Examining palpebral conjunctiva for anemia assessment with image processing methods.

Area of Science:

• Medical Imaging
• Computational Biology
• Hematology

Background:

• Anemia diagnosis traditionally relies on hemoglobin levels via blood tests.
• Palpebral conjunctiva color analysis is a non-invasive diagnostic standard but suffers from subjective interpretation.
• Automated, objective anemia screening is needed to improve diagnostic accessibility.

Purpose of the Study:

• To develop and evaluate two distinct algorithms for automated anemia diagnosis using digital image analysis of the palpebral conjunctiva.
• To provide both a simple, fast algorithm and a more sophisticated, robust algorithm for diverse application needs.
• To establish a feasible, non-invasive method for anemia screening.

Main Methods:

• Algorithm 1: Two-stage classification using High Hue Rate (HHR) and Pixel Value in the Middle (PVM) features with a minimum distance classifier.
• Algorithm 2: Feature selection from 18 candidates (including entropy) to identify 3 independent features (entropy, binarized HHR, PVM of G component) for classification via Support Vector Machine or Artificial Neural Network.
• Performance evaluation using sensitivity, specificity, and Kappa values.

Main Results:

• Both proposed algorithms demonstrated good classification performance in anemia detection.
• The experimental results validate the feasibility of using automated image analysis for anemia diagnosis.
• The developed methods offer a promising alternative to traditional blood-based anemia testing.

Conclusions:

• Automated analysis of palpebral conjunctiva color is a viable approach for anemia screening.
• The developed algorithms provide effective, non-invasive tools for anemia diagnosis.
• This research supports further investigation into digital health solutions for anemia detection.