The Study Of The Clinical Value Of Texture Analysis Based On MRI Images To Judge The Degree Of Cervical Cancer Differentiation

Objective:To study the efficacy and clinical value of texture analysis of conventional MRI images in determining the degree of differentiation of cervical cancerMethods:Retrospective inclusion in the regular MRI examination of our hospital from December 2016 to December 2019,and 47 cases of cervical cancer?20 cases of low differentiation,27 cases of medium and high differentiation?confirmed by pathological results,and the real part of tumor on the multi-sequence MRI images sketched the tumor real part of the largest diameter level of the tumor as ROI,using Image J software,using histogram and grayscale symbiotic matrix?GLCM?texture analysis method,measures the mean,maximum,minimum value,standard deviation,kurtosis,skewness,modal grayscale value,ASM,contrast,deficit moment,entropy value,self-correlation and other texture parameters,and with postoperative pathological results for comparative analysis.The difference sytostic graph and grayscale symbiotic matrix parameters of the lower differentiation and medium-high differentiation of the MRI images were tested by independent sample t-test,Mann-Whitney U test.ROC curve is drawn on statistically significant parameters,and its effectiveness in predicting the degree of differentiation of cervical cancer is analyzed.Results:?1?Among the main histograms parameters of ADC images,the standard deviation of low-differentiated cervical cancer(SD_ADC)was 21.998?18.691,24.628?and the standard deviation of medium-differentiated cervical cancer was 15.994?13.918,20.833?.The standard deviation(SD_ADC)of low-differentiated cervical cancer(SD_ADC)is greater than that of medium-high-differentiated cervical cancer,and the difference between the two groups is statistically significant?Z value was-2.410,P respectively 0.016?.?2?In grayscale symbiotic matrix parameters,the contrast of the enhanced T1WI and T2WI sequences of low-differentiated cervical cancer(Con_CE-T1WI,Con_T2WI)was 160.717?109.615,198.950?,444.153±128.199,and medium-high-differentiated cervical cancer was 210.074?149.318,262.359?,554.993±166.978.Low-differentiated cervical cancer enhanced the contrast of T1WI,T2WI sequence(Con_CE-T1WI,Con_T2WI)is smaller than medium-high-differentiated cervical cancer,the difference between the two groups is statistically significant?t value were-2.238,-2.475,P respectively 0.025,0.017?.?3?Draw the ROC curves of Con_T2WI,Con_CE-T1WIE-T1WI and SD_ADCDC respectively,with the area?AUC?at 0.685,0.693 and0.707 respectively,and the combined Con_T2WI,Con_CE-T1WIE-T1WI and SD_ADCDC are the best in determining the effectiveness of cervical cancer differentiation,AUC reached 0.831.Conclusion:Texture analysis can provide multi-quantitative information,and can judge the degree of cervical cancer in a more accurate measure.Con_T2WI,Con_CE-T1WIE-T1WI and SD_ADCDC have better identification effectiveness for low-differentiation and medium-high differentiation cervical cancer,and better identification of joint parameters.