Tensor Decomposition for Colour Image Segmentation of Burn Wounds

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Accurate burn wound assessment, including burn area and depth estimation, is critical for effective surgical management and reducing patient mortality.
• Current clinical methods for burn parameter estimation are prone to errors, highlighting the need for automated solutions.
• Automated burn area calculation, known as image segmentation, is essential for improving assessment accuracy.

Purpose of the Study:

• To propose a novel automated segmentation method for burn wound images.
• To improve the accuracy and efficiency of burn area quantification.
• To leverage tensor decomposition for enhanced texture feature extraction in burn wound classification.

Main Methods:

• Utilized tensor decomposition of color images for burn wound segmentation.
• Extracted effective texture features from segmented images for classification.
• Developed a new image segmentation technique specifically for burn assessment.

Main Results:

• The proposed tensor decomposition method demonstrated superior segmentation accuracy compared to existing methods.
• The new method achieved faster computational speeds than conventional approaches.
• Experimental results validated the effectiveness of the proposed technique.

Conclusions:

• The developed tensor decomposition-based image segmentation method offers a significant advancement in automated burn wound assessment.
• This technique improves both the accuracy and speed of burn area calculation, aiding clinical decision-making.
• Further research in automated burn assessment tools is crucial for advancing burn care and patient outcomes.