DeramScan — AI Burn Severity Assessment System

DeramScan — AI-Powered Burn Severity Assessment System Inspiration DeramScan was inspired by a real clinical uncertainty I experienced after suffering a burn injury on my hand. My diagnosis was based solely on visual inspection, and I received conflicting medical opinions from different doctors. This revealed a critical problem in emergency medicine: burn severity assessment is subjective, inconsistent, and highly experience-dependent. This led to a core question: How can we build a fast, objective, and reliable system that supports clinicians in accurately assessing burn severity within seconds? DeramScan was created to address this real-world gap in emergency care.

What it does DeramScan is a real-time multimodal AI-assisted burn triage system designed for emergency and pre-hospital environments. It: Captures RGB visual data Captures thermal imaging signals Measures skin temperature differentials Fuses all inputs using a confidence-weighted decision model Output: Burn severity classification Triage priority (Red / Yellow / Green) Confidence score for decision reliability Performance: Inference time: < 5 seconds Designed for offline / edge operation The system is strictly a clinical decision support tool, not an autonomous diagnostic system. How we built it DeramScan is built as a multisensor embedded intelligence system combining hardware sensing and algorithmic decision-making. Hardware Components: MLX90618 Infrared Temperature Sensor → non-contact skin temperature detection DS18B20 Temperature Sensor → precise temperature validation TCS34725 Color Sensor → skin color and tissue change detection ESP32 / Arduino Microcontroller → real-time data processing unit OLED Display → instant visual output of results Bluetooth Module → wireless transmission to mobile device System Pipeline: Data acquisition (thermal + RGB + temperature) Preprocessing and normalization Multisensor feature fusion Confidence-based classification logic Real-time output generation AI/Decision Logic: DeramScan uses a multimodal confidence-weighted fusion system where: Each sensor produces a severity estimate Each estimate includes a confidence score Final decision is computed using weighted aggregation This improves robustness in: variable lighting conditions sensor noise environmental distortion Human-in-the-Loop Decision (CRITICAL SAFETY LAYER) DeramScan is designed as a human-supervised AI system. Decision Control: AI provides recommendations only Final clinical decision remains with healthcare professionals Safety Mechanism: Confidence ≥ 80% → AI recommendation displayed Confidence < 80% → mandatory human review required This ensures safe deployment in high-risk medical environments and prevents over-reliance on automation. Responsible AI DeramScan was designed with medical safety and fairness as core principles. Identified Risks: Skin tone and lighting bias in visual analysis Thermal sensor noise and environmental interference Limited generalization across diverse patient datasets Risk of misclassification in extreme burn cases Mitigation Strategies: Multimodal fusion (RGB + thermal + physiological data) reduces single-source bias Confidence scoring system detects uncertainty and prevents overconfident predictions Human override requirement for low-confidence outputs Edge computing design ensures reliability without network dependency ⚖️ Ethical Principle: DeramScan is intentionally designed to augment clinicians, not replace them, ensuring transparency, accountability, and safe decision support. Challenges we ran into Achieving reliable accuracy using non-contact sensors only Defining medically meaningful thresholds between burn degrees Balancing simplicity with clinical relevance Ensuring real-time processing under hardware constraints Minimizing environmental interference (light, heat noise) Accomplishments we're proud of Transforming a real medical experience into a functional prototype concept Designing a full multimodal diagnostic pipeline (hardware + logic) Building a fast, non-invasive emergency triage system Creating a solution with direct relevance to emergency healthcare workflows Integrating safety-aware decision logic into the system design What we learned Medical diagnosis can vary significantly without objective tools Early burn assessment has a major impact on patient outcomes Multisensor fusion significantly improves reliability over single-sensor systems Real innovation often comes from personal pain points Embedded systems can solve meaningful healthcare problems when combined with intelligent decision logic What’s next for DeramScan Future development will expand DeramScan into a full intelligent emergency assistant: AI-powered deep learning burn classification model Mobile application for doctors and patients First-aid recommendation engine based on severity Voice alerts for critical burn cases (third-degree) Patient recovery tracking and monitoring dashboard Final Summary DeramScan is a multimodal AI-assisted burn triage system that combines thermal sensing, RGB vision, and physiological temperature analysis into a confidence-based decision model for emergency care. It reduces subjectivity, accelerates triage decisions, and improves consistency in high-pressure environments while maintaining full human clinical control.

1. Problem Understanding & Context Burn injuries affect millions of people worldwide and remain one of the most challenging emergencies to assess accurately in the first minutes after injury. Current burn assessment often relies on visual inspection, which can be subjective and highly dependent on the experience of the healthcare provider. Misjudging burn severity may lead to delayed treatment, unnecessary transfers, improper triage, and increased patient risk. DeramScan was created to address this critical gap by providing a fast, objective, non-contact, and AI-assisted method for evaluating burn severity directly at the point of care. The goal is to support healthcare professionals with rapid and reliable information when every second matters.
2. AI Reasoning & Appropriateness Artificial Intelligence is the core component that transforms raw sensor measurements into meaningful clinical insights. DeramScan uses computer vision and multimodal sensor fusion to analyze multiple indicators simultaneously, including skin color changes, thermal patterns, and temperature measurements. Unlike human visual assessment, which can be affected by lighting conditions, fatigue, experience level, and subjective judgment, AI applies the same evaluation criteria consistently for every patient. Unlike traditional devices that rely on fixed thresholds or lookup tables, AI can recognize complex relationships between multiple variables, learn from clinical burn datasets, adapt to variations in skin tones and burn appearances, and continuously improve its performance as additional data becomes available. This enables more accurate, scalable, and standardized burn assessment, making AI not only appropriate but essential for solving the problem effectively.
3. Solution Design & Architecture DeramScan is designed as a practical emergency-care solution that combines speed, reliability, and intelligent decision support. Inputs:
4. Infrared thermal measurements
5. Digital temperature readings
6. Skin color and surface appearance data Processing Layer:
7. ESP32/Arduino-based processing unit
8. Sensor fusion engine for combining multiple physiological signals AI Layer:
9. Computer vision analysis
10. Pattern classification model
11. Confidence scoring system
12. Computer vision, multimodal sensor fusion, pattern classification, and confidence scoring for burn severity assessment. Outputs:
13. Burn severity classification (1st, 2nd, or 3rd degree)
14. Confidence score
15. Emergency Priority Index (Green, Yellow, Red)
16. OLED display results Connectivity:
17. Bluetooth communication
18. Mobile monitoring interface
19. Future physician review integration The complete assessment process is designed to deliver results within seconds, supporting rapid triage and treatment decisions in emergency situations.
20. Impact & Decision Value DeramScan improves decision-making during the most critical stage of burn management: the first assessment. The system helps healthcare professionals, paramedics, and emergency responders quickly determine burn severity and treatment urgency. By reducing uncertainty and standardizing assessment procedures, DeramScan can improve triage efficiency, support earlier intervention, and potentially improve patient outcomes. The solution is particularly valuable in ambulances, rural healthcare settings, disaster-response situations, and locations where burn specialists may not be immediately available.
21. Responsibility, Ethics & Limitations DeramScan is designed as a clinical decision-support tool and not as a replacement for professional medical judgment. To ensure safe and responsible deployment:
22. The system provides preliminary assessments only.
23. Users are clearly informed that professional medical evaluation remains necessary.
24. Final diagnosis and treatment decisions remain the responsibility of qualified healthcare professionals.
25. If the AI confidence score falls below 80%, the system automatically flags the case for mandatory human review before any treatment decision is made.
26. Assessment results are intended to support, not replace, clinical expertise.
27. Future development will include training on larger clinical datasets, continuous validation, and improved calibration across diverse patient populations. These safeguards ensure that DeramScan remains ethical, transparent, and trustworthy while delivering meaningful clinical value.
28. AI Signals
29. AI Signals DeramScan uses multiple AI signals to improve burn assessment accuracy and reliability. These signals include:
30. Skin color variation and discoloration patterns
31. Surface appearance characteristics
32. Infrared temperature distribution
33. Digital temperature measurements
34. Temperature differences between affected and surrounding tissue The AI model combines these signals through multimodal sensor fusion to identify patterns associated with different burn severities. By analyzing several indicators together rather than relying on a single measurement, the system can produce more robust and clinically useful assessments.
35. Human Role Humans remain central to every stage of the DeramScan workflow. Healthcare professionals are responsible for:
36. Performing the physical examination
37. Positioning and operating the device
38. Reviewing AI-generated results
39. Interpreting findings within the patient's clinical context
40. Making final diagnosis and treatment decisions The AI system functions solely as a decision-support tool. If the confidence score falls below 80%, the system automatically requests mandatory human review before any action is taken. This ensures that medical expertise always remains the final authority.
41. Pitch DeramScan is an AI-powered burn assessment system designed to provide fast, objective, and non-contact evaluation of burn severity. By combining computer vision, thermal sensing, and intelligent pattern classification, the system helps healthcare professionals make faster and more consistent decisions during critical emergency situations. Unlike traditional visual assessment methods, DeramScan reduces subjectivity, improves triage efficiency, and provides actionable clinical insights within seconds. Our vision is to make accurate burn assessment accessible everywhere—from ambulances and emergency rooms to rural clinics and disaster-response environments—ultimately helping healthcare providers deliver faster care and improve patient outcomes.

Built With

• a
• accurate
• an
• and
• arduino
• bluetooth
• body
• burns.
• by
• caused
• changes
• color
• communication
• contact.
• data
• data.
• detects
• digital
• direct
• display
• ds18b20
• esp32
• examination
• for
• infrared
• main
• measures
• microcontroller
• mlx90618
• module
• oled
• or
• phone
• processes
• readings.
• receives
• results
• screen
• sensor
• show
• skin
• small
• tcs34725
• temperature
• that
• the
• to
• transfers
• used
• wirelessly
• without