上海科技大学知识管理系统

Minimally invasive surgery is widely used for spine surgery, however the commonly used optical endoscopes cannot identity tissues under surface. In this study, a forward-oriented ultrasound endoscopic system was proposed to detect and identity different types of tissues for surgical approaches. A total of 150 ultrasound image data were collected from 6 types of intervertebral disc tissue using a custom-developed endoscopic probe. The gray-level co-occurrence matrix (GLCM) properties including energy (angular second moment, ASM), contrast, entropy, and homogeneity (inverse difference moment, IDM) were calculated on the acquired ultrasound images, and the single-parameter and combined parameter were applied for tissue classification. The classification accuracies of fibrous ring, nerve roots and bone were 100%, and the overall accuracy for all tissues was 73.33%. The results indicated that the combined parameter method provided more accurate classification output. It demonstrated that the proposed endoscopic ultrasonography system had the potential of identifying different tissues under surface during the endoscopic spine surgery.Clinical Relevance-This study establishes that the forward-oriented ultrasound endoscopic system was feasible to identify different types of tissues under surface during the endoscopic spine surgery. © 2023 IEEE.