Research On Quantitative Analysis Of Narrow-band Blood Vessel Image In V-PDT

Vascular targeted photodynamic therapy（V-PDT）has been widely used clinically because of its good target selection characteristics.However,clinical studies have shown that the therapeutic effect of vascular targeted photodynamic therapy not only differs significantly between individuals,but also in different areas of the same individual.In order to further accurately evaluate and predict the efficacy of V-PDT,this thesis uses computer image processing technology to quantitatively analyze the narrow-band blood vessel images collected during the V-PDT treatment of the dorsal skin windows chamber（DSWC）in vivo model of ICR mice.In order to overcome the difficult problem that traditional blood vessel segmentation methods are difficult to effectively segment small blood vessels in narrow-band blood vessel images.Therefore,this thesis proposed a set of methods for blood vessel segmentation,skeleton extraction and blood vessel quantification suitable for narrow-band blood vessel images.The main research work completed is as follows:（1）A judgment hybrid registration method is proposed.In view of the slight and severe deformation of the narrow-band blood vessel image in the V-PDT due to the mice’s breathing,heartbeat and involuntary movement.The slight and severe deformations were firstly distinguished according to the mean square difference（MSD）of registration groups.For slight deformation,the Log-Demons algorithm was directly adopted for registration of images.For severe deformation,SURF-TPS algorithm was utilized for global registration,and then the Log-Demons algorithm was used for realizing local registration.The experimental results show that the method proposed in this thesis not only has higher registration accuracy,but also has higher time processing efficiency.（2）A new method of blood vessel segmentation based on dual-scale fusion is proposed.Because of the high contrast between the wide blood vessel and the background,the traditional algorithm can extract the thick blood vessel of the narrow-band image before and after V-PDT.The small-scale Gaussian matched filter can enhance the thin vascular network in the narrowband light image of the V-PDT.For the pre-treatment image,the deformed Frangi model can further remove the background in the result of the small-scale Gaussian matched filter.For the image under treatment,At the previous moment,the background in the blood vessel binary image was registered and inversely mapped to further remove the background;then the edge of the blood vessel was extracted and the blood vessel binarization process was performed to obtain the thin blood vessel binary image;finally,the wide and thin blood vessel binary image was merged to realize the blood vessel segmentation.The experimental results show that the method proposed in this thesis can more completely extract the blood vessel network of the narrowband light image in V-PDT.（3）The quantification method of vascular skeleton and vascular damage in V-PDT is proposed.The thinning algorithm is utilized to extract the blood vessel skeleton based on the binary image of the blood vessel.On the basis of the binary image of the blood vessel and the blood vessel skeleton,the quantification of vascular diameter,vascular density and vascular fractal dimension in V-PDT is realized,and it will provide more accurate and comprehensive reference information for doctors to develop personalized and precise treatment plans in V-PDT.