The Comparative Study Of Bi-exponential Model And Traditional Mono-exponential Apparent Diffusion Coefficient In The Diagnostic Performance Of Invasive Ductal Carcinoma

Objective To explore the value of Bi-exponential model and traditional Mono-exponentialapparent diffusion coefficient in the diagnosis of invasive ductal carcinoma and try to find a bestdiagnostic cutoff.Methods Retrospectively analyze the findings of35patients with44lesions of invasiveductal carcinoma and17patients benign tumors with24lesions diagnosed by pathology or otherways. At the same time, selected30healthy women as control group. All patients underwent aroutine MRI examination and traditional mono-exponential model MR DWI、IVIM technologybi-exponential model of multi-b-value DWI examination. The ADC, fast ADC and slow ADC ofthree groups were obtained. Using SPSS17.0and MedCalc software for statistical analysis. Theone-way ANOVA and independent-samples t test was used for comparison. Draw the ROC curveand try to find the best cutoff. Calculate the sensitivity, specificity and accuracy of ADC and slowADC. To compare ADC and slow ADC area under the ROC curve between benign and malignantlesions with Z test.Results1. The one-way ANOVA results showed that the ADC and slow ADC hadstatistically significance, fast ADC had no statistically significance among groups.2. The ADC value of invasive ductal carcinoma was (0.989±0.145)×10-3mm2/s, the slowADC value was (0.375±0.130)×10-3mm2/s, this had statistical significance compared withnormal mammary and benign lesions(P <0.01).3. The area under the ROC curve of benign and malignant lesions were0.893and0.929respectively. The best diagnostic cutoff was1.06×10-3mm2/s and0.475×10-3mm2/s. The diagnostic sensitivity, specificity and accuracy were87.5%and91.7%,77.3%and81.8%,77.3%and86.4%respectively.4. The difference between the two ROC curve of ADC and slow ADC had no statisticalsignificance (Z=0.764, P=0.445).Conclusion The diagnostic value in the diagnosis of invasive ductal carcinoma ofslow ADC based on IVIM technology of multi-b-value bi-exponential model and traditional ADCbased on mono-exponential model was higher, but compared with the traditional ADC based onthe mono-exponential model, the diagnostic sensitivity, specificity and accuracy of the slow ADCwere all improved.