Case B: AI-Based Enhancement and Analysis of Laser Speckle Contrast Imaging for Intraoperative Tissue Perfusion Assessment

Background: Laser speckle contrast imaging (LSCI) is a non-invasive optical imaging technique used to assess tissue perfusion by visualizing blood flow dynamics in real time. The method relies on the speckle pattern generated when coherent laser light scatters from moving red blood cells. Increased motion of blood cells leads to greater blurring of the speckle pattern, which corresponds to higher tissue perfusion. By analyzing spatial or temporal variations in speckle contrast, LSCI produces high-resolution, wide-field maps of relative blood flow.

Objective: LSCI has significant potential for intraoperative use, as it can be integrated into existing laparoscopic video systems and surgical workflows. This integration enables surgeons to receive real-time visual feedback on tissue perfusion during surgical procedures. Such information may support better clinical decision-making. Despite these advantages, several challenges limit the widespread clinical adoption of LSCI. First, LSCI images are acquired using red laser illumination, resulting in monochromatic images that make anatomical interpretation and surgical navigation difficult. Second, LSCI data often suffer from artifacts such as specular reflections, motion blur, and occlusions caused by surgical instruments, which can obscure perfusion maps. Third, the lack of clear anatomical boundaries in red laser images makes it difficult for surgeons to assess perfusion levels of specific organs or tissue regions. This research aims to address these challenges through the development of AI-based methods for image colorization, artifact removal, and anatomical segmentation in LSCI data.