Détection automatique des signes radiologiques par la mammographie numérique tridimensionnelle.

Accueil || Parcours || Recherche || S'enregistrer || Mon Compte || Contacts || Aide || Langues

Peters, Gero (2007) Détection automatique des signes radiologiques par la mammographie numérique tridimensionnelle. Doctorat Signal et Image, ENST - TSI Traitement du Signal et des Images, ENST p.215.

Plein texte disponible en tant que : - PhD_GP.pdf ( 13305 Kb )

Licence: Copyright

Résumé

Digital Breast Tomosynthesis (DBT) is a new three-dimensional limited-angle tomography breast imaging technique that will substantially overcome the superimposition problem for lesion detection in digital mammography. This work focuses on developing an automated detection scheme for the very recent images obtained with DBT.

Since DBT is an emerging modality, clinical data sets are still rare, and few works have been published in the domain of computer-aided detection for these data sets. The three dimensional nature of the data offers a number of advantages. At the same time, issues like a reduced patient dose per scan, the limited-angle acquisition geometry and the resulting reconstruction artifacts, as well as the problem of an adapted reconstruction technique need to be addressed.

We develop an algorithm for reconstruction-independent computer-aided detection on DBT data. Working directly on the tomographic projection images, we make use of fuzzy set theory to preserve the ambiguity present in the images until the information from the entire set of projected views can be jointly considered. A framework is proposed that allows for detection of different kinds of radiological findings using the same general framework.

We present a method for fuzzy contour extraction for mass lesions. The originality of our method lies in the use of a set of dedicated hybrid active contour models. Each model is constructed using a priori knowledge about a class of objects of interest. For each radiological finding, we apply the different active contour models resulting in a set of fuzzy contours for each object of interest. A feature vector is extracted for each fuzzy contour. Using the extension principle we establish a set of fuzzy attributes based on this feature vector.

We propose an aggregation strategy for combining the information corresponding to a given three-dimensional object over the entire set of projected views. To this end we introduce a partial defuzzification operation that provides a pixel-based representation of the fuzzy information extracted for each object. This operation, combined with a back-projection/re-projection step, provides the links between the particles in the projection images, that correspond to the same three-dimensional particle. Cumulated fuzzy attributes and a confidence degree for each particle are computed. A fuzzy decision tree working on fuzzy data is then applied to obtain a final decision.

The presented approach has been validated on a number of simulated images as well as on clinical images. We show that the algorithm is capable of distinguishing between different radiological signs. The availability of a larger clinical database will allow a better quantitative evaluation of the algorithm performance in a near future.

Table des Matières

Acknowledgements v

Table of Contents vii

List of Figures x

List of Tables xvi

Abstract xix

Résumé xxi

Synthèse xxiii

1 Introduction 1

2 Context and Motivation 5

2.1 The Breast - 5

2.1.1 Anatomy of the Breast - 5

2.1.2 Breast Cancer - 7

2.2 Breast Imaging - 8

2.2.1 Mammography - 8

2.2.2 Digital Mammography - 11

2.2.3 Digital Breast Tomosynthesis - 12

2.2.4 Other Breast Imaging techniques - 18

2.3 Computer-Aided Detection - 18

2.3.1 CAD for 2D Mammography - 19

2.3.2 CAD for DBT - 24

3 Extracting Image Information 29

3.1 Introduction - 29

3.2 Markers - 30

3.2.1 Candidate Calcification Detection - 30

3.2.2 Candidate Mass Detection - 32

3.3 Segmentation - 36

viii Table of Contents

3.3.1 Active Contour Models - 38

3.3.2 Experiments and Results - 44

3.3.3 Fuzzy Contour Extraction - 51

3.3.4 A New Class of Fuzzy Contours: Fuzzy Active Contours - 55

3.4 Feature Extraction - 63

3.4.1 Extraction of 2D Attributes - 63

3.4.2 Computing 2D Fuzzy Attributes - 64

3.5 Conclusion - 68

4 Information Fusion 69

4.1 Image Reconstruction - 69

4.1.1 Back-pro jection Methods - 70

4.1.2 Transform Algorithms - 71

4.1.3 Algebraic Reconstruction Techniques - 72

4.1.4 Statistical Reconstruction Techniques - 73

4.2 Aggregation of Fuzzy Data - A New Pixel-Based Approach - 74

4.2.1 Partial Defuzzification - 75

4.2.2 Aggregating 2D Fuzzy Particle Maps - 78

4.3 Aggregation of Fuzzy Data - A New Particle-Based Approach - 84

4.3.1 Linking 2D Fuzzy Particles - 84

4.3.2 A Novel Approach for Aggregating Fuzzy Attributes - 87

4.4 Conclusion - 91

5 Decision Making 93

5.1 Introduction - 93

5.2 Decision Trees - 95

5.2.1 Introduction - 95

5.2.2 Tree Construction - 95

5.2.3 Classification of Test Data - 98

5.3 Fuzzy Decision Trees - 99

5.3.1 Crisp Input Data - 99

5.3.2 Fuzzy Input Data - 105

5.4 An Improved Fuzzy Decision Tree - 111

5.4.1 Synthetic Data - 111

5.4.2 Node Tests - 112

5.5 Classifying Mass Attributes from DBT Pro jection Data - 121

5.6 Performance Evaluation and Comparison - 126

5.7 Conclusion - 126

6 CAD for Clinical DBT Data 129

Table of Contents ix

6.1 The Database - 129

6.2 The Algorithm - 132

6.3 Algorithm Performance - 134

6.3.1 Feature Selection - 135

6.3.2 Results Using the Leave-One-Out Method - 136

6.3.3 Obtaining a Final Decision - 137

6.3.4 The Contribution of Fuzzy Logic - 139

6.3.5 Conclusion - 141

7 Conclusion 143

7.1 Main Contributions - 143

7.2 Suggestions for Further Work - 145

A Criteria to Compare Volumes and Surfaces 151

A.1 Volume comparison - 151

A.2 Surface comparison - 152

A.3 Illustrative examples - 153

Appendices

List of Patents and Publications 157

Bibliography 158

Statistiques de consultation

Administrateurs de l'archive uniquement : éditer cet enregistrement