Automated and Semi Automated Analysis of Simulated Wounds Using Image Processing Techniques
H Dhanush, Harish Vikram P, Monish Kumar R, J. B. Jeeva · 2025
Precise segmentation and assessment of artificial wounds are critical for medical diagnosis and treatment preparation. This research investigates four different segmentation techniques: the semi-automatic Local Graph Cut, manual region of interest (ROI) selection, and automatic methods like K-means clustering and morphological segmentation, all implemented using MATLAB. To emulate real wound features, artificial wounds were generated on synthetic tissue, and images were taken using a 12 MP smartphone camera positioned 30 centimeters above the wounds. A calibration factor of 0.0025 cm/pixel was applied to convert pixel-based measurements—area, perimeter, major axis length, and minor axis length—into centimeters. For the ROI-based Graph Cut technique, additional metrics such as average intensity and standard deviation were computed to improve precision. Findings show that the Graph Cut method, particularly with ROI input, provides accurate boundary segmentation, while K-means presents a fully automated method with slightly reduced accuracy.