level; S6: If the recognition is successful (confidence level ≥ preset threshold), the operator confirms that the recognition is successful; if the recognition fails, the warning module triggers a real-time warning and blocks the process until the recognition is successful. S7: After the operation is completed, the system will associate and store the video file, recognition results, and related information, and generate a quality control report.

8. A quality control method for medical device package inspection based on video surveillance and AI recognition according to claim 6, characterized in that, In step S5, the AI ​​recognition module uses a specially trained YOLOv5 target detection model to identify the morphological features of the chemical indicator card and the biological monitoring card.

9. A quality control method for medical device package inspection based on video surveillance and AI recognition according to claim 6, characterized in that, In step S6, the preset threshold is 0.8, which can be adjusted according to the actual application scenario.

10. A quality control method for medical device package inspection based on video surveillance and AI recognition according to claim 6, characterized in that, The naming rules for the video files include the instrument package barcode, operator ID, and inspection date and time information, facilitating traceability and querying.

Images