脑缺血和脑胶质瘤的高光谱成像研究
发布时间：2018-08-25 07:41
【摘要】：背景与目的： 神经外科手术中实时评估脑组织的血液灌注状态、分辨恶性肿瘤的边界对指导手术、改善预后具有重要意义。而术者肉眼观察和目前的影像检查技术(如CT、MRI等)往往难以达到这一要求。高光谱成像技术不仅能实时获取所测物体的空间图像，而且能通过光谱分析提取物质的本征信号，可能成为实时引导手术、实现精准切除的新方法。 为了探讨生物组织在可见近红外波段的漫反射光谱特性，本研究测量了不同吸收和散射性质溶液的光谱，分析了其变化规律；为了探寻缺血脑组织及脑胶质瘤的高光谱成像方法，本课题测量了缺血脑组织和胶质瘤组织的特征光谱，分析了二者与正常脑组织光谱的差异，筛选了可有效辨别缺血脑组织及脑胶质瘤的高光谱成像参数，建立了图像处理算法。以期为高光谱成像的临床应用奠定基础。 材料与方法： 1.采用光纤光谱仪测量了血液和脂肪乳混合溶液的可见近红外漫反射光谱，分析了不同血红蛋白浓度、不同血氧饱和度及不同脂肪乳浓度的光谱曲线变化规律。 2.建立了SD大鼠右侧大脑中动脉阻塞(MCAO)模型，采用光纤光谱仪在体测量了正常及缺血1h、3h、6h、12h、24h脑组织的可见近红外漫反射光谱，分析了缺血脑组织与正常脑组织的特征性光谱差异。 3.采用高光谱成像仪结合手术显微镜对离体正常脑组织及缺血1h、3h、6h、12h、24h脑组织进行了光谱成像研究，提取了正常和缺血脑组织光谱，，分别用主成分分析(PCA)及光谱比值算法处理高光谱图像，并与TTC染色和HE染色进行对照。 4.建立了裸鼠皮下C6、GL261、U87胶质瘤移植模型，C57小鼠颅内GL261胶质瘤移植模型,采用光纤光谱仪在体测量了裸鼠正常脑组织、皮下C6、GL261、U87胶质瘤组织、C57小鼠颅内GL261胶质瘤组织及临床手术病人胶质瘤的可见近红外漫反射光谱，分析了胶质瘤组织与正常脑组织的特征性光谱差异。 5.采用高光谱成像仪结合手术显微镜对离体和在体颅内GL261胶质瘤组织进行了光谱成像研究，提取了正常脑组织和胶质瘤的光谱，用光谱比值算法处理高光谱图像，并与HE染色、MRI成像、RGB成像进行了比较。 主要结果： 1.血红蛋白在542nm和577nm波段具有特征的吸收峰，500~600nm波段光谱曲线的变化可反映血红蛋白浓度、血氧饱和度的变化；700~900nm波段光谱曲线可反映样本的散射性质。 2.缺血后1h开始，梗塞区脑组织在400~900nm波段的光谱特征与正常脑组织便存在明显差异，且缺血时间越长(3h、6h、12h、24h)差异越显著。 3.基于主成分分析的高光谱成像可有效识别缺血后1h、3h、6h、12h、24h脑组织；基于R545/R560光谱比值的高光谱成像可清晰显示缺血6h、12h、24h脑组织区域，计算面积(28.09±4.81、50.80±5.31、60.95±6.27mm2)与TTC染色(26.06±4.26、48.68±4.31、60.29±5.96mm2)高度吻合。 4.裸鼠皮下C6、GL261、U87胶质瘤，C57小鼠颅内GL261胶质瘤以及手术中人胶质瘤在400~900nm波段的光谱曲线均与正常脑组织存在明显差异。 5.基于R700/R545光谱比值的高光谱成像能清晰显示离体与在体颅内GL261胶质瘤区域，与HE染色显示的肿瘤区域比较，光谱比值R700/R545高光谱成像对肿瘤的识别精度最高(92.40±2.50%)、高于MRI T2成像(84.39±4.69%)和RGB成像(肉眼，81.93±4.47%)。 结论： 1.可见近红外漫反射光谱可鉴别不同组织血红蛋白浓度、血氧饱和度、结构成分等的差异，从而有效识别缺血脑组织与胶质瘤组织。 2.基于主成分分析的高光谱成像可探测脑组织的早期缺血；基于R545/R560光谱比值的高光谱成像可实现缺血区域的精确定位；基于R700/R545光谱比值的高光谱成像可有效辨别脑胶质瘤边界。 3.高光谱成像可能成为神经外科术中实时、无标记、在体探测与成像缺血脑组织和胶质瘤的新方法。
[Abstract]:Background and purpose:
In neurosurgery, it is important to evaluate the blood perfusion status of brain tissue in real time and to distinguish the boundary of malignant tumor for guiding surgery and improving prognosis. Images, which can also analyze the intrinsic signals of the extracts by spectroscopy, may become a new method for real-time guided surgery and precise excision.
In order to study the diffuse reflectance spectra of biological tissues in visible and near infrared bands, the spectra of different absorption and scattering properties of solutions were measured and analyzed. The spectral difference between ischemic brain tissues and normal brain tissues was analyzed. The hyperspectral imaging parameters which can effectively distinguish ischemic brain tissues and gliomas were selected. The image processing algorithm was established to lay a foundation for the clinical application of hyperspectral imaging.
Materials and methods:
1. The near-infrared diffuse reflectance spectra of mixed solution of blood and fat emulsion were measured by optical fiber spectrometer. The spectral curves of different hemoglobin concentration, different oxygen saturation and different fat emulsion concentration were analyzed.
2. The right middle cerebral artery occlusion (MCAO) model of SD rats was established. The visible near-infrared diffuse reflectance spectra of normal and ischemic brain tissues were measured in vivo by optical fiber spectrometer at 1, 3, 6, 12, and 24 h after ischemia. The characteristic spectral differences between ischemic brain tissues and normal brain tissues were analyzed.
3. Spectral imaging of normal and ischemic brain tissues in vitro for 1, 3, 6, 12, and 24 hours was studied with hyperspectral imager and operating microscope. Spectra of normal and ischemic brain tissues were extracted. The hyperspectral images were processed by principal component analysis (PCA) and spectral ratio algorithm, and compared with TTC staining and HE staining.
4. The models of subcutaneous C6, GL261, U87 glioma transplantation in nude mice and intracranial GL261 glioma transplantation in C57 mice were established. The visible near infrared diffuse reflectance spectra of normal brain tissue, subcutaneous C6, GL261, U87 glioma tissue, C57 mice intracranial GL261 glioma tissue and clinical operation patients were measured by optical fiber spectrometer in vivo. The difference of characteristic spectrum between glioma tissue and normal brain tissue.
5. Spectral imaging of GL261 glioma tissue in vitro and in vivo was studied with hyperspectral imager and operating microscope. Spectra of normal brain tissue and glioma were extracted. Hyperspectral images were processed with spectral ratio algorithm, and compared with HE staining, MRI imaging and RGB imaging.
Main results:
1. Hemoglobin has characteristic absorption peaks in 542 nm and 577 nm bands. The changes of spectral curves in 500-600 nm bands can reflect the changes of hemoglobin concentration and blood oxygen saturation, and the spectral curves in 700-900 nm bands can reflect the scattering properties of samples.
2. From 1 hour after ischemia, the spectral characteristics of the infarcted brain tissue in 400-900 nm band were significantly different from that of the normal brain tissue, and the longer the ischemic time (3h, 6h, 12h, 24h) was, the more significant the difference was.
3. The hyperspectral imaging based on principal component analysis can effectively identify the brain tissues at 1h, 3h, 6h, 12h and 24h after ischemia; hyperspectral imaging based on R545/R560 spectral ratio can clearly display the brain tissues at 6h, 12h and 24h after ischemia, and the calculated area (28.09, 50.80, 5.31, 60.95, 6.27mm2) and TTC staining (26.06, 48.68, 4.31, 60.29, 5.96mm2) were highly kissed. Close.
4. The spectral curves of subcutaneous C6, GL261 and U87 gliomas in nude mice, intracranial GL261 gliomas in C57 mice and human gliomas in operation at 400-900 nm were significantly different from those of normal brain tissues.
5. Hyperspectral imaging based on the spectral ratio of R700/R545 can clearly display the region of GL261 glioma in vitro and in vivo. Compared with the tumor region displayed by HE staining, the spectral ratio of R700/R545 hyperspectral imaging has the highest accuracy of tumor identification (92.40+2.50%), higher than MRI T2 imaging (84.39+4.69%) and RGB imaging (naked eye, 81.93+4.47%).
Conclusion:
1. Near infrared diffuse reflectance spectroscopy can distinguish the difference of hemoglobin concentration, oxygen saturation and structural components in different tissues, thus effectively identifying ischemic brain tissue and glioma tissue.
2. Hyperspectral imaging based on principal component analysis can detect early cerebral ischemia; hyperspectral imaging based on R545/R560 spectral ratio can accurately locate the ischemic region; hyperspectral imaging based on R700/R545 spectral ratio can effectively distinguish the boundary of glioma.
3. Hyperspectral imaging may become a new method for detecting and imaging ischemic brain tissues and gliomas in vivo in neurosurgery.
【学位授予单位】：第三军医大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R739.41
本文编号：2202218
[Abstract]:Background and purpose:
In neurosurgery, it is important to evaluate the blood perfusion status of brain tissue in real time and to distinguish the boundary of malignant tumor for guiding surgery and improving prognosis. Images, which can also analyze the intrinsic signals of the extracts by spectroscopy, may become a new method for real-time guided surgery and precise excision.
In order to study the diffuse reflectance spectra of biological tissues in visible and near infrared bands, the spectra of different absorption and scattering properties of solutions were measured and analyzed. The spectral difference between ischemic brain tissues and normal brain tissues was analyzed. The hyperspectral imaging parameters which can effectively distinguish ischemic brain tissues and gliomas were selected. The image processing algorithm was established to lay a foundation for the clinical application of hyperspectral imaging.
Materials and methods:
1. The near-infrared diffuse reflectance spectra of mixed solution of blood and fat emulsion were measured by optical fiber spectrometer. The spectral curves of different hemoglobin concentration, different oxygen saturation and different fat emulsion concentration were analyzed.
2. The right middle cerebral artery occlusion (MCAO) model of SD rats was established. The visible near-infrared diffuse reflectance spectra of normal and ischemic brain tissues were measured in vivo by optical fiber spectrometer at 1, 3, 6, 12, and 24 h after ischemia. The characteristic spectral differences between ischemic brain tissues and normal brain tissues were analyzed.
3. Spectral imaging of normal and ischemic brain tissues in vitro for 1, 3, 6, 12, and 24 hours was studied with hyperspectral imager and operating microscope. Spectra of normal and ischemic brain tissues were extracted. The hyperspectral images were processed by principal component analysis (PCA) and spectral ratio algorithm, and compared with TTC staining and HE staining.
4. The models of subcutaneous C6, GL261, U87 glioma transplantation in nude mice and intracranial GL261 glioma transplantation in C57 mice were established. The visible near infrared diffuse reflectance spectra of normal brain tissue, subcutaneous C6, GL261, U87 glioma tissue, C57 mice intracranial GL261 glioma tissue and clinical operation patients were measured by optical fiber spectrometer in vivo. The difference of characteristic spectrum between glioma tissue and normal brain tissue.
5. Spectral imaging of GL261 glioma tissue in vitro and in vivo was studied with hyperspectral imager and operating microscope. Spectra of normal brain tissue and glioma were extracted. Hyperspectral images were processed with spectral ratio algorithm, and compared with HE staining, MRI imaging and RGB imaging.
Main results:
1. Hemoglobin has characteristic absorption peaks in 542 nm and 577 nm bands. The changes of spectral curves in 500-600 nm bands can reflect the changes of hemoglobin concentration and blood oxygen saturation, and the spectral curves in 700-900 nm bands can reflect the scattering properties of samples.
2. From 1 hour after ischemia, the spectral characteristics of the infarcted brain tissue in 400-900 nm band were significantly different from that of the normal brain tissue, and the longer the ischemic time (3h, 6h, 12h, 24h) was, the more significant the difference was.
3. The hyperspectral imaging based on principal component analysis can effectively identify the brain tissues at 1h, 3h, 6h, 12h and 24h after ischemia; hyperspectral imaging based on R545/R560 spectral ratio can clearly display the brain tissues at 6h, 12h and 24h after ischemia, and the calculated area (28.09, 50.80, 5.31, 60.95, 6.27mm2) and TTC staining (26.06, 48.68, 4.31, 60.29, 5.96mm2) were highly kissed. Close.
4. The spectral curves of subcutaneous C6, GL261 and U87 gliomas in nude mice, intracranial GL261 gliomas in C57 mice and human gliomas in operation at 400-900 nm were significantly different from those of normal brain tissues.
5. Hyperspectral imaging based on the spectral ratio of R700/R545 can clearly display the region of GL261 glioma in vitro and in vivo. Compared with the tumor region displayed by HE staining, the spectral ratio of R700/R545 hyperspectral imaging has the highest accuracy of tumor identification (92.40+2.50%), higher than MRI T2 imaging (84.39+4.69%) and RGB imaging (naked eye, 81.93+4.47%).
Conclusion:
1. Near infrared diffuse reflectance spectroscopy can distinguish the difference of hemoglobin concentration, oxygen saturation and structural components in different tissues, thus effectively identifying ischemic brain tissue and glioma tissue.
2. Hyperspectral imaging based on principal component analysis can detect early cerebral ischemia; hyperspectral imaging based on R545/R560 spectral ratio can accurately locate the ischemic region; hyperspectral imaging based on R700/R545 spectral ratio can effectively distinguish the boundary of glioma.
3. Hyperspectral imaging may become a new method for detecting and imaging ischemic brain tissues and gliomas in vivo in neurosurgery.
【学位授予单位】：第三军医大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R739.41
【参考文献】
相关期刊论文 前1条
1 李庆利;薛永祺;刘治;;基于高光谱成像技术的中医舌象辅助诊断系统[J];生物医学工程学杂志;2008年02期
本文编号：2202218
本文链接：https://www.wllwen.com/yixuelunwen/shenjingyixue/2202218.html
最近更新
教材专著
