Visualization of dynamic subcutaneous vasomotor response by computer-assisted thermography

Two computer-assisted methodologies for the visualization of peripheral subcutaneous vasomotor events are described. The first approach, which utilizes a three-stage segmentation strategy based on edge detection, can be used to visualize temperature differences of approximately 3.5 degrees C between the subcutaneous vessel boundaries and surrounding tissue. The second approach requires user interaction with an adaptive filtering algorithm that selectively enhances vascular patterns in the thermogram while decreasing background noise artifacts. The user interactively selects decision thresholds used by the algorithm to develop symbolic, axiomatic models of homogeneous and bimodal local contrast regions. The result of this trained filter is then employed in a technique called digital subtraction thermographic venography for the extraction of subcutaneous venous patterns. This second approach shows less ambiguity and higher sensitivity than the edge detection approach in resolving subtle temperature differences of approximately 1.2 degrees C between the vessel and surrounding tissue. Computer-processed frames from both of these approaches are used for the dynamic visualization of normal and pathological vasomotor responses to thermal challenges, thereby providing diagnostic visual cues which are unavailable in the original thermograms.