Hyperspectral Imaging Of Cerebral Ischemia And Glioma

Background and objective:It is important to real-time assess tissue perfusion and define tumor margin duringoperation in neurosurgery, which is beneficial to guide treatment, and improve prognosis.Anyway, it is difficult to meet this requirement for surgeon only with naked eyes or bymeans of current imaging techniques, such as CT, MRI, etc. Hyperspectral imagingtechnology could not only obtain a spatial image of measured object but also extractintrinsic signal by spectral analysis, which may become a new method for intraoperativereal-time guidance and precise excision.In order to investigate the spectral characteristics of diffuse reflection in the visiblenear-infrared range of biological tissue, this study measured the spectrum of varioussolution of different absorption and scattering properties and analyzed its variations. Inorder to find hyperspectral imaging method for identifying the ischemic brain tissue andglioma，this research measured the spectroscopy of ischemic brain tissue and glioma tissuesand analysed the spectral difference between them and normal brain tissue, and exploredhyperspectral imaging parameters to establish image processing algorithms of hyperspectralimaging. In total, it provides a foundation for clinical applications of hyperspectral imaging.Materials and methods:1. The reflectance spectroscopy in the visible near-infrared spectra of the blood andintralipid mixed solution was measured by fiber optic spectrometer, and the spectral curvevariation of different hemoglobin concentration, different oxygen saturation and differentintralipid concentration were analyzed.2. The rat model of right middle cerebral artery occlusion (MCAO) was establishedand the visible and near-infrared reflectance spectroscopy of normal and ischemic1h,3h,6h, 12h,24h brain tissue were measured by fiber optic spectrometer, and the spectraldifferences between ischemic and normal brain tissues were analyzed.3. Hyperspectral imager combined with surgical microscope system was used toexplore the imaging method of normal and ischemic brain tissues at1h,3h,6h,12h and24h in vitro, meanwhile, the spectra of normal and ischemic brain tissue were respectivelyextracted and the hyperspectral images were processed using principal component analysis(PCA) and spectral ratios algorithm with TTC staining and HE staining as control.4. The subcutaneous transplantation glioma model with C6, GL261and U87in nudemice and intracranial transplantation with GL261in C57mice were established. Thereflectance spectroscopy in the visible and near-infrared spectra of normal brain tissue,subcutaneous C6, GL261, U87glioma and intracranial GL261glioma and human glioma inoperation were measured in vivo by fiber optic spectrometers, and spectral differences wereanalyzed between glioma and normal brain tissues.5. Hyperspectral imager combined with surgical microscope system was used toexplore the imaging method of intracranial GL261glioma in vitro and in vivo. The differentspectra between normal and glioma tissues were extracted and hyperspectral images wereprocessed using spectral ratios algorithm with comparison to HE staining, MRI imaging andRGB imaging.Main results:1. The characteristical absorption for hemoglobin was at542nm and577nm. Thechanges of hemoglobin concentration and oxygen saturation could be reflected at500~600nm. While，the scattering properties of the sample could be indicated at700~900nm inspectral curve.2. The spectral characteristics at400~900nm between brain tissue in infarction areaand normal brain tissue were significantly different after1h of ischemia and with time goingon(3h,6h,12h,24h)，the difference is becoming more obvious.3. Hyperspectral images processed with principal component analysis (PCA) wereuseful and accurate to identify ischemia at1h,3h,6h,12h and24h, while, with spectralratio R545/R560processing could clearly showed the region of ischemia6h,12h and24h.Meanwhile, the calculating area of ischemic tissue on basis of R545/R560image (28.09±4.81,50.80±5.31,60.95±6.27mm2) and TTC staining (26.06±4.26,48.68± 4.31,60.29±5.96mm2) showed highly consistent.4. The spectral curve of subcutaneous transplantation glioma model with C6, GL261and U87, intracranial transplantation glioma model with GL261and human glioma tissuesin operation indicated significant difference with normal brain tissues at400~900nm.5. R700/R545hyperspectral imaging diagram could clearly indicated the tumor areaand glioma margin in vitro and in vivo. At the same time, R700/R545hyperspectral imagingcould effectively quantified tumor region(92.40±2.50%), which was higher than MRIT2(84.39±4.69%)and RGB imaging(naked eyes，81.93±4.47%)compared with HEstaining, respectively.Concusions:1. The visible near-infrared spectra could identify differences in tissue hemoglobinconcentration, oxygen saturation, and structural components, thereby effectively identifyischemic brain tissue and glioma tissue.2. Hyperspectral images processed with principal component analysis (PCA) couldidentify early brain ischemia. Hyperspectral images processed with spectral ratioR545/R560could accurately showed the region of ischemia. And, hyperspectral imagingwith spectral ratio R700/R545could effectively recognise the margin of glioma.3. Hyperspectral imaging could possibly become a new method for real-time, label freedetecting and imaging ischemic brain tissues and glioma tissues in vivo during operation inneurosurgery.