The Application Of Digital Software To Analyze Ki-67 Index In The Diagnosis Of Glioma And Meningioma

Objective:Through comparing the digital image analysis(DIA),visual evaluation method(VA),manual counting method(MA)three meningioma and glioma ki-67 index differences,and ki-67 results of differences between different subjects,to explore with AI methods advantages and disadvantages,and can replace manual counting or visual evaluation method for the diagnosis of clinical pathology doctors help,and for the interaction development of artificial intelligence and medical reference and provide new ideas.Methods:Sections of meningiomas and gliomas stained by immunohistochemical KI-67 in the Department of Pathology,the First Affiliated Hospital of Nanchang University from 2016 to 2019 were collected,including 54 meningiomas and 67 gliomas,which were grouped according to WHO classification and KI-67 values.The digital slice image was recorded by the digital slice scanner,the result of digital image analysis was detected by the AI analysis software,and the results were counted by visual assessment and manual counting respectively.Spss22.0 software and SPSSAU20.0 software were used for statistical analysis.Intergroup correlation coefficient was used to test the consistency of the results of each counting method among different subjects.Wilcoxon test of paired samples was used to compare the results of two brain tumors under digital image analysis,pathologist visual assessment and manual counting.Results:(1)In meningiomas and gliomas,the results of DIA and MA among different subjects were highly reproducible(ICC>0.900).The repeatability of using VA to evaluate the KI-67 index for meningioma was relatively high(ICC=0.762),while that of using VA to evaluate the KI-67 index for glioma was relatively high(ICC=0.719).(2)In WHOI,II and III meningiomas,the difference between VA and MA was not statistically significant(P>0.05),the differences between DIA and VA methods were statistically significant(P<0.05),the differences between DIA and MA were statistically significant(P<0.05).(3)In WHOI,II,III and IV gliomas,there was no statistically significant difference between VA and MA methods(P>0.05);In WHOI,II,and III gliomas,the differences between DIA and VA methods were statistically significant(P<0.05),the difference between DIA method and MA method was statistically significant(P<0.05);In WHOIV gliomas,the difference between DIA and VA was not statistically significant(P>0.05),there was no statistical difference between DIA method and MA method(P>0.05).(4)In meningiomas with low,medium and high KI-67 values,the difference between VA method and MA method was not statistically significant(P>0.05),the differences between DIA and VA methods were statistically significant(P<0.05);The difference between DIA and VA in low and medium KI-67 meningiomas was statistically significant(P<0.05);In meningiomas with high KI-67 values,there was no statistically significant difference between DIA and MA(P>0.05).(5)In gliomas with low,medium and high KI-67 values,the difference between VA method and MA method was not statistically significant(P>0.05);In low and medium KI-67 gliomas,the differences between DIA and VA methods were statistically significant(P<0.05),the difference between DIA method and MA method was statistically significant(P<0.05);In high KI-67 value gliomas,there was no statistically significant difference in the results of the three counting methods(P>0.05).Conclusion:(1)The results of digital image analysis(DIA)and manual counting(MA)between the two experimenters have a strong consistency.The results of visual assessment(VA)used by doctors with different seniority were consistent.(2)There was good consistency between the results of visual assessment(VA)and manual counting(MA)in the two brain tumors with different pathological grades and ki-67 numerical groups.Compared with the visual assessment method(VA)and manual counting method(MA),the results of digital image analysis method(DIA)under most groups are higher.The accuracy of digital image analysis method(DIA)is improved with the improvement of glioma pathological grades and ki-67 values of meningiomas and gliomas.