back

HaemoPixel

Development and Validation of a Mobile Application Using Irrigation Fluid Haemoglobin Colour Analysis to Quantify Intraoperative Blood Loss in Urological Endoscopic Surgery

Dr Samudra Dharmasiri

Overall Study Concept

This research will evolve from a visual haemoglobin colour scale into a mobile application that:

• Captures an image of irrigation fluid
• Objectively analyses colour intensity
• Calculates haemoglobin concentration
• Estimates total blood loss in real time

The app will reduce observer bias, improve accuracy, and provide instant intraoperative decision support, especially valuable in resource-limited operating theatres.

---

Study Structure (Multiphase Design)

Phase I – Digital Colour Scale Calibration & Algorithm Development

Aim

To create a validated digital haemoglobin colour calibration model for irrigation fluid analysis using smartphone cameras.

Methods

1. Reference Sample Preparation

• Serial dilutions of venous blood in standard irrigation fluid
• Known haemoglobin concentrations (0.2–10 g/dL)
• Measured using laboratory spectrophotometry (gold standard)

2. Image Acquisition Protocol

• Samples placed in standardized transparent containers
• Controlled lighting conditions:
  • White background
  • Fixed distance
  • No flash
• Images captured using multiple smartphone models (Android & iOS)

3. Colour Data Extraction

• RGB / HSV / LAB colour values extracted from images
• Analysis of:
  • Mean pixel intensity
  • Red-channel dominance
  • Optical density equivalents

4. Algorithm Development

• Regression or machine learning model:
  • Input: Colour metrics
  • Output: Estimated haemoglobin concentration
• Model internally validated using cross-validation

Outputs

• Digital haemoglobin colour calibration curve
• Core computational algorithm for the app

---

Phase II – Mobile Application Development

App Features

Core Functions

• Camera-based irrigation fluid image capture
• Automated colour analysis
• Haemoglobin concentration estimation
• Blood loss calculation using:

$$\text{Blood Loss (mL)} = \frac{\text{Hb}_{\text{fluid}} \times \text{Total Irrigation Volume}}{\text{Patient Pre-op Hb}}$$

User Inputs

• Procedure type (TURP / TURBT / PCNL)
• Preoperative patient haemoglobin
• Total irrigation volume used

Outputs

• Estimated blood loss (mL)
• Risk stratification (Low / Moderate / High)
• Alerts for excessive blood loss
• Data export (PDF / CSV)

Technical Specifications

Specification	Details
Platform	Cross-platform (Flutter / React Native)
Connectivity	Offline functionality
Security	End-to-end encryption
Privacy	No patient identifiers stored (anonymised case ID)

---

Phase III – Clinical Validation Study

Study Design

Prospective diagnostic accuracy study

Participants

100–150 adult patients undergoing endoscopic urological procedures

Methodology

• Irrigation fluid collected as per protocol
• App-based estimation performed intraoperatively
• Parallel assessments:
  • Laboratory haemoglobin of irrigation fluid
  • Manual colour scale estimation
  • Perioperative haemoglobin changes recorded

Outcome Measures

Primary Outcome:

• Agreement between app-estimated blood loss and laboratory-calculated blood loss

Secondary Outcomes:

• Accuracy compared to visual colour scale
• Time to estimation
• Inter-device variability
• User satisfaction (surgeons & anaesthetists)

Statistical Analysis

• Bland–Altman analysis
• Intraclass correlation coefficient
• ROC curves for detection of significant blood loss (>500 mL)
• Subgroup analysis by procedure

---

Phase IV – Clinical Utility & Implementation Study

Aim

To assess whether app use improves intraoperative decision-making.

Endpoints

• Reduction in unnecessary transfusions
• Earlier recognition of excessive bleeding
• Improved documentation of blood loss
• User adoption rate

---

Ethical & Regulatory Considerations

• ✅ Ethics approval for clinical data collection
• ✅ App classified as clinical decision support tool
• ✅ Compliance with:
  • GDPR / local data protection laws
  • ISO 13485 principles (if expanded)
• ⚠️ Explicit disclaimer: "Adjunct tool—not a substitute for clinical judgment"

---

Innovation & Strengths

Strength	Description
🥇 First of its kind	First endoscopy-specific blood loss quantification app
🎯 Objective	Real-time, objective measurement
💰 Low-cost	Scalable and affordable
🌍 Accessible	Ideal for LMIC settings and emergency surgery

---

Limitations & Risk Mitigation

Limitation	Mitigation Strategy
Lighting variation	In-app lighting calibration
Camera variability	Device-specific correction
Turbid fluids	Algorithm training on real samples
Extreme bleeding	Upper-limit warning

---

Expected Deliverables

• ✅ Validated mobile app (prototype → clinical version)
• 📄 Peer-reviewed publications:
  • Algorithm development
  • Clinical validation
  • Implementation outcomes
• 💡 Potential patent / intellectual property

---

Future Extensions

• 🔗 Integration with anaesthesia monitors
• ☁️ Cloud-based analytics
• 🩺 Expansion to hysteroscopy, arthroscopy, GI endoscopy