Medical visualization researchers have shown that wound regions segmented via active contour models can be modified manually before using automated camera calibration to measure wound size. A manual component to wound segmentation is useful in contexts where precise segmentation is ambiguous. For example, in applications that have varying requirements for visualizing differing wound tissues and the surrounding skin. To meet these intricacies, we propose using marching ants and a resizeable quick selection tool similar to what is available in Photoshop.

Before the end user interactively modifies the wound region, we use an automated un- supervised algorithm to provide a preliminary segmentation to work with. In so doing, we obtain the best of both worlds. Unsupervised wound segmentation quickly hones in on the target wound region, and quick selection allows for easy modification and customization. Once the wound is segmented from the image, the segmented wound image is stored for later retrieval such as would be required by the majority of the wound assessment apps discussed in this paper.