Title: Quantitative evaluation of fibroglandular tissue for estimation of tissue- differentiated absorbed energy in breast tomosynthesis.
Other Titles: Kwantitative evaluatie van het fibroglandulaire weefsel voor de schatting van weefsel-gedifferencieerde energie in borst tomosynthese
Authors: Geeraert, Nausikaä; S0160423
Issue Date: 6-Oct-2014
Abstract: Breast cancer is an important disease, accounting worldwide for 25.2% of all female cancers and for 16% of cancer deaths in adult women. Early detection of suspicious lesions via breast screening increases survival chances. Mammography is the corner- stone of population-based breast cancer screening. The quality of screening tools is strictly controlled andcontinuously improved. A possible improvement of the current screening programs is an even more specific identification of the population at risk. Today, most screening programs invite women only based on and sometimes on patient or family history. Another improvement could be a better estimation of the associated radiation risk. For both improvements, breast density, a quantitative evaluation of fibroglandular tissue, differing in amount and distribution within the population, is a key parameter.Breast density has been identifiedas a risk factor for breast cancer by many studies. In addition an abundant dense tissue tends to mask suspicious lesions and reduces their detection. In this thesis we developed a volumetric breast density computation method in mammographic images, based on calibrating the image chain with breast-equivalent phantoms, and the acquisition data stored in the image-header. We applied and published a new validation method for breast density computation methods based on regular thorax CT images.In 1979 Hammerstein et al. stated that mammary gland is tissue at high risk for radiation damage, as opposed to skin, fat and connective tissue, which are not at high risk. They concluded with the proposition of total energy absorbed in glandular tissue as the most relevant indicator of riskin mammography. The average glandular dose (AGD), the currently accepted measure of mammographic dose, is not a function of the amount of tissue at risk, and as a consequence is not a good indicator for the radiation risk from a specific mammographic examination. In the individualized evaluation of the radiation related risk in mammography, the different radiation sensitivities of glandular and adipose tissues should be taken into account. This also requires the knowledge of the amount and localization of each of these components of the breast at an individual level. Thanks to methods as the volumetric breast density computation it is possible to assess to the amount of glandular tissue. Thanks to recent developments in breast imaging such as breast tomosynthesis it ispossible to partly overcome the problem of the glandular tissue localization. Hence, breast tomosynthesis reconstructs a 3D volume of the breast from a limited number of projections over a small angular range. In this thesis we apply a method to estimate the tissue-differentiated absorbed energy for a Senographe Essential configuration with the SenoClaire Digital Tomosynthesis attachment (GE Healthcare, Chalfont,UK). From the 0◦ projection image we computed the volumetric breast density. Contrary to CT scanning that produces Hounsfield units, the pixel values in reconstructed DBT volumes have no physical meaning andtherefore segmentation of the glandular versus adipose tissue cannot bereadily performed using simple thresholding techniques. We have proposed and worked out a procedure in which the volumetric breast density computation is used to label the local tissue, based on the conservation of glandular tissue percentage between the projection and the 3D volume. Wethen computed the locally imparted energy by Monte Carlo simulations applied to the reconstructed volume. Combining the labeled volume and the local imparted energy leads to the total imparted energy to the glandular and adipose tissues separately for the given breast tomosynthesis exam. For daily use more research is necessary to facilitate the calibrationof the system, the computation of locally imparted energy avoiding Monte Carlo simulations and an improvement of the reconstructed volume with for goal glandular segmentation.
Table of Contents: List of abbreviations vii
Abstract ix
Résumé xi
Présentation du projet xi
Densité volumique du sein xii
Etat de l’art xii
Nos contributions xv
Energie déposée dans la glande xviii
Etat de l’art xviii
Nos contributions xix
Conclusion xxii

Samenvatting xxv
Probleemstelling xxv
Volumetrische borstdensiteit xxvi
State of the art xxvi
Onze bijdragen xxix
Energie geabsorbeerd door het klierweefsel xxxii
State of the art xxxii
Onze bijdragen xxxiii
Conclusie xxxvii

1 Introduction 5
1.1 General problem statement 5
1.2 State of the art 6
1.2.1 Breast anatomy and histology 6
1.2.2 Breast density 7
1.2.3 Breast dose 12
1.2.4 Conclusion 16
1.3 Research objectives 17

2 Validation of breast equivalent phantom material 23
2.1 Breast equivalent material for VBD computation 23
2.2 CT measurements of the breast equivalent phantom 24 2.2.1 CIRS 24
2.2.2 Configurations 25
2.2.3 Measuring the CIRS characteristics 26
2.3 Breast equivalent phantom characteristics 27
2.4 Breast equivalent phantom in mammography 30
2.5 Conclusion 32
3 Breast density computation 33
3.1 Volumetric breast density: introduction 33
3.2 VBD: from theory to practice 33
3.2.1 Theoretical model for the computation of V BDMX from digital mammographic images 33
3.2.2 Implementation of the V BD computation 36
3.2.3 Application to mammographic images 40
3.3 Calibration of the VBD computation 43
3.3.1 VBD for phantoms 43
3.3.2 VBDMX for the database of mammographic images 49
3.4 Conclusion: VBD computation is possible 52
4 New validation of the breast density computation 57
4.1 CT versus mammography 57
4.2 VBD computation in CT 57
4.2.1 Theoretical derivation 57
4.2.2 Database of mammographic and CT images 59
4.3 VBDCT versus VBDMX 59
4.3.1 Calibration of the CT method 59
4.3.2 VBDCT for the databases 60
4.3.3 Correlation between V BDCT and V BDMX 61
4.4 Same breast, same VBD 63
Discussion Part I 65

5 Validation of the Monte-Carlo simulation tool CatDose 71 5.1 Monte-Carlo simulations 71
5.2 TG 195 AAPM manual 72
5.3 Computation of dose conversion factors 73
5.3.1 Dance conversion factor 73
5.3.2 Boone conversion factors 75
5.4 Limits and strengths 76
6 Evaluation of irradiation in mammography 79
6.1 Dosimetry for individuals 79
6.2 Use the right density 82
6.3 New quantity used for individualized risk 83
6.4 Computation of the local GIE 86
6.5 Discussion and conclusion on individual risk assessment 89
7 Segmentation of glandular tissue from tomosynthesis 93 7.1 Brief introduction to tomosynthesis 93
7.2 Tomosynthesis limitations 94
7.3 Segmentation method 96
7.4 Reconstructions 99
7.5 Textured phantom 105
7.6 Real patient cases 107
7.7 Conclusion on tomosynthesis segmentation 109
Discussion Part II 111
8 General conclusions and perspectives 113

A Breast statistics 115
B Level estimators 119
List of publications 121
Bibliography 122
Publication status: published
KU Leuven publication type: TH
Appears in Collections:Medical Physics & Quality Assessment (+)

Files in This Item:
File Status SizeFormat
PhD2014_NGeeraert.pdf Published 21193KbAdobe PDFView/Open


All items in Lirias are protected by copyright, with all rights reserved.