Feature Extraction Based Machine Learning for Human Burn Diagnosis From Burn Images.

Area of Science:

• Medical technology
• Computer vision
• Machine learning in healthcare

Background:

• Burn diagnosis relies heavily on expert clinical experience, often requiring specialized facilities.
• Accessibility to expert medical assessment can be limited in non-clinical settings.
• Automated burn assessment tools are needed to aid diagnosis where experts are unavailable.

Purpose of the Study:

• To develop an automated machine learning model for burn diagnosis and classification.
• To create a feature extraction model for classifying burn severity.
• To evaluate the performance of a support vector machine (SVM) based classification model.

Main Methods:

• Utilized automated machine learning, specifically a support vector machine (SVM) model.
• Trained the SVM model on the burns-BIP_US database, classifying images into graft and non-graft categories.
• Evaluated the model using 74 test images against ground truth.

Main Results:

• The proposed SVM-based model achieved an accuracy of 82.43% in burn classification.
• This accuracy is higher than the 79.73% achieved by previous multidimensional scaling analysis (MDS) methods.
• The model demonstrated effectiveness in classifying burns requiring grafts versus those not.

Conclusions:

• Automated machine learning, particularly SVM, offers a viable approach for accurate burn assessment.
• The developed model can assist in burn diagnosis, especially in resource-limited environments.
• This technology has the potential to improve patient outcomes by enabling timely and accurate burn severity classification.