Evaluation of stageⅠB cervical cancer by apparent diffusion coefficient histogram of MR diffusion weighted imaging
10.3760/cma.j.issn.1005-1201.2015.05.007
- VernacularTitle:采用MR扩散加权成像表观扩散系数值直方图诊断ⅠB期宫颈癌的价值
- Author:
Yuning LIN
;
Hui LI
;
Ziqian CHEN
;
Ping NI
;
Qun ZHONG
;
Ming MA
;
Shangwen XU
- Publication Type:Journal Article
- Keywords:
Cervical neoplasms;
Magnetic resonance imaging
- From:
Chinese Journal of Radiology
2015;(5):349-353
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the diagnostic value of ADC histogram obtained from MR DW imaging for stage ⅠB cervical cancer. Methods Seventy three patients diagnosed by cervical smear screening as cervical cancer without priortreatment were included prospectively in the patient group, and staged according to the international federation of gynecology and obstetrics (FIGO) staging system. Forty three patients with uterine leiomyoma detected by gynecologic examination, ultrasonography or CT and with negative result of cervical smear screening who were scheduled for hysterectomy were included prospectively in the control group. The patients of both groups underwent routine pelvic MR sequences, dynamic contrast enhanced imaging and DWI before hysterectomy. ADC histograms of the entire tumor and cervix volume were generated by post-processing software. Features of ADC histogram for the 2 groups were observed. Histogram parameters such as mean ADC (ADCmean), median ADC (ADCmedian), the 25th percentile of ADC (ADC_25th), the 75th percentile of ADC (ADC_75th), skewness and kurtosis were recorded. Student's t test or Mann-Whitney U test depending on homogeneity of variance was employed for the comparison of
those parameters. ROC analysis was employed for assessing the diagnostic performance of ADC histogram in distinguishing the 2 groups. Results Thirty five patients in the patient group were staged as FIGO IB. Five patients in the control group ended up with pathologic findings of cervical intraepithelial neoplasia grade 3. Therefore 38 patients in the control group were investigated. ADC histograms of the patient group were mostly skewed positively, while the curves were largely skewed negatively. ADCmean, ADCmedian, ADC_25th, ADC_75th, skewness and kurtosis for the IB stage patient group were (1.10±0.21)×10-3mm2/s, (1.05±0.21)× 10-3 mm2/s, (0.90 ± 0.19) × 10-3mm2/s, (1.26 ± 0.23) × 10-3mm2/s, 0.83 (median) and 1.25 (median) respectively. ADCmean, ADCmedian, ADC_25th, ADC_75th, skewness and kurtosis for the control group were (1.62 ± 0.25)×10-3mm2/s, (1.64±0.24)×10-3mm2/s, (1.42±0.24)×10-3mm2/s, (1.84±0.27)×10-3mm2/s,-0.11(median) and 0.29 (median) respectively. All parameters showed statistically different (t values were -9.693,- 11.117, -10.255, and -9.988 for ADCmean, ADCmedian, ADC_25th and ADC_75th respectively;Z values were -6.360 and -4.445 for skewness and kurtosis respectively; P< 0.01). ROC analysis indicated that ADCmedian had the highest diagnostic accuracy for differentiating the 2 groups, with the area under the curve being 0.97, a cutoff value of 1.21×10-3mm2/s, and a sensitivity of 95.6%and a specificity of 89.3%. Conclusion ADC histogram of DWI may be valuable for diagnosing stage IB cervical cancer by distinguishing stage IB cervical cancer from normal cervix or cervical benign lesions.