| Part 1 The relationship between breast type, exposure conditions and the detection rate of lesions on digital breast tomosynthesis[Objective]Most of the countries worldwide, the incidence rate of female breast cancer is growth and the incidence rate of female breast cancer in China is also significant growth. Breast cancer has become the first incidence rate of malignant tumor within Chinese women. Breast X-ray photography has been recognized as the preferred examination in breast cancer. In recent years, with the development of imaging equipment constantly updated, a new kind of photography technology known as Digital Breast Tomosynthesis (DBT), it appears that the breast cancer early detection and diagnosis has been further improved. In the present work, difficult to make optimized to the patient doses and lesions displayed in the same time.This part of the study aimed to investigate the dose in follow "in reasonably achievable levels as low as possible (ALARA)" on the basis of the principles used in manual mode, adjust the exposure conditions, for different types of breast lesions reference provides the optimized display, make the diagnosis of clinical breast disease more easily, and to promote the wide application of digital breast tomosynthesis the.[Materials and Methods]Human breast tissue specimens were selected from Nanfang Hospital of Southern Medical University, digital breast tomosynthesis examination in the Department of radiology and clinic diagnosed as breast cancer which is nesscery for breast modified radical mastectomy. The original lesion has been removed, the breast density classification of digital breast tomography in the preoperative has four kinds:fat type, a small amount of glandular type, large amount of glandular type and compat type. Then, use the USA standard breast imaging special detection ACR-156 model to combined with different breast types of human breast tissue samples in simulation, the simulation of breast tissue thickness in 42mm.Simulation of breast tissue composition containing a fibre-like lesions, calcification-like lesions and mass-like lesions of different sizes.1. Using Hologic Selenia Dimensions digital mammary machine, on four different anthropomorphic breast tissue combined phantom.First,use the auto exposure mode (AEC) exposure, kVp, mAs, AGD were recorded respectively. Radiography both choose tungsten target, aluminum filtration. In the same breast type, according to the AEC pattern exposure tube voltage obtained (kVp) value, through a decrease or increase the tube current (mAs) for manual exposure is not higher than the standard average glandular dose, selects the mAs at the time of AEC exposure values were 140%,130% 120%,110%,100%,90%,80%,70%,60%,50%, to obtain the same thickness in four breast type combination phantom, image and data in ten different exposure conditions. Image reading and evaluation of diagnostic workstation in Hologic, Barco 5M displays.2. Please three radiologists according to America Radiological Society (ACR) score method recommended. For each simulation lesions built in breast phantom, score from large to small according to the three system (score of 1,0.5,0), the average value of all kinds of simulated lesions final scored by three radiologists.3.Phantom lesion details description:(1)the fibro-like details can identify the fiber length, the position and direction for 1 points, can only identify the full-length half and the location of the correct direction for 0.5 points; (2)the calcification-like details can be neat identify more than 4 calcification scoring 1, can identify 2-3 calcification point 0.5; (3)the mass-like details can count 1 points in the line position identification and circumferential mass density difference boundary correct, a density difference but no peripheral boundary line 0.5.4. On four kinds of combination phantom different breast types, on ten different exposure conditions all image fiber lesions, calcification lesions and mass lesions were scored as image quality assessment, qualified standard to calculate the total detection rate that not less than 60%, the detection rate calculation formula:The total detection rate of lesion score= score of classification (fiber/calcification/mass) divided by the total score of classification (fiber/calcification/mass)5. Statistical analysisApplication SPSS20.0 software for analysis. Used univariate analysis of variance (One-Way ANOVA) for different breast type under different exposure conditions, the average glandular dose, the total detection rate of homogeneity to variance test. If the homogeneity of variance, under the same thickness, the sample mean of glandular dose and total detection rate of lesions on Paired t-test to FFDM and DBT, test level is set to 0.05, P<0.05 was considered statistically significant. The measurement data for different simulation breast types of exposure conditions and total detection rat of lesions using linear correlation and regression (Linear Regression) analysis.[Results]1. The relationship between the different simulation breast type exposure condit ions, average glandular dose and total detection rate of lesions.For the same simulated breast thickness, different breast types, manually increase and decrease the number of breast tomography exposure conditions, the measured average glandular dose, calculated the total detection rate of lesions. The exposure conditions and average glandular dose of compat breast type is maximum, the second is a large number of glandular type, the third is a small amount of glandular type, exposure conditions and average glandular dose of fat type is minimum on FFDM and DBT. And then, the sample mean of average glandular dose and total detection rate of lesions on FFDM less than the sample mean of average glandular dose and total detection rate of lesions on DBT.Under the same thickness, the sample mean of average glandular dose on DBT is (1.60±0.51) mGy, the sample mean of average glandular dose on FFDM is (1.37± 0.26) mGy, the difference for sample mean of average glandular dose on FFDM and DBT is statistically significant (t=18.354, P<0.01). the sample mean of DBT’s total detection rate of lesions is 80.57%±10.62%, the sample mean of FFDM’s total detection rate of lesions is 77.58%±9.23%, the difference for sample mean of total detection rate of lesions on FFDM and DBT is statistically significant (t=12.53, P <0.01).2. The linear correlation and regression analysis of different exposure conditions, breast type and the total detection rate of lesions on digital breast tomosynthesis (DBT).(1) Fat type under DBT exposure conditions and the total detection rate of lesions for the regression equation:The total detection rate of lesions (fat type)=0.459+0.007* exposure conditions (mAs)(2) Small glandular type under DBT exposure conditions and the total detection rate of lesions for regression equation:The total detection rate of lesions (small glandular type)=0.431+0.007* exposure conditions (mAs)(3) Large quantities of glandular type under DBT exposure conditions and the total detection rate of lesions for regression equation:The total detection rate of lesions (large quantities of glandular type)=0.477+0.006* exposure conditions (mAs)(4) Compact type under DBT exposure conditions and the total detection rate of lesions for regression equation:The detection rate of total lesions (compact type)=0.531+0.005* exposure conditions (mAs)The fat type, small glandular type, large quantities of glandular type were linear correlation between exposure conditions of digital breast tomosynthesis and the detection rate of lesions, the correlation is significant; the detection rate of lesions and exposure conditions of digital breast tomosynthesis were also exist linear correlation in compact type which have a high correlation and the regression equation of B values are greater than 0. Therefore, the study data of this group indicate that different simulation breast type of exposure conditions and the total detection rate of lesions had a linear relationship and positive correlation on DBT.[Conclusions]1. When the tube current (mAs) is within the range of 50%-100% for AEC exposure, increasing the tube current (mAs) can be improved to detection rate of different lesions with varying degrees among the different breast type in FFDM.2. For fat type and a small amount of glandular type, when the tube current (mAs) within the range of 100%-140% for AEC exposure, increasing the tube current (mAs) does not improve the detection rate of different lesions on DBT.3. For large amount of glandulars type and compat type, when the tube current (mAs) within the range of 50%-100% for AEC exposure, the increase in tube current (mAs) can improve the detection of lesions at different rates of FFDM and DBT.When tube current (mAs) is within the range of 100%-140% of AEC exposure, increasing the tube current (dose) dose not improve the detection rate of different lesions on FFDM.4. The same exposure conditions, detection rate of different lesions on DBT is more than FFDM.Part 2 The relationship between breast type, reconstruction slice thickness and the detection rate of lesions on digital breast tomosynthesis[Objective]The choice of reconstruction slice thickness in tomography images is special processing functions in digital breast tomosynthesis system, it can choose different reconstruction slice thickness in tomography images of breast lesions were observed and make the images to reach optimization. In the daily work, a part of the function is to the use of digital breast tomosynthesis essential for imaging diagnosis breast disease. In the current work, the reconstruction slice thickness of the image display optimization have not formed the attention.In this part of research, using of Hologic digital mammary machine combo mode with four different breast types of anthropomorphic phantom photography,3D tomographic images had reconstructed by different thickness in Hologic diagnostic workstation, explore relationship between the different reconstruction slice thickness and different breast types phantom display optimization in all kinds of lesions, provides the reference for the the clinical diagnosis of breast imaging.[Materials and Methods]Human breast tissue specimens were selected from Nanfang Hospital of Southern Medical University, digital breast tomosynthesis examination in the Department of radiology and clinic diagnosed as breast cancer which is nesscery for breast modified radical mastectomy. The original lesion has been removed, the breast density classification of digital breast tomography in the preoperative has four kinds:fat type, a small amount of glandular type, large amount of glandular type and compat type. Then, use the USA standard breast imaging special detection ACR-156 model to combined with different breast types of human breast tissue samples in simulation, the simulation of breast tissue thickness in 42mm. Simulation of breast tissue composition containing a fibre-like lesions, calcification-like lesions and mass-like lesions of different sizes.Secondly, using Hologic Selenia Dimensions digital mammary machine, on four different anthropomorphic breast tissue combined phantom, automatic exposure mode combo conditions are chosen, photography tungsten target, aluminum filtration,2D and 3D images obtained at the same time. Making different reconstruction slice thickness for the three-dimensional tomographic image, select the layer thickness 1mm,2mm,3mm,4mm,5mm,6mm,7mm,8mm,9mm,10mm to reconstruction respectively. Image reconstruction, reading and evaluation of diagnostic workstation in Hologic, Barco 5M displays.Third, please three radiologists base on America Radiological Society (ACR) score method recommended. According to the different slice thickness of the reconstructed image, combination of breast phantoms had each simulation lesions, scored from large to small according to the three system (score of 1,0.5,0). Each image was two scored. One without image adjusting, scored the original image on the display directly; Another may display parameter image of arbitrary regulation, including the regulation of brightness, contrast, window width and windowing, amplification, inversion that workstations can realize the function until observers believed to show the best state, the average scores value of all kinds of simulated lesion standarded for three doctors scoring finally.Fourth, phantom lesion details description:(1)the fibro-like details can identify the fiber length, the position and direction for 1 points, can only identify the full-length half and the location of the correct direction for 0.5 points; (2)the calcification-like details can be neat identify more than 4 calcification scoring 1, can identify 2-3 calcification point 0.5; (3)the mass-like details can count 1 points in the line position identification and circumferential mass density difference boundary correct, a density difference but no peripheral boundary line 0.5.Fifth, on four kinds of combination phantom different breast types, on ten reconstruction slice thickness all image fiber lesions, calcification lesions and mass lesions were scored as image quality assessment, qualified standard to calculate the total detection rate that not less than 60%, the detection rate calculation formula: The total detection rate of lesion score= score of classification (fiber/calcification/ mass) divided by the total score of classification (fiber/calcification/mass).Sixth, application of SPSS 20.0 software to carry on the analysis. Through single factor analysis of variance (One-Way ANOVA), in the same simulation breast type, the analysis different reconstruction slice thickness, average glandular dose, detection rate of lesion by homogeneity of variance test. If the analysis of homogeneity of variance, use the one-way Anova.If the heterogeneity of variance, use of multiple comparison for the sample mean.Different breast type of various disease score mean for multiple comparison (Kruskal Wallis test), P<0.05 considered statistically significant.[Results]1. The statistical results of different breast types and detection rate of lesions with different reconstruction slice thickness under the same simulation thickness.In fat type, under different reconstruction slice thickness that fiber-like lesions scores (4.89±0.47), calcification-like scores (4.16±0.44), mass-like scores (3.78±0.51), the total detection rate (80%±9%);In small amount of glandular type, under different reconstruction slice thickness that fiber-like lesions scores (4.26±0.80), calcification-like scores (4.01±0.58), mass-like scores (3.86±0.52), the total detection rate (76%±11%); In large quantities of glandular type, under different reconstruction slice thickness that fiber-like lesions scores (3.97±0.57), calcification-like scores (3.96±0.42), mass-like scores (3.48±0.60), the total detection rate (71%±10%); In compact type, under different reconstruction slice thickness that fiber-like lesions scores (4.05±0.67), calcification-like scores(3.85±0.58), mass-like scores (3.46±0.61), the total detection rate (72%±11%).2. The relationship between different breast types, reconstruction slice thickness and detected rate of lesions.The same simulation breast thickness, different simulation breast types lesions that calcification-like, mass-like, fiber-like scores and total detection rate had no significant difference (P>0.05), this may be due to the experiment sample size is not large enough and differences in mammary gland distribution become smaller. In the different reconstruction slice thickness, calcification-like lesions, fiber-like lesions, mass like lesions and the total detection rate of leisons were statistically significant (P value<0.01), indicating that the difference of different reconstruction slice thickness detection rate of lesion.Detection rate of lesions under different simulation of breast types varies with different reconstruction slice thickness. Different simulation breast type with the increase various kinds of the layer reconstruction thickness, the detected rate of leisons increased firstly and then decreased, the difference was statistically significant (P value<0.01). On the same simulation breast thickness, whatever the type of the breast, when the slice thickness is lmm, the highest detection rate is calcification-like lesions, the higher detection rate is fibrous lesions, the minimum detection rate is mass-like lesions. With the reconstruction slice thickness have increased, the change of detection rate in calcification lesions showed a trend of gradually decreased,the change of detection rate in fiber-like lesions and mass-like lesions showed a trend of increased at first,and then decreased.When the reconstruction slice thickness is 3-4mm, the detection rate of fiber-like lesions is highest. When the reconstruction slice thickness is 5-6mm, mass-like lesions reached the highest. For the different reconstruction slice thickness, the detection rate of lesions were more than 60% and reached the diagnostic requirements.[Conclusions]1. For calcified-like lesions, under four breast types, when the reconstruction slice thickness is 1mm, detection rate of lesions is highest. With the increasing of reconstruction silce thickness, the change trend of detection rate of calcified-like lesions gradually decreased.2. For fiber-like lesions, under four breast types, with increasing of reconstruction slice thickness, fiber-like detection rate of leisons increased at first and then decreased.When the reconstruction slice thickness is 3mm-4mm, the detection rate of fiber-like lesions is highest.3. For mass-like lesions, under four breast types, with increasing of reconstruction slice thickness, mass-like detection rate of leisons increased at first and then decreased.When the reconstruction slice thickness is 5mm-6mm, the detection rate of mass-like lesions is highest.4. With different reconstruction slice thickness, the detection rate of each lesions greater than 60%, reached the diagnostic requirements. Effects of detection rate of lesions by different reconstruction slice thickness could be due to the difference of simulated lesions that caused by size, density, shape and spatial distribution. Understanding these characteristics, can provide a reference for reading image in clinical diagnosis, improved diagnostic accuracy. |
PDF Full Text Request