The Value Of Digital Breast Tomosynthesis (DBT) In The Diagnosis Of Breast Lesions And Radiation Dose-Related Research

Part I Evaluation on the ability of DBT to display simulated lesions of the breast by using ACR-156 ModelObjective To evaluate the ability of DBT to display simulated lesions of the breast by using ACR-156 Model.Materials and Method Different quantities of organic glass attenuation boards and ACR-156 model are selected to simulate the different equivalent thicknesses of breast oppression (19mm,28mm,37mm,47mm,55mm and 65mm). ComboHD model is adopted to shoot this model under the condition of automatic exposure control. According to the scoring method recommended by ACR, three evaluators score the fiber ribbon, micro-calcification and lump of ACR-156 model respectively. Value of Kendall’s W is applied to evaluate the consistency among different evaluators. Variance analysis is applied to analyze the scores of fiber ribbon, micro-calcification and lump of DBT,2D image which is composited by DBT and FFDM. It is concluded that LSD method should be adopted and P<0.05 indicates that the difference is statistically meaningful.Result (1) There is no statistical meaning between the three evaluators’scores (P>0.05). (2) Score comparison of fiber ribbon:Through comparing the scores of DBT and FFDM with different oppression thicknesses, it is concluded that the difference has no statistical meaning (P>0.05). When oppression thickness equals to 19mm,28mm,37mm or 47mm, there is no statistical meaning (P>0.05) as to the difference between composite 2D and FFDM. When oppression thickness equals to 55mm or 65mm, the score of composite 2D is lower than the score of FFDM, thus it has statistical meaning (P<0.05). (3) Score comparison of micro-calcification:Through comparing the scores of micro-calcification of DBT and FFDM with different oppression thicknesses, it is concluded that the difference has no statistical meaning (P>0.05). When oppression thickness equals to 19mm,28mm,37mm,47mm or 55mm, there is no statistical meaning (P>0.05) as to the difference between composite 2D and FFDM. When oppression thickness equals to 65mm, the score of composite 2D is lower than the score of FFDM, thus it has statistical meaning (P<0.05). (4) Score comparison of lump:When oppression thickness equals to 19mm or 28mm, the score of DBT is higher than the scores of FFDM and composite 2D respectively, thus this difference is statistically meaningful (P<0.05). When oppression thickness equals to 37mm,47mm or 55mm, DBT, FFDM and composite 2D have the same score. When oppression thickness equals to 65mm, the scores of DBT and FFDM are respectively higher than the score of composite 2D, thus this difference is statistically meaningful (P<0.05).Conclusion Under ACE control:(1) When the oppression thickness is relatively small, the ability of DBT to display lump is superior to the ability of FFDM. As the oppression thickness increases, the ability of DBT to display lump is equivalent to the ability of FFDM. (2) The ability of DBT to display fiber ribbon and micro-calcification is similar to the ability of FFDM. (3) When the oppression thickness is relatively small, the ability of composite 2D to display fiber ribbon, micro-calcification and lump is similar to the ability of FFDM. As the oppression thickness increases, the ability of composite 2D to display fiber ribbon, micro-calcification and lump becomes is inferior to the ability of FFDM.Part II The value of DBT in clinical diagnosis of breast lesionsObjective To evaluate the value of DBT in clinical diagnosis of breast lesions.Materials and Method 250 consecutive patients with °253 breast lesions are selected. All lesions are confirmed pathologily. All pationts has and who have had both DBT and FFDM exanimation before getting pathology. Meanwhile,212 patients with 212 breast lesions in total accept ultrasound examination simultaneously. Based on the gold standard of pathological diagnosis, FFDM, DBT, US, DBT combined with FFDM and DBT combined with US are diagnosed respectively. Category of BI-RADS 1～3 is defined as negative and category of BI-RADS 4～5 is defined as positive. McNemar test is adopted to compare the two imaging methods or sensitivity, specificity, coincidence rate, PPV and NPV among groups. ROC curve analysis imaging method is adopted to diagnose breast lesion and Z test is adopted to compare them. Kappa test is adopted to analyze the consistency between imaging diagnosis and final diagnosis. Pearson liner correlation is adopted to analyze the correlation between measured value of breast cancer range and pathological measured value on DBT and FFDM.Result (1) The Kappa coefficients of DBT, FFDM, DBT combined with FFDM are 0.783,0.664 and 0.789 respectively (P< 0.001). Areas under ROC curve (95% credibility interval) are 0.890 (0.884～0.925),0.833 (0.781～0.877) and 0.890 (0.844～ 0.925) respectively. The sensibility and coincidence rate of DBT and DBT combined with FFDM are both higher than FFDM’s (comparison of sensitivity P=0.001, comparison of coincidence rate P<0.001). The comparison difference has no statistical meaning at the aspects of specificity, PPV and NPV of FFDM, DBT and FFDM combined with DBT. (2) In the group with breast density> 50%, group with age≤50 and non-menopause group, areas under ROC curve of DBT and DBT combined with FFDM all larger than FFDM that the difference is statistically meaningful (P<0.05). In the group with breast density≤50%, group with age> 50 and menopause group, areas under ROC curve of DBT, FFDM and DBT combined with FFDM are compared by pairs and all differences have no statistical meaning (P>0.05). (3) The correlation coefficients of measured value of breast cancer range and pathological measured value on DBT and FFDM are 0.905 and 0.849 respectively (P<0.001). (4) Kappa coefficients of DBT, US and DBT combined with US are 0.761,0.774 and 0.761 respectively (P<0.001). Areas under ROC curve (95% credibility interval) are 0.884 (0.833～0.924),0.876 (0.824～ 0.917) and 0.865 (0.812～0.908) respectively. Through pair comparison of the areas under ROC curve, it finds out that their differences have no statistical meaning (P>0.05). (5) In the group with breast density> 50%, the difference of sensibility or specificity between DBT and US is not statistically meaningful. The sensibility of DBT combined with US is higher than DBT’s while its specificity is lower than DBT’s, thus their differences have statistical meaning (P=0.008). In the group with breast "density<50%, the sensibility or specificity of DBT, US or DBT combined with US is similar.Conclusion DBT has clinical diagnostic value on breast lesions:(1) In the subgroup with breast density> 50%, subgroup with age ≤50 and non-menopause subgroup, their diagnostic efficiency is superior to FFDM’s. (2) The measurement accuracy of DBT on breast cancers is superior to FFDM’s. (3) The efficiency of DBT to diagnose breast lesions is similar to US’s. When breast density> 50%, the sensitivity of DBT combined with US is superior to DBT’s with lower specificity.Part III Relationship between AGDDBT and breast density or oppression thicknessObjective To discuss the relationship between AGDDBT and breast density or oppression thickness.Materials and Method:271 female patients who had bilateral breast photography under the mode of DBT Combo (taking photos of the same breast and same oppression thickness with DBT and FFDM) in our hospital are collected.542 images of unilateral breast CC with DBT and 542 images of unilateral breast CC with FFDM are analyzed (AGD, breast density and oppression thickness). According to the percentage of glandular breast density, breasts were divided into four categories:<25%,25% to 50%, 51% to 75% and> 75%. Pair t test is adopted to compare AGD between two groups. Spearman rank correlation is adopted to respectively analyze the correlation between AGDDBT and breast density and the correlation between△AGD and breast density (AGDDBT minus AGDFFDM). Pearson correlation is adopted to analyze the correlation between AGDDBT and oppression thickness and the correlation between△AGD and oppression thickness. Multiple linear regression is adopted to analyze the influence of breast density and oppression thickness on AGDDBT and △AGD.Result (1) Oppression thickness and AGDDBT have a strong and positive correlation (r=0.920, P<0.001). Breast density and AGDDBT have a weak and positive correlation (r=0.263, P<0.001). Oppression thickness and breast density both can independently influence AGDDBT (B oppression thickness=0.046, B breast density=0.084, P< 0.001). Oppression thickness has a larger influence on AGDDBT than breast density (std B oppression thickness =0.94, std B breast density=0.143). (2) Range of △ AGD is -2.35～1.23 mGy. Oppression thickness and △ AGD have a weak and negative correlation (r=-0.287, P<0.001). Breast density and △ AGD have a medium-strong and negative correlation (r=-0.616, P< 0.001). Breast density and oppression thickness both can independently influence △ AGD (B oppression thickness=-0.012, B breast density=-0.219, P< 0.001). Breast density has a larger influence on △ AGD than oppression thickness (std B oppression thickness=-0.366, std B breast density=-0.580). (3) When breast density>75% and oppression thickness is 56mm～ 75 mm or 76～95 mm, the mean value of AGDDBT is lower than the mean value of AGDFFDM, thus difference has statistical meaning (P<0.05). When 25%≤breast density≤50% and oppression thickness is 76-95 mm, the mean value of AGDDBT is similar to the mean value of AGDFFDM (P> 0.05). When 50%<breast density≤75% and oppression thickness is 56-75 mm, the mean value of AGDDBT is similar to the mean value of AGDFFDM (P> 0.05). In other groups with different breast density and oppression thickness, the mean values of AGDDBT are all higher than the mean values of AGDFFDM, thus difference has statistical meaning (P<0.05).Conclusion (1) Breast density and oppression thickness are the independent influence factors of AGDDBT.AGDDTB have a positive correlation with breast density or oppression thickness. Meanwhile, oppression thickness has a larger influence on AGDDBT than breast density. (2) Breast density and oppression thickness are the independent influence factors of △AGD. △AGD have a negative correlation with breast density or oppression thickness. Meanwhile, breast density has a larger influence on △ AGD than oppression thickness. (3) According to this set of data, when breast density>75% and oppression thickness is 56-95 mm, the mean value of AGDDBT is lower than that of AGDFFDM.In other groups with different breast density and oppression thickness (mean oppression thickness≤95mm), the mean values of AGDDBT are all higher than or equal to that of AGDFFDM.