Open-Sourced Vessel Extraction Algorithm Using Structural Dissimilarity Mapping

Visualizing tumor blood vessels is critical for monitoring cancer, but current methods for angiography present following challenges. Standard intensity-based imaging techniques often require injected contrast agents for angiography, which can be invasive and cause side effects, while doppler functions of wave-based imaging techniques are angle-dependent. Additionally, other angiography technologies can involve radiation or have limited penetration depth. To address these issues, this study introduces a new structural dissimilarity (DSSIM) mapping algorithm. This method produces high-contrast, three-dimensional blood flow maps from a series of standard structural scans, removing the need for contrast agents. This approach was demonstrated by imaging a benign oral lesion. The resulting DSSIM map successfully extracted detailed microvascular features, such as increased vessel density surrounding an ulcer. This simple and effective technique provides valuable information about complex vascular networks from repeated intensity datasets. The DSSIM mapping method shows significant potential to enhance diagnostic and monitoring tools for cancer and other conditions across various medical imaging fields.