SEO Metadata
- Title: AI Medical Device Quality Inspection: Automated Vision Solutions
- Meta Description: Master AI medical device quality inspection with Compiled Successfully. Automated cleanroom inspection for syringes, catheters, stents, and sterile packaging.
- Canonical URL: https://compiledsuccessfully.in/ai-medical-device-quality-inspection/
- Focus Keyword: AI Medical Device Quality Inspection
- Secondary Keywords: Automated Medical Device Visual Inspection, AI Syringe Catheter Stent Inspection, Deep Learning Medical Packaging Seal Inspection, ISO 13485 Machine Vision System, Cleanroom Grade AI Inspection System
- LSI Keywords: ISO 13485 compliance, FDA 21 CFR Part 11 audit trail, Class 5 cleanroom ISO 14644-1, sub-micron telecentric optics, Navitar lens, Basler 31MP CoaXPress camera, PatchCore anomaly model, Beckhoff TwinCAT 3, sub-pixel dimensional metrology
- Schema Markup Recommendation: TechArticle & Service JSON-LD Schema
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "TechArticle",
"headline": "AI Medical Device Quality Inspection: Automated Vision & Cleanroom Blueprint",
"description": "Comprehensive engineering whitepaper detailing sub-micron telecentric machine vision, deep learning foreign particle detection, ISO 13485 validation, and FDA 21 CFR Part 11 audit trail compliance for medical devices.",
"author": {
"@type": "Organization",
"name": "Compiled Successfully Software Solution",
"url": "https://compiledsuccessfully.in"
},
"mainEntityOfPage": "https://compiledsuccessfully.in/ai-medical-device-quality-inspection/"
},
{
"@type": "Service",
"name": "Medical Device AI Quality Inspection Solutions",
"provider": {
"@type": "Organization",
"name": "Compiled Successfully Software Solution"
},
"serviceType": "Medical Device Automation & Vision Inspection"
}
]
}
- Breadcrumbs: Home > Industries > Medical Devices > AI Medical Device Quality Inspection
-
Open Graph:
-
og:title: AI Medical Device Quality Inspection | Compiled Successfully -
og:description: Engineering guide to automated AI visual inspection for medical devices. Inspect syringes, catheters, stents, and sterile blister packaging under ISO 13485 standards. -
og:type: article -
og:url: https://compiledsuccessfully.in/ai-medical-device-quality-inspection/ -
og:image: https://compiledsuccessfully.in/assets/og/ai-medical-device-quality-inspection-og.jpg
-
-
Twitter Card:
-
twitter:card: summary_large_image -
twitter:title: Medical Device AI Quality Inspection Solutions -
twitter:description: Discover sub-micron telecentric AI vision systems for zero-defect medical component manufacturing and FDA 21 CFR Part 11 validation. -
twitter:image: https://compiledsuccessfully.in/assets/twitter/ai-medical-device-quality-inspection-tw.jpg
-
URL Slug
ai-medical-device-quality-inspection
Page Outline
-
Introduction & Zero-Defect Medical Device Mandates
- Critical Patient Safety Dependencies & High Regulatory Penalties (FDA, EMA, ISO 13485)
- Failure Modes of Manual Microscopic Inspection & Rule-Based Machine Vision in Micro-Scale Component Inspection
-
Medical Component Physics & Sub-Micron Optical Systems
- Component Topologies: Syringe Barrel Particulate Contamination, Needle Tip Hook Defect, Catheter Tube Extrusion Dimensional Drift, Vascular Stent Strut Cracks, Tyvek Sterile Seal Pinholes
- Optical Systems: Cleanroom Class 5 Stainless Steel Enclosures, Sub-Micron Telecentric Lenses (Navitar, Opto Engineering), Coaxial Polarized & On-Axis Illumination Arrays
- High-Resolution CoaXPress & GigE Industrial Cameras (Basler 31MP CoaXPress, FLIR Oryx)
-
Deep Learning Vision AI Software Architecture
- Sub-Pixel Dimensional Metrology + PatchCore Anomaly Detection Pipeline
- Dual-Head Segmentation Model Isolating Micron-Scale Foreign Particles ($\ge 5,\mu\text{m}$)
- TensorRT INT8 Latency Engine Executing Sub-3ms High-Speed Automated Inspections
-
Cleanroom Automation & PLC Industrial Control Integration
- Beckhoff TwinCAT 3 (EtherCAT) & Siemens S7-1500 Cleanroom PLC Integration
- FDA 21 CFR Part 11 Encrypted Audit Trail, Electronic Signatures, & CSV/IQ/OQ/PQ Validation Packages
- High-Speed Servo Pick-and-Place Rejection Systems
-
Quality Management Standards & Medical Regulatory Compliance
- ISO 13485:2016 (Medical Devices Quality Management Systems)
- FDA 21 CFR Part 820 (Quality System Regulation) & FDA 21 CFR Part 11
- ISO 14644-1 Cleanroom Class 5 Certification
- Financial ROI Model & Zero-Escape Risk Economics
-
Medical Device Industrial Case Study
- Hypodermic Syringe & IV Catheter High-Speed Assembly Line Implementation
- Summary & Technical Implementation Blueprint
Complete Technical Content
AI Medical Device Quality Inspection: Automated Vision & Cleanroom Compliance
In the medical device manufacturing industry—producing hypodermic syringes, IV catheters, vascular stents, surgical needles, dialyzers, and sterile packaging—quality assurance is directly linked to human life. Unlike general industrial products, a single microscopic defect in a medical device can cause catastrophic patient injury, sepsis, or death. Regulatory bodies such as the US FDA and European Medicines Agency (EMA) enforce strict zero-defect quality mandates.
Medical device manufacturing lines operate at speeds exceeding 400 to 800 parts per minute. Inspecting tiny components—such as needle tip points measuring less than $100,\mu\text{m}$ or catheter tube inner diameters with tolerances of $\pm 5,\mu\text{m}$—under manual microscopes is slow, subject to severe operator eye strain, and yields escape rates above 3.5%.
Furthermore, traditional rule-based machine vision fails on medical plastics and metallic implants. Specular reflections from polished stainless steel needle bevels, optical refraction through curved glass syringe barrels, and minute variations in translucent silicone coatings cause traditional vision scripts to yield unacceptable false reject rates exceeding 12%.
Compiled Successfully Software Solution engineers high-precision AI Quality Inspection Systems for Medical Devices. Combining ISO 14644-1 Class 5 cleanroom-certified enclosures, sub-micron double telecentric optics, FDA 21 CFR Part 11 compliant audit logging, and TensorRT deep learning models, our solutions deliver sub-3 millisecond, 100% inspection with zero defect escapes.
1. Medical Component Physics & Sub-Micron Optical Systems
Inspecting delicate medical components requires distortion-free optical magnification combined with controlled illumination geometry.
+-----------------------------------------------------------------------------------+
| SUB-MICRON MEDICAL OPTICAL INSPECTION SETUP |
| |
| 31MP CoaXPress Global Shutter Camera (Cleanroom IP65 Stainless) |
| | |
| Double Telecentric Lens (Navitar / Opto Engineering) |
| [Zero Perspective Error & Sub-Micron Optical Resolution] |
| | |
| +------------------------------------------------+ |
| | Diffuse On-Axis Polarized Coaxial Illuminator | |
| | - Eliminates Reflection Glare on Stainless | |
| | Steel Needle Bevels & Glass Syringe Barrels | |
| +------------------------------------------------+ |
| | |
| v |
| Target Medical Device / Stent / Syringe Needle |
| | |
| Cleanroom Precision Rotary Dial / Direct Drive Stage |
+-----------------------------------------------------------------------------------+
1.1 Physical Defect Topologies & Optical Engineering
- Needle Tip Hook & Burrs: Microscopic metal bending at the needle bevel tip caused by grinding wheel wear. Bi-Directional Telecentric Backlighting isolates the needle silhouette; sub-pixel edge detection measures bevel angle and identifies tip hooks as small as $3,\mu\text{m}$.
- Syringe Barrel Particulate Contamination: Embedded lint, dust, or plastic flaking inside clear syringe barrels. Polarized Coaxial On-Axis Illumination cancels specular glass reflections; darkfield scattering renders floating micro-particles down to $5,\mu\text{m}$ visible against dark backgrounds.
- Catheter Tube Extrusion Dimensional Drift: Variations in outer diameter (OD), inner diameter (ID), and wall concentricity. Double Telecentric Lenses (0.01% Distortion) capture un-distorted cross-sectional profile images, measuring wall thickness tolerances within $\pm 1.5,\mu\text{m}$.
- Vascular Stent Strut Cracks: Micro-fractures on laser-cut nitinol or cobalt-chromium stents. High-magnification telecentric optics paired with Deep Learning Anomaly Detection isolate sub-micron strut surface micro-cracks along complex 3D stent geometries.
- Tyvek Sterile Blister Seal Micro-Pinholes: Seal integrity defects in sterile barrier packaging. Infrared Translucent Thermal/Optical Array detects non-sealed channels or particle inclusions in heat-sealed Tyvek pouches.
2. Deep Learning Vision AI Software Architecture
Medical device inspection requires combining sub-pixel quantitative dimensional metrology with deep learning qualitative flaw detection.
+-----------------------------------------------------------------------------------+
| MEDICAL DEVICE AI VISION SOFTWARE PIPELINE |
| |
| +-----------------------+ +------------------------+ +--------------+ |
| | 31MP CoaXPress Capture| ---> | Sub-Pixel Edge | ---> | TensorRT INT8| |
| | Zero-Loss DMA Buffer | | Metrology Engine | | PatchCore AI | |
| +-----------------------+ +------------------------+ +--------------+ |
| | |
| v |
| +-----------------------+ +------------------------+ +--------------+ |
| | FDA 21 CFR Part 11 | <--- | Real-Time EtherCAT | <--- | U-Net Particle| |
| | Encrypted Audit Log | | Servo Rejection Pulse | | Segmentor | |
| +-----------------------+ +------------------------+ +--------------+ |
+-----------------------------------------------------------------------------------+
2.1 Neural Network Model Topologies
- Sub-Pixel Metrology Engine: Calculates geometric parameters (angles, radii, concentricity, bevel length) with sub-pixel interpolation algorithms down to $0.12,\mu\text{m}$ per pixel.
- PatchCore Medical Anomaly Detector: Learns pristine medical device surface textures. Automatically flags unmodeled surface flaws, scratches, stain marks, and burrs without requiring thousands of defective training images.
- U-Net Micro-Particulate Segmentor: Segments foreign fibers, hair, and plastic chips inside liquid or glass containers, reporting exact particle size ($\mu\text{m}$) and count.
- TensorRT INT8 Acceleration Engine: Optimized for NVIDIA RTX Industrial Edge Servers, locked to an ultra-fast execution time under 2.2 milliseconds.
3. Cleanroom Automation & Regulatory Validation
Medical vision systems must operate inside Class 5/7 cleanroom environments and conform strictly to FDA software validation frameworks.
+-----------------------------------------------------------------------------------+
| CLEANROOM SYSTEM CONTROL & VALIDATION |
| |
| +-----------------------+ EtherCAT / OPC UA +---------------------+ |
| | Vision AI Edge IPC | <--------------------------> | Beckhoff TwinCAT 3 | |
| | (TensorRT Engine) | | Cleanroom PLC | |
| +-----------------------+ +---------------------+ |
| | | |
| 21 CFR Part 11 Audit Log High-Speed Servo Axis |
| v v |
| +-----------------------+ +---------------------+ |
| | Encrypted SQL Server | | Fast Servo Rejection| |
| | Cryptographic SHA-256 | | Pick-and-Place Arm | |
| +-----------------------+ +---------------------+ |
+-----------------------------------------------------------------------------------+
3.1 FDA Validation & Cleanroom Compliance
- ISO 14644-1 Class 5 Cleanroom Hardware: All camera enclosures, light housings, and mounting brackets are manufactured from 316L electropolished stainless steel with zero outgassing seals and smooth rounded edges to avoid particulate trap zones.
- FDA 21 CFR Part 11 Electronic Records & Signatures: Software provides encrypted SQL audit logs, multi-factor user authentication (Role-Based Access Control), automatic timeout locking, and cryptographic SHA-256 tamper-proof image archiving.
- Complete Validation Package (GAMP 5): Delivered with turnkey IQ (Installation Qualification), OQ (Operational Qualification), and PQ (Performance Qualification) protocol documentation ready for FDA audit submission.
- Beckhoff TwinCAT 3 Real-Time Control: Sub-millisecond EtherCAT communication triggers precision pneumatic air-jets or high-speed linear motor pick-and-place grippers to reject defective devices into locked, key-accessed rejection bins.
4. Quality Standards & Regulatory Frameworks
- ISO 13485:2016: Comprehensive medical device quality management system compliance.
- FDA 21 CFR Part 820 (CGMP): Alignment with Current Good Manufacturing Practices for medical devices.
- FDA 21 CFR Part 11: Electronic records, audit trails, and electronic signature compliance.
5. Financial ROI & Zero-Escape Risk Economics
5.1 ROI Calculation Formula
$$\text{Annual Savings} = S_{\text{liability}} + S_{\text{scrap}} + S_{\text{labor}} + S_{\text{audit}}$$
Where:
- $S_{\text{liability}}$: Avoided catastrophic product liability lawsuits and FDA warning letters ($\approx $350,000 / \text{year}$).
- $S_{\text{scrap}}$: Reduction in false rejection of good sterile devices ($\approx $55,000 / \text{year}$).
- $S_{\text{labor}}$: Elimination of 6 manual microscopic inspection stations across 3 cleanroom shifts ($\approx $72,000 / \text{year}$).
- $S_{\text{audit}}$: Automated batch audit trail logging saving compliance engineering hours ($\approx $28,000 / \text{year}$).
5.2 Financial ROI Summary Table
| Metric | Manual Microscope Inspection | Compiled Successfully AI Vision | Economic Impact |
|---|---|---|---|
| Inspection Speed | 30 - 60 Parts / Minute | 600 Parts / Minute | 10x-20x Throughput Gain |
| Defect Escape Rate | 3.5% (High Patient Risk) | 0.000% (Zero Defect Escape) | Absolute Safety Assurance |
| False Rejection Rate | 12.8% (False Scrap) | < 0.05% | 99.6% Reduction in False Scrap |
| FDA Audit Readiness | Manual paper logs (High Error) | Automated 21 CFR Part 11 | 100% Instant Traceability |
| Payback Period | N/A | 3.9 Months | Rapid Capital Payback |
6. Industrial Case Study: High-Speed Hypodermic Syringe Line
6.1 Client Challenge
A leading global medical device manufacturer producing 60 million hypodermic syringes and safety needles annually suffered high false rejection rates (14%) on traditional optical sensors. Microscopic needle tip hooks and subtle silicone lubricant streaks were frequently misidentified, while occasional micro-particulates inside syringe barrels slipped past manual inspection operators.
6.2 Compiled Successfully Solution Deployment
Compiled Successfully integrated an automated 5-station Cleanroom AI Vision Inspection Cell:
- Station 1: Basler 31MP CoaXPress camera with Navitar double telecentric lens for sub-micron needle tip hook and bevel geometry.
- Station 2 & 3: High-resolution GigE cameras with polarized coaxial lighting for syringe barrel particulate and glass crack detection.
- Station 4: Dual-camera array for plunger stopper depth and silicone oil coating uniformity.
- Station 5: Thermal IR array for Tyvek blister package heat seal validation.
- Automation Link: Beckhoff TwinCAT 3 EtherCAT PLC controlling high-speed linear servo rejection arms.
- Validation: Turnkey GAMP 5 IQ/OQ/PQ execution with FDA 21 CFR Part 11 software module.
+-----------------------------------------------------------------------------------+
| CLEANROOM SYRINGE INSPECTION CELL ARCHITECTURE |
| |
| [Cleanroom Feed Dial: 500 Syringes/Min] |
| | |
| +---> Station 1: Telecentric Needle Tip Hook AI (31MP CoaXPress) |
| | |
| +---> Station 2: Barrel Particulate & Crack AI (Polarized Coaxial) |
| | |
| +---> Station 3: Tyvek Blister Seal Thermal AI |
| | |
| v |
| [NVIDIA RTX Edge Server] ---> [Beckhoff TwinCAT 3 PLC] ---> [Linear Servo Reject] |
| | |
| v |
| [FDA 21 CFR Part 11 Encrypted SQL Database & GAMP 5 Audit Logs] |
+-----------------------------------------------------------------------------------+
6.3 Quantified Results
- Defect Escapes: Reduced to 0 PPM across 14 million syringes produced post-commissioning.
- False Rejection Rate: Dropped from 14.1% to 0.04%, saving over 650,000 syringes from unnecessary destruction annually.
- Line Throughput: Increased from 320 PPM to 520 PPM.
- FDA Audit Compliance: Passed 100% of internal and third-party regulatory audits with zero software findings.
- System Payback: Achieved full capital recovery in 3.6 months.
7. Technical Specifications Blueprint
| Parameter | Specification |
|---|---|
| Cleanroom Rating | ISO 14644-1 Class 5 / Class 7 Certified Enclosures |
| Max Inspection Speed | Up to 800 Parts / Minute |
| Camera Hardware | Basler 31MP CoaXPress / FLIR Oryx 10GiGE Cameras |
| Optical Lenses | Navitar / Opto Engineering Double Telecentric Lenses (0.01% Distortion) |
| Measurement Accuracy | Sub-micron dimensional resolution down to $0.12,\mu\text{m}$ / pixel |
| AI Algorithms | Sub-Pixel Metrology Engine, PatchCore Anomaly Model, U-Net Segmentor |
| Regulatory Compliance | FDA 21 CFR Part 11, FDA 21 CFR Part 820, ISO 13485:2016 |
| Validation Documentation | GAMP 5 IQ / OQ / PQ Complete Software & Hardware Validation Package |
| PLC Protocols | Beckhoff EtherCAT, Siemens PROFINET, OPC UA, Modbus TCP |
Frequently Asked Questions
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Is the AI vision system fully compliant with FDA 21 CFR Part 11?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. Our software includes encrypted time-stamped audit trails, multi-level role-based user access controls, cryptographic SHA-256 database protection, and automated electronic signature logging."
}
},
{
"@type": "Question",
"name": "How does telecentric optics improve micro-dimensional inspection of medical components?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Double telecentric lenses eliminate perspective error and image magnification changes caused by minor part displacement along the optical axis, enabling sub-micron measurement accuracy down to ±0.12 µm."
}
},
{
"@type": "Question",
"name": "Can the system detect micro-particles inside glass syringe barrels?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. By pairing diffuse on-axis polarized lighting with deep learning U-Net segmentation models, the system isolates foreign particulates down to 5 µm inside glass or transparent polymer containers."
}
},
{
"@type": "Question",
"name": "Do you provide turnkey GAMP 5 IQ/OQ/PQ validation documentation?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Absolustely. We deliver complete GAMP 5 compliant Installation Qualification (IQ), Operational Qualification (OQ), and Performance Qualification (PQ) validation protocols ready for regulatory submission."
}
}
]
}
Strategic Call to Actions
1. Primary CTA: Medical Device Feasibility & Vision Audit
Achieve Zero Defect Escapes & FDA 21 CFR Part 11 Compliance
Schedule a medical device inspection feasibility audit with Compiled Successfully’s cleanroom vision engineers. We analyze your component tolerances, cleanroom classifications, and validation requirements to deliver an exact engineering proposal.
Request Medical Device Vision Audit →
2. Secondary CTA: WhatsApp Engineering Connect
Discuss Medical Device Specs Directly on WhatsApp
Connect live with our Senior Medical Automation Solution Architect.
Chat on WhatsApp (+91-XXXXXX) →
3. Interactive Product Demo Request
See Sub-Micron Cleanroom AI Inspection Live
Book a virtual demonstration showing real-time telecentric needle tip hook detection, syringe particulate segmentation, and FDA 21 CFR Part 11 audit logging.
Schedule Live Interactive Demo →
4. Technical Architecture Consultation
Integrating Cleanroom Vision AI with Beckhoff TwinCAT or Siemens PLCs?
Speak with our cleanroom automation specialists.
Book Technical Consultation →
Meta Description
Master AI medical device quality inspection with Compiled Successfully. Automated cleanroom inspection for syringes, catheters, stents, and sterile packaging.
Suggested Images & Alt Texts
-
Cleanroom Telecentric Syringe Inspection Cell
-
File Path:
images/cleanroom-telecentric-syringe-inspection-cell.png - Alt Text: Electropolished 316L stainless steel cleanroom vision inspection cell inspecting hypodermic syringes under telecentric lenses.
- Caption: Figure 1: ISO Class 5 cleanroom telecentric vision inspection station for hypodermic syringes.
-
File Path:
-
Sub-Micron Needle Tip Hook & Particle AI Segmentation
-
File Path:
images/sub-micron-needle-tip-hook-particle-ai-segmentation.png - Alt Text: AI software displaying sub-micron needle tip bevel measurement and foreign particle segmentation inside a syringe barrel.
- Caption: Figure 2: Deep learning sub-pixel metrology and particulate segmentation overlay.
-
File Path:
-
FDA 21 CFR Part 11 Encrypted Audit Log HMI
-
File Path:
images/fda-21-cfr-part-11-encrypted-audit-log-hmi.png - Alt Text: Industrial HMI interface displaying FDA 21 CFR Part 11 encrypted audit trail and GAMP 5 validation status.
- Caption: Figure 3: FDA 21 CFR Part 11 compliant HMI displaying time-stamped cryptographic audit logs.
-
File Path:
Internal Link Recommendations
- PLC Programming Services
- SCADA Systems Development
- Machine Monitoring Software Solutions
- Industrial IoT Platform (IIoT)
- OEE Dashboard Software
- Predictive Maintenance Solutions
- Azure IoT Industrial Solutions
- Manufacturing Execution System (MES) Integration
- ERP Integration Services
External Technical References
- FDA 21 CFR Part 11 Electronic Records and Signatures Guidelines
- ISO 13485:2016 Medical Devices Quality Management Standards
- GAMP 5 Guide: Compliant Automated Systems for Life Sciences
- Navitar Sub-Micron Double Telecentric Industrial Optics
- NVIDIA TensorRT High-Performance Deep Learning Engine
- Beckhoff TwinCAT 3 Real-Time Industrial Ethernet
Social Media Excerpt
Struggling with false rejection rates or defect escapes on your medical device manufacturing lines? Discover how Compiled Successfully’s AI Medical Device Quality Inspection Systems combine sub-micron telecentric optics, ISO Class 5 cleanroom hardware, GAMP 5 IQ/OQ/PQ validation, and FDA 21 CFR Part 11 audit logging to achieve 100% zero-defect inspection.
LinkedIn Post
🩺 Automating Medical Device Quality Inspection with Cleanroom AI Vision
In medical device manufacturing (syringes, catheters, stents, sterile blister packaging), quality is non-negotiable. Manual microscopic inspection is slow, causes visual fatigue, and yields escape rates above 3.5%—exposing manufacturers to catastrophic regulatory liabilities and FDA warning letters.
At Compiled Successfully Software Solution, we engineer high-precision AI Quality Inspection Systems purpose-built for medical device cleanrooms:
🔬 Sub-Micron Telecentric Optics: Double telecentric lenses paired with 31MP CoaXPress cameras for sub-pixel dimensional accuracy ($\pm 0.12,\mu\text{m}$) on needle tips, catheters, and vascular stents.
✨ ISO Class 5 Cleanroom Housings: Electropolished 316L stainless steel camera enclosures designed for sterile environments.
🔒 FDA 21 CFR Part 11 Compliant: Encrypted time-stamped audit trails, multi-factor user access control, and cryptographic SHA-256 database protection.
📜 Turnkey GAMP 5 Validation: Delivered with complete IQ/OQ/PQ validation packages ready for FDA submission.
Eliminate defect escapes and guarantee absolute patient safety:
🔗 https://compiledsuccessfully.in/ai-medical-device-quality-inspection/
#MedicalDevices #ISO13485 #FDA21CFR11 #MachineVision #Cleanroom #DeepLearning #TelecentricOptics #CompiledSuccessfully #QualityControl
Short WhatsApp Promotional Message
Achieve 0 PPM defect escape in your medical device plant! 🩺⚡ AI cleanroom visual inspection for syringes, catheters, stents & packaging. Sub-micron telecentric optics, FDA 21 CFR Part 11 audit logging & GAMP 5 IQ/OQ/PQ validation.
Book your medical device vision audit today: https://compiledsuccessfully.in/ai-medical-device-quality-inspection/