Precise Tissue Segmentation in Orthopedic Surgery Using Hyperspectral Imaging

Advances in image-guided surgery increasingly depend on accurate intraoperative visualization of anatomical structures. In orthopedic procedures, distinguishing cartilage from bone is crucial to optimize surgical planning, implant positioning, and patient outcomes. Hyperspectral imaging (HSI), with its ability to capture rich spectral signatures, offers a powerful alternative to conventional imaging for detailed tissue characterization. In this study, we evaluate the use of HSI combined with machine learning for pixel-wise segmentation of bone, cartilage, and surrounding tissues. Experimental results show that a multilayer perceptron model achieved a segmentation accuracy of 97.03% while processing a full hyperspectral cube in just 3 seconds. These findings demonstrate the feasibility of realtime intraoperative guidance and highlight the potential of HSI-based AI solutions to enhance surgical precision and decision-making in orthopedic contexts.