Skip to content

limitations.md

Latest commit

 

History

History
184 lines (140 loc) · 5.23 KB

limitations.md

File metadata and controls

184 lines (140 loc) · 5.23 KB
Raw

Limitations and Constraints

ECG Axis Lab is a visualization and learning tool, not a clinical system. This document outlines its limitations.

Scope Limitations

What This Tool Does NOT Do

  1. Not a diagnostic system

    • Does not classify rhythm (normal sinus, AFib, etc.)
    • Does not detect pathology (MI, LVH, etc.)
    • Does not provide axis categories (normal, left deviation, etc.)
    • Does not provide medical decision support

  2. Not a production ECG reader

    • No real-time monitoring
    • No integration with hospital records
    • No audit trail or regulatory compliance
    • Not validated for clinical use

  3. Not a signal processing toolkit

    • No automatic beat detection
    • No P/QRS/T delineation
    • No artifact removal or filtering
    • No noise reduction

  4. Not PDF/image capable

    • Cannot scan or OCR ECG documents
    • Only accepts tabular data formats

Mathematical Limitations

  1. Single-point projection

    • Uses one summary value per lead (e.g., signed area)
    • Does not track temporal evolution of the vector during the QRS
    • Assumes "average" vector over the window

  2. 2D geometry only

    • Frontal plane only (0°, ±90°, 180°)
    • Ignores anterior-posterior depth (sagittal plane)
    • Precordial leads shown as context, not geometry

  3. Linear model

    • No saturation curves
    • No amplitude normalization
    • Direct scalar projection onto lead axes

  4. Simplified Einthoven

    • Assumes perfect 60° triangle
    • No accounting for anatomical variation
    • No body habitus correction

Data Limitations

  1. Sample format only

    • Requires numeric time-series data
    • Cannot infer from image or PDF
    • Cannot parse unstructured text reports

  2. No lead validation

    • Trusts user-provided lead labels
    • Cannot confirm lead placement quality
    • No detection of swapped or reversed leads

  3. No preprocessing

    • User must provide clean data
    • No automatic baseline drift removal
    • No 60 Hz filter or high-frequency rejection

  4. Implicit assumptions

    • Assumes valid lead placements (standard positions)
    • Assumes normal electrical axis range concept (not enforced)
    • Assumes data is normalized to typical mV scale

Confidence Limitations

  1. Confidence score is heuristic

    • Based on number of leads and residual error
    • Not validated against clinical gold standard
    • Should not be interpreted as clinical accuracy

  2. Single-window estimate

    • Uses one averaged value per lead
    • Does not consider beat-to-beat variation
    • Does not track temporal evolution

  3. Residual-based confidence

    • Low residual does not guarantee correct vector
    • May fit noise as well as signal
    • Not a measure of clinical relevance

Input Data Issues

  1. No quality assessment

    • Does not detect poor recording quality
    • Does not identify excessive noise
    • Does not detect lead disconnection

  2. Limited error handling

    • Silently skips unknown leads
    • Truncates mismatched lead lengths
    • May produce nonsensical results from bad inputs

  3. No delineation

    • Cannot identify QRS window automatically
    • Requires user to crop to QRS window (future feature)
    • May include T wave or other components if not specified

Visualization Limitations

  1. Static displays

    • No beat-to-beat animation
    • No loop display (Lissajous patterns)
    • No 3D rotation

  2. Limited geometric rendering

    • SVG-based, not GPU-accelerated
    • No interactive 3D projection
    • No point cloud or density visualization

  3. No comparative features

    • Cannot overlay multiple ECGs
    • Cannot show before/after
    • Cannot track changes over time

Performance Limitations

  1. Sample count

    • Optimized for ~500-5000 samples per lead
    • High sample counts (>100k) may be slow
    • No streaming or windowed processing

  2. Number of leads

    • Supports up to 12 leads (standard ECG)
    • Not tested with >12 leads
    • Geometry rendering limited to 6 frontal leads

  3. Computational precision

    • Double-precision floating-point
    • ~15-17 significant digits
    • Sufficient for medical ranges

Future Improvements (Post-MVP)

  • Automatic QRS window detection
  • Beat delineation and classification
  • Precordial plane visualization
  • Time-series animation (vector evolution)
  • Multi-patient comparison
  • Advanced noise/artifact filtering
  • Lead validity checking
  • Clinical validation studies

Recommended Use Cases

✓ Learning and teaching ECG concepts ✓ Exploring geometric relationships ✓ Understanding lead projections ✓ Simulation and parameter study ✓ Signal visualization and inspection ✓ Research and pedagogy

Not Recommended For

✗ Clinical diagnosis ✗ Patient monitoring ✗ Decision support ✗ Regulatory or compliance purposes ✗ Automated diagnosis systems ✗ Real-time cardiac monitoring

Always Remember

This tool is for exploration and learning, not clinical care.

If interpreting real patient ECGs:

  • Always involve qualified healthcare professionals
  • Follow established clinical protocols
  • Use FDA-cleared or hospital-validated systems
  • Do not rely solely on ECG Axis Lab output

For questions or to report issues with this tool's behavior, please open an issue on GitHub.