基于EMPIRE算法的数字乳腺断层合成X线成像结合重建二维及三维图像对乳腺良恶性病灶的鉴别诊断价值
|
摘要
目的:探讨基于EMPIRE算法的数字乳腺断层合成X线成像(DBT)结合重建二维(s2D)及重建三维(s3D)图像对乳腺良恶性病灶的鉴别诊断价值。方法:前瞻性纳入经超声诊断为乳腺影像报告和数据系统(BI-RADS)分类为4类及以上、拟行手术或穿刺活检的126例患者,所有患者均行全屏数字乳腺X线成像(FFDM)和DBT检查,通过DBT图像后处理获得s2D、s3D图像。分别对FFDM、DBT结合s2D、DBT结合s3D、DBT结合FFDM图像进行分析,记录腺体的平均辐射剂量,对病灶的可见性进行分级,同时对病灶进行BI-RADS分类;对穿刺及手术标本进行病理分析,获得病变的组织类型。采用单因素方差分析比较四种不同组合图像的平均腺体辐射剂量,采用配对符号秩和检验比较四种组合图像的可见性评级,以病理学为金标准,采用受试者工作特征(ROC)曲线比较四种组合图像的诊断效能。结果:4例为穿刺病理证实,122例为手术病理证实。良性病灶61例,恶性病灶65例。FFDM、DBT-FFDM、DBT-s2D、DBT-s3D图像的单个体位平均腺体辐射剂量分别为(1.72±0.48)、(4.81±1.12)、(3.28±0.79)和(3.28±0.79)mGy,其差异具有统计学意义(P<0.0001)。DBT-FFDM(116/126)、DBT-s2D(116/126)及DBT-s3D(117/126)较单独FFDM(98/126)显著提高病灶的可见性,其中DBT-s3D对病灶的可见性最高(93%)。FFDM、DBT-FFDM、DBT-s2D及DBT-s3D对乳腺良恶性病灶诊断的ROC曲线下面积分别为0.749,0.804,0.832,0.864。结论:基于EMPIRE迭代算法的DBT结合重建二维及三维图像能在一定程度上增加乳腺病灶的可见性,并提高乳腺良恶性病灶的鉴别诊断效能。
Objective:To explore the diagnostic performance of digital breast tomosynthesis(DBT) based on EMPIRE algorithm combined with synthesized two-dimensional(2 D) and three-dimensional(3 D) images in differentiating benign from malignant breast lesions.Methods:A cohort of 126 patients who were classified as breast imaging reporting and data system(BI-RADS) category 4 and above under ultrasound examination and underwent surgery or needle biopsy in our hospital between June 2016 and April 2017 were prospectively enrolled.All patients underwent full-field digital mammography(FFDM) and DBT examinations before surgery or biopsy.Synthesized 2 D(s2 D) and 3 D(s3 D) images were reconstructed using DBT data.The images with FFDM alone,DBT plus s2 D(DBT-s2 D),DBT plus s3 D(DBT-s3 D),DBT plus FFDM(DBT-FFDM) were analyzed in the average glandular dose,lesion visibility and lesion BI-RADS classification.The pathological results were obtained by surgery or needle biopsy after imaging examinations.One-way ANOVA analysis was used to compare the average glandular dose among these four groups.Wilcoxon matched-pairs signed ranks test was applied for the comparison of the lesion visibility.Taking pathological criteria as the gold standard,the diagnostic performance of the four different image combinations was compared by the receiving operating characteristic(ROC) analysis.Results:Four cases were pathologically confirmed by biopsy and 122 cases were pathologically confirmed by surgery.There were 61 benign lesions and 65 malignant lesions.The average glandular dose of FFDM,DBT-FFDM,DBT-s2 D,DBT-s3 D were(1.72±0.48)mGy,(4.81±1.12)mGy,(3.28±0.79)mGy,and(3.28±0.79)mGy.The difference was statistically significant(P<0.0001).DBT-FFDM(116/126,92.1%),DBT-s2 D(116/126,92.1%) and DBT-s3 D(117/126,93%) significantly improve the lesion visibility compared with FFDM alone(98/126,77.8%).The area under the ROC curves(AUC) of FFDM alone,DBT-FFDM,DBT-s2 D,DBT-s3 D were 0.749,0.804,0.832,and 0.864 respectively.Conclusion:DBT combined with reconstruction of 2 D and 3 D images based on EMPIRE iterative algorithm can increase the lesion visibility and improve the differential diagnosis of benign and malignant breast lesions.
Objective:To explore the diagnostic performance of digital breast tomosynthesis(DBT) based on EMPIRE algorithm combined with synthesized two-dimensional(2 D) and three-dimensional(3 D) images in differentiating benign from malignant breast lesions.Methods:A cohort of 126 patients who were classified as breast imaging reporting and data system(BI-RADS) category 4 and above under ultrasound examination and underwent surgery or needle biopsy in our hospital between June 2016 and April 2017 were prospectively enrolled.All patients underwent full-field digital mammography(FFDM) and DBT examinations before surgery or biopsy.Synthesized 2 D(s2 D) and 3 D(s3 D) images were reconstructed using DBT data.The images with FFDM alone,DBT plus s2 D(DBT-s2 D),DBT plus s3 D(DBT-s3 D),DBT plus FFDM(DBT-FFDM) were analyzed in the average glandular dose,lesion visibility and lesion BI-RADS classification.The pathological results were obtained by surgery or needle biopsy after imaging examinations.One-way ANOVA analysis was used to compare the average glandular dose among these four groups.Wilcoxon matched-pairs signed ranks test was applied for the comparison of the lesion visibility.Taking pathological criteria as the gold standard,the diagnostic performance of the four different image combinations was compared by the receiving operating characteristic(ROC) analysis.Results:Four cases were pathologically confirmed by biopsy and 122 cases were pathologically confirmed by surgery.There were 61 benign lesions and 65 malignant lesions.The average glandular dose of FFDM,DBT-FFDM,DBT-s2 D,DBT-s3 D were(1.72±0.48)mGy,(4.81±1.12)mGy,(3.28±0.79)mGy,and(3.28±0.79)mGy.The difference was statistically significant(P<0.0001).DBT-FFDM(116/126,92.1%),DBT-s2 D(116/126,92.1%) and DBT-s3 D(117/126,93%) significantly improve the lesion visibility compared with FFDM alone(98/126,77.8%).The area under the ROC curves(AUC) of FFDM alone,DBT-FFDM,DBT-s2 D,DBT-s3 D were 0.749,0.804,0.832,and 0.864 respectively.Conclusion:DBT combined with reconstruction of 2 D and 3 D images based on EMPIRE iterative algorithm can increase the lesion visibility and improve the differential diagnosis of benign and malignant breast lesions.
引文
[1] 杨蕾,宋俊峰,李静,等.数字乳腺断层融合X线成像中平均腺体剂量与乳腺密度、压迫厚度的关系[J].放射学实践,2015,5(14):494-497.
[2] 张云燕,顾雅佳,彭卫军,等.数字乳腺断层合成X线成像结合合成二维图像对乳腺疾病的诊断价值[J].中华放射学杂志,2016,50(11):833-837.
[3] Svahn TM,Houssami N,Sechopoulos I,et al.Review of radiation dose estimates in digital breast tomosynthesis relative to those in two-view full-field digital mammography[J].Breast,2015,24(2):93-99.
[4] Mertelmeier T,Orman J,Haerer W,et al.Optimizing filtered backprojection reconstruction for a breast tomosynthesis prototype device[J].Proceedings of SPIE-The International Society for Optical Engineering,2006,19(5):303-315.
[5] Rodriguez-Ruiz A,Teuwen J,Vreemann S,et al.New reconstruction algorithm for digital breast tomosynthesis:better image quality for humans and computers[J].Acta radiol,2018,59(9):1051-1059.
[6] Tani H,Uchiyama N,Machida M,et al.Assessing radiologist performance and microcalcifications visualization using combined 3D rotating mammogram (RM) and digital breast tomosynthesis(DBT)[M].Breast Imaging,Springer,International Publishing,2014:142-149.
[7] Gur D,Zuley ML,Anello MI,et al.Dose reduction in digital breast tomosynthesis (DBT) screening using synthetically reconstructed projection images:an observer performance study[J].Acad radiol,2012,19(2):166-171.
[8] Zuckerman SP,Conant EF,Keller BM,et al.Implementation of synthesized two-dimensional mammography in a population-based digital breast tomosynthesis screening program[J].Radiology,2016,281(3):730-736.
[9] Skaane P,Bandos AI,Eben EB,et al.Two-view digital breast tomosynthesis screening with synthetically reconstructed projection images:comparison with digital breast tomosynthesis with full-field digital mammographic images[J].Radiology,2014,271(3):655-663.
[10] Aujero MP,Gavenonis SC,Benjamin R,et al.Clinical performance of synthesized two-dimensional mammography combined with tomosynthesis in a large screening population[J].Radiology,2017,283(1):70-76.
[11] Choi JS,Han BK,Ko EY,et al.Comparison between two-dimensional synthetic mammography reconstructed from digital breast tomosynthesis and full-field digital mammography for the detection of T1 breast cancer[J].Eur radiol,2016,26(8):2538-2546.
[2] 张云燕,顾雅佳,彭卫军,等.数字乳腺断层合成X线成像结合合成二维图像对乳腺疾病的诊断价值[J].中华放射学杂志,2016,50(11):833-837.
[3] Svahn TM,Houssami N,Sechopoulos I,et al.Review of radiation dose estimates in digital breast tomosynthesis relative to those in two-view full-field digital mammography[J].Breast,2015,24(2):93-99.
[4] Mertelmeier T,Orman J,Haerer W,et al.Optimizing filtered backprojection reconstruction for a breast tomosynthesis prototype device[J].Proceedings of SPIE-The International Society for Optical Engineering,2006,19(5):303-315.
[5] Rodriguez-Ruiz A,Teuwen J,Vreemann S,et al.New reconstruction algorithm for digital breast tomosynthesis:better image quality for humans and computers[J].Acta radiol,2018,59(9):1051-1059.
[6] Tani H,Uchiyama N,Machida M,et al.Assessing radiologist performance and microcalcifications visualization using combined 3D rotating mammogram (RM) and digital breast tomosynthesis(DBT)[M].Breast Imaging,Springer,International Publishing,2014:142-149.
[7] Gur D,Zuley ML,Anello MI,et al.Dose reduction in digital breast tomosynthesis (DBT) screening using synthetically reconstructed projection images:an observer performance study[J].Acad radiol,2012,19(2):166-171.
[8] Zuckerman SP,Conant EF,Keller BM,et al.Implementation of synthesized two-dimensional mammography in a population-based digital breast tomosynthesis screening program[J].Radiology,2016,281(3):730-736.
[9] Skaane P,Bandos AI,Eben EB,et al.Two-view digital breast tomosynthesis screening with synthetically reconstructed projection images:comparison with digital breast tomosynthesis with full-field digital mammographic images[J].Radiology,2014,271(3):655-663.
[10] Aujero MP,Gavenonis SC,Benjamin R,et al.Clinical performance of synthesized two-dimensional mammography combined with tomosynthesis in a large screening population[J].Radiology,2017,283(1):70-76.
[11] Choi JS,Han BK,Ko EY,et al.Comparison between two-dimensional synthetic mammography reconstructed from digital breast tomosynthesis and full-field digital mammography for the detection of T1 breast cancer[J].Eur radiol,2016,26(8):2538-2546.