复合染色方法可以解决传统宫颈癌联合筛查无法在同一张涂片上实现的问题。针对复合染色带来的DNA吸光度干扰和色彩还原失真等问题, 本文建立了一套多波段光源的显微光谱成像系统.组合光源的设计兼顾彩色图像的还原和吸光度模型的建立, 采用多元线性回归法建立多光谱剥离模型, 利用该模型从混叠的多光谱图像中剥离出福尔根染料的真实吸光度用于DNA定量分析, 同时合成符合医生诊断习惯和要求的伪彩色图像. 实验结果表明: 选用的四个波段的光源的带宽为12 nm/12 nm/14 nm/19 nm, 是符合要求的窄带单色光；以显微测微尺为标尺, 以原光源测量参数为标准, 测得组合光源的光斑偏移量最大不超过3个刻度, 其平行度和同轴度满足成像系统要求；合成的伪彩色图像颜色明亮饱满、背景干净, 不会对阅片医生诊断造成干扰和障碍；吸光度剥离模型可以成功将福尔根染料真实吸光度剥离出, 能够准确计算DNA含量.选用小鼠肝脏涂片作为测量标本, 测量计算的DI值符合肝细胞多倍体性的特点；选择宫颈阳性样本进行对比实验, 两种方法测得的二倍体细胞和四倍体细胞DI值相吻合.本系统具有较高的准确性和可靠性.

Compound staining method can solve the problem that traditional combination screening of cervical cancer can not be carried out on the same smear. In view of the problems of DNA absorbance interference and color restoration distortion caused by compound staining, a microscopic spectral imaging system with multi-band illuminants was established in this paper. The design of composite illuminants gave consideration to both the restoration of color image and the establishment of absorbance model, multi-spectral unmixing model were constructed by multiple linear regression method, and by using this modle Feulgen′s real absorbance were unmixed from the mixing muti-spectral images for DNA quantitative analysis , and pseudo color images were also composed to meet the doctors′ requirements and habits at the same time. Experimental results show that: the Full Width at Half Maximum (FWHM) of the chosed four-band illuminants were 12 nm/12 nm/14 nm/19 nm, and they were satisfactory narrowband monochromatic lights; The measured maximum offset of spot does not exceed 3 scale by using a microscope micrometer as a ruler and the original illuminant as a standard, so parallelism and coaxiality of the composite illuminants met the requirements of imaging system; The synthesis pseudo color images were bright and full , and have clean background, they will not cause interference and obstacles to doctor′s diagnosis.The absorbance unmixing model can successfully unmixing the real absorbance of feulgen and accurately calculate the content of DNA: The liver smear of mice was used as the measuring specimen and the calculated DNA index (DI) values accorded with the polyploidy of liver cells; In contrast experiments using a cervical positive sample, the DI values of the diploid and tetraploid cells calculated by the two methods were almost equal, therefore, this method has higher accuracy and reliability.