ASSESSMENT OF ULCER WOUNDS USING 3D SKIN SURFACE IMAGING

In medical care, ulcer wound refers to open wound or sore in which certain conditions exist that impede healing. Nonhealing wounds can cause economical and psychological distress for patients. Wound size measurement (top area, true surface area, depth, and volume) is an objective indicator for wound healing. Top area measurement is useful for the follow up of shallow wounds, while true surface area if done accurately can work for all types of wounds. Calculating ulcer volume is crucial since studies showed that wounds start healing from the bottom. Overestimation in top area and true surface area measurement can be solved by digitizing the traced part. The objective of this research is to develop computer algorithms to measure ulcer wound size using 3D surface imaging. The wounds of interest are the wounds located at the leg. The algorithms should construct wound models and compute volume without getting affected by irregularities on wound surface and they should model leg curvature. Two algorithms for constructing wound models and volume computation are developed and evaluated; namely midpoint projection and convex hull approximation (Delaunay tetrahedralization). Parameters that describe the wounds are developed based on real ulcer wound surface images for wound modelling. Wound models representing possible ulcer wounds developed using AutoCAD software are used to investigate the performance of solid reconstruction methods. Results and analysis show that, for volume computation midpoint and convex hull methods can compute volume of leg ulcer without getting affected by irregularities in the healthy skin around the wound. The results show that, for convex hull low errors are produced in cases of regular boundary models excluding the elevated base models. Overestimation in volume for convex hull method can either be due to irregular boundary and/or elevation at the base (both global and local). Surface division is performed prior to convex hull approximation so that the high curvature of the leg and irregularity at the boundary can be represented using a number of linear segments. With the increase in surface division, error due to irregular boundary is reduced. In the case of global curvature, the reconstructed model using convex hull preceded by surface division simulates the leg curvature. Midpoint outperforms convex hull for models excluding elevated base models. Midpoint can construct solids for wound surfaces with local curvature while for surfaces with high global curvature the error is high. Midpoint method is not suitable for shallow and very large wounds.