| Objective:multi-parameters MRI(mpMRI)is currently recognized as the most accurate noninvasive diagnostic method for prostate cancer.The ADC map is recommended by the second version of the prostate imaging reporting and data system as the most important sequence for the identification of lesions.However,the use of quantitative analysis of ADC diagrams has not yet reached a unified conclusion.As we know,pathological section is the gold standard for the diagnosis of prostate cancer.However,the whole organ slices contain more detailed information about the shape,location and size of the lesion.The aim of the present study is to explore the precise matching process of whole prostate organ slices and MRI by taking the whole body slice of prostate as the gold standard for image recognition.Besides,the relationship between ADC value and the possibility of extracapsular invasion of suspected lesions is also analyzed.Furthermore,we will explore the relationship between the ISUP group and the pathological group by Extracting the characteristics of different ADC graphs.Methods:we randomly selected two groups of specimens.According to the conventional method and IUNU method,the sampling and registration were finished respectively.Then,the number and size of tissue block obtained from two groups of specimens were compared.Besides,the similarity between the final organ pathology and the corresponding magnetic resonance images in the prostate parameters was compared.Thus,the advantages and disadvantages of the two methods in matching pathology and imaging were evaluated.After this,a group of specimens were assigned and matched according to the modified IUNU method.Then,delineate the slice focus and ADC map to select the lesions that were contacted with the prostatic capsule without obvious external invasion.Finally,extract the ADC value in the region of interest of its corresponding ADC graph to explore the correlation between ADC value and the positive pathological area of suspicious external invasion.The next stage is to extract all markers or clinically significant tumor foci and the ADC value of the region of interest of the ADC map corresponding to the non-cancer region.After this,the ADC values of regions of interest were classified according to different percentile intervals.And the histogram parameters corresponding to the region of interest of each lesion including the summit value,four division interval value were extracted.This stage was to explore the relationship between ISUP grouping and lesion groupingResults:176 tissue blocks were obtained by conventional methods,and 198 tissue blocks were obtained by IUNU method.The average thickness of the tissue block obtained is thinner and more uniform than that of the conventional method(2.98±0.247mm vs.3.64±0.49mm,P<0.0001).The skew of the section is smaller(1.252±1.080mm vs.0.582±0.521mm,P<0.001).Compared with the routine method,there were 92 pathology sections which are matched to different mpMRI images by two different methods of all 374 whole-mount histopathology.The final matching scores of the two methods were 3.250 vs 4.129(P<0.001).A total of 49 peripheral zones of interest were included in the study.Among them,29 cases(16 cases of external invasion of the capsule confirmed by pathology,and 13 were not violated).Then,ROC curve revealed that the area under the curve was 0.745,and the sensitivity and specificity values were 0.625 and 0.769 respectively.When drawing ROI with the amount of pixels more than 40,the area under the curve would be 0.9.Once the cut-off value was 61.8,the maximum sensitivity and specificity were 0.867 and 0.875 respectively.While drawing ROI with the amount of pixels less than 40,the incidence of external invasion was only 16.7%.In the comparison of ADC value and ISUP grouping,a total of 72 lesions(49 in the peripheral zone and 23 in the shift belt)were included.Two factor analysis of variance analysis revealed that there is no interaction in the impact on the ADC value between location of lesion and ISUP grouping corresponding to lesion.F(4,64)=1.394,P=0.246.However,the main effect analysis of both ADC values was statistically significant.Flocation(1,64)=8.921,P=0.04;FISUP(2,64)=6.659,P=0.000.The percentile intervals of all ROI ADC values in the 94.5%peripheral zone was significant different from pathological ISUP groups.Among the interval of 50%-60%,the correlation between ADC mean and ISUP grouping is the highest(r=-0.340,p=0.003).The percentile intervals of all ROI ADC values in the 96.2%transitional zone was significant different from pathological ISUP grouping.Among the interval of 0%-40%,between ADC mean and ISUP grouping is the highest(r=-0.714,p=0.00013).In the four interval of the histogram of the ADC value of the peripheral zone lesion,the significances values of ADC1/4,ADCmiddle and ADC3/4 were 0.024,0.069 and 0.097 respectively.In the transitional zone,the significances values of ADC1/4,ADCmiddle and ADC3/4 were 0.001,0.009 and 0.084 respectively.The summit values of the ISUP 1-5 groups in the peripheral zone were 68.75±9.2,60.00±3.33,52.73±7.54,53.6364±11.42 and 51.25±8.54 respectively.F(4,39)=0.01<0.05,the intergroup analysis indicated that there was statistical difference between ISUP grade 1 and ISUP grade 3/4/5,the p values were 0.001,0.005 and 0.003 respectively.According to the summit value ADC value corresponding to ROI and pathological ISUP grade which were equal to 1 or more than 2 to draw ROC curve.Then,the area under the curve would be 0888.Once the cut-off value was 62.5,the maximum value of sensitivity and specificity was 1.691(0.875+0.816).The summit values of ISUP 1-5 at the transitional zone were 88.33±7.64,73.13±15.80,56.25±6.29,55±0.00 and 55±5.00 respectively.F(4,17)=0.041<0.05,the analysis indicated that there was statistical difference between ISUP 1 and ISUP 3/4/5 and ISUP 2 and ISUP3/4/5 level.The p values were 0.011,0.007,0.025,0.013 and 0.031,the under the curve would be 0.904.Once the cut-off value was 75,the maximum value of sensitivity and specificity would be 1.895(1+0.895).Conclusions:The IUNU method has significantly advantages in matching hispathology section to mpMRI images.We can improve the alignment after introduced this method into the future work.When the suspicious relationship between lesions and prostatic pseudo capsule on ADC map is suspicious,the possibility of external invasion of the lesion is less if the lesion is smaller(pixels<40).On the contrary,it is likely that the invasion of the capsule is more likely to occur if the lesion is bigger(pixels>61.8).The ADC value of the lesion has a certain correlation with the ISUP grouping of the corresponding lesion.Furthermore,no parameters that can effectively distinguish all ISUP grouping are found while the summit value using ADC histogram has a good degree of discrimination.If the lesion is located in the peripheral zone and the corresponding summit value is greater than 62.5,then the lesion is more likely to be a ISUP 1 group.Similarly,if the lesion is located in the transitional zone and the corresponding the summit value is greater than 75,than the possibility of this focus is also the ISUP 1 group. |