摘　要：为了探索一种智能、无损检测黑色素瘤边界的方法，开发了一套具有紧凑结构和自动化软件的光纤光谱仪系统，并将其应用于皮肤色素沉积的鉴别。从样品中采集散射光谱后，用标准光谱进行相关测试，系统会根据相关系数自动给出判断。首先通过体外试验验证光纤光谱仪（FOS）系统快速智能检测色素沉积的可行性。用FOS系统对一个红、绿、蓝三色墨水染色的琼脂模型进行光谱测量，其鉴定准确率达96%。进一步用FOS系统检测墨水皮下注射构建的小鼠黑色素瘤模型，结果显示，墨水注射区与正常区光谱有显著差异。与标准墨水的光谱比对，墨水注射区域的相关系数为0.83±0.07，而正常组织的相关系数为0.18±0.05。从相关系数构建的图上可以清楚地识别出皮肤中墨水着色区域的边界。体内和体外试验都表明，FOS系统对皮肤沉积色素的检测具有较高的敏感性和特异性，预示着其在黑色素瘤的临床诊断中具有广阔的应用前景。
关键词：光纤光谱仪；活体；黑色素瘤；边界检测
中图分类号：R-33　　　　　　   文献标志码：A　　　　　　   DOI: 10.3969/j.issn.1007-7146.2020.06.005

Abstract: In order to explore an intelligent, non-destructive method for detecting the boundary of melanoma. In this study, a fiber optic spectrometer (FOS) system combined an integrated structure with an automatic software was developed and employed for distinguishing the skin pigmentation. After collecting scattering spectrum from sample, a correlation test will be performed with standard spectrum. The system will automatically give a judgment based on the correlation coefficient. The feasibility of the FOS system for rapid and intelligent detecting the pigmentation was first tested by an ex vivo experiment. A tricolor agar phantom stained by red, green and blue inks could be identified by the FOS system with 96% accuracy. Furthermore, a melanoma mouse model induced by subcutaneous injection of ink was examined by the FOS system in vivo. There were significant differences in spectrum between the ink-injection region and the normal region. Comparison with the standard spectrum of injected ink, the correlation coefficient of the ink injection area was 0.83±0.07, while the normal tissue was 0.18±0.05. The boundary of the ink pigmentation area in the skin could be clearly identified from the correlation coefficient map. Both the ex vivo and in vivo experiment demonstrated the FOS system could identify pigmentation with high sensitivity and specificity, and indicated its great potential to be used in the clinical diagnosis of melanoma.
Key words: fiber optic spectrometer; in vivo; melanoma; margin detecting
CLC number: R-33　　　　　 Document code: A　　　　　 DOI: 10.3969/j.issn.1007-7146.2020.06.005