La lecture à portée de main
Découvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDécouvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDescription
Sujets
Informations
Publié par | ruprecht-karls-universitat_heidelberg |
Publié le | 01 janvier 2011 |
Nombre de lectures | 15 |
Langue | Deutsch |
Poids de l'ouvrage | 10 Mo |
Extrait
Dissertation
submitted to the
Combined Faculties for the Natural Sciences and for Mathematics
of the Ruperto-Carola University of Heidelberg, Germany
for the degree of
Doctor of Natural Sciences
Put forward by
Dipl.-Phys. Tobias Sebastian Schwarz
Born in: Heidelberg
Oral examination: 01. Feb. 2011
COUPLED SHAPE MODELS
FOR THE DIAGNOSIS OF
ORGAN MOTION RESTRICTION
Referees: Prof. Dr. Bernd Jähne
Prof. Dr. Hans-Peter Meinzer
Zusammenfassung
Annähernd 30% der weltweiten Todesfälle sind auf Erkrankungen des Herzens
und der Lunge zurückzuführen, wobei die meisten dieser Erkrankungen während
ihres Verlaufs die Mobilität des betroffenen Organs verändern. Viele dieser To-
desfälle könnten durch eine frühzeitige Erkennung und Behandlung der Erkran-
kung vermieden werden. Deshalb wurden im Zuge dieser Arbeit Methoden ent-
wickelt, um aus Segmentierungen von dynamischen Magnetresonanztomogra-
phie-Daten quantitative Kennzahlen für die funktionale Analyse der Herz- und
Lungenbewegung zu generieren. Ein automatisiertes Segmentierungsverfahren
basierend auf gekoppelten Formmodellen wurde entwickelt, welches wechsel-
seitige Informationen der Form und Geometrie mehrerer korrelierter Objekte
mit einbezieht, und somit 40% bessere Ergebnisse im Vergleich zur Verwendung
einzelner Modelle erzielte. Im Fall des Herzens wurde ein Volumenberechnungs-
fehler von unter 13% erreicht, was in der Größenordnung der Interobserver-
Variabilität liegt. Für die Lunge konnte ein Volumenfehler von unter 70ml gezeigt
werden. Aus den Segmentierungsergebnissen wurden funktionale Parameter der
lokalen Organdynamik abgeleitet und visualisiert, die gegen konventionelle Diag-
nosemethoden evaluiert wurden und dabei gute Übereinstimmung zeigen, darü-
ber hinaus jedoch eine lokal und regionale Mobilitätscharakterisierung erlauben.
Abstract
Approximately 30% of deaths worldwide originate from diseases of the heart and
lungs, whereby most of which alter mobility of the organ during their course.
Many of these deaths could be avoided by early detection and treatment of the
disease. Therefore, in this thesis, methods have been developed for the analysis
of dynamic magnetic resonance imaging data, and the generation of quantitative
measures for the functional cardiac and pulmonary analysis from segmentation
of these image sequences. An automated coupled shape model segmentation
scheme has been developed that incorporates mutual information on shape and
geometry of correlated objects to cope with the difficulties found in the image
data, showing 40% better results compared to single models. For the heart, a
volumetric error of below 13% was achieved, which is in the magnitude of inter-
observer variability. For the lungs, a volume calculation error of below 70ml
could be shown. From the segmentation results, functional parameters describ-
ing the local organ dynamics have been derived and visualized. The quantitative
parameters were evaluated against conventional diagnostic techniques and
showed good agreement, but with the benefit of a local and regional mobility
characterization.
Table of contents
Table of contents .............................................................................. vii
List of figures ...................... xi
List of tables ..................................................................................... xiv
1 Introduction ................ 1
1.1 Motivation .......................................................................... 1
1.2 Objectives ........... 2
1.3 Structure of the Thesis ........................................................ 4
2 Background ................................................. 5
2.1 Anatomy of the Human Heart ............................................ 5
2.2 Cardiovascular Diseases .................... 11
2.2.1 Carditis ...................................... 11
2.2.2 Heart Failure ............................. 12
2.2.3 Valvular Insufficiency ................................................ 12
2.2.4 Valvular Stenosis ....................... 13
2.2.5 Coronary Heart Disease (CHD) .................................. 14
2.3 Cardiac Diagnosis .............................................................. 15
2.4 Physiological Parameters of the Heart ............................. 16
2.4.1 Volumetry ................................. 16
2.4.2 Wall mass .................................. 17
2.4.3 Wall motion............................... 18
2.4.4 Wall Thickness ........................................................... 18
2.4.5 Wall Thickening ......................... 18
2.5 Bull's Eye Diagram ............................. 18
2.6 AHA Standard .................................................................... 19
2.7 Respiratory System ........................... 21
vii
2.7.1 Anatomy of the Respiratory System ......................... 21
2.7.2 Physiology of the Respiratory System....................... 23
2.7.3 Lung Volumes and Capacities ................................... 24
2.8 Restrictive Respiratory Diseases ....... 27
2.8.1 Pleural effusion ......................................................... 28
2.8.2 Adhesive pleurisy ...................... 28
2.8.3 Pulmonary Fibrosis .................... 28
2.8.4 Pulmonary Edema ..................................................... 28
2.8.5 Lung Cancer ............................... 29
2.9 Respiratory Diagnosis ....................... 30
2.9.1 Spirometry ................................................................ 30
2.9.2 Bronchospasmolytic Test .......... 32
2.9.3 Body Plethysmography ............................................. 32
2.9.4 Arterial Blood Gas ..................... 33
3 State of the Art.......................................... 35
3.1 Medical Image Segmentation ........................................... 35
3.1.1 Multiple Organ Segmentation .. 35
3.2 Cardiac Diagnosis .............................................................. 37
3.2.1 Commercially Available Products ............................. 37
3.2.2 LV Segmentation and Functional Analysis ................ 38
3.3 Pulmonary Diagnosis ......................................................... 41
3.3.1 MRI Lung Segmentation ............ 43
3.4 Interactive Model Correction............................................ 44
4 Methods .................................................... 45
4.1 Deformable Shape Model ................. 45
4.1.1 Shape......................................................................... 45
4.1.2 Statistical Variability of Shape .. 46
4.1.3 Shape Representation ............... 46
4.1.4 Shape Space .............................................................. 47
4.1.5 Shape Model Construction ....................................... 49
viii
4.2 Model Search .................................................................... 51
4.2.1 Local Appearance Model .......... 51
4.2.2 Search Algorithm....................................................... 52
4.3 Coupled Models ................................ 53
4.3.1 Model Initialization ................... 54
4.3.2 Shape Space Coupling ............................................... 56
4.3.3 Joint Shape Space...................... 56
4.3.4 Unified Shape Space ................. 57
4.3.5 Shape Parameter application .................................... 58
4.3.6 Geometrical Coupling ............... 60
4.4 Summary of Model Search ................................................ 62
4.5 Interactive Correction ....................... 63
4.6 Quantitative Analysis of Cardiac Motion .......................... 66
4.7 Cardiac Diagnosis .............................................................. 68
4.7.1 Implementation of the Bull’s Eye Diagram ............... 68
4.7.2 Projection .................................. 68
4.7.3 Parameter calculation ............... 72
4.8 Quantitative Analysis of Pulmonary Motion ..................... 74
4.8.1 Virtual Spirometry ................................ 74
4.8.2 Extraction of quantitative Measures ........................ 75
4.8.3 Motion field calculation ............................................ 76
4.9 2D+t Pulmonary Function Analysis ... 79
4.10 2D+t image analysis .......................................................... 80
5 Results ....................................................... 83
5.1 Evaluation of Segmentation Quality ................................. 83
5.1.1 Evaluation Metrics .................................................... 83
5.2 Segmentation of the Left Ventricle ... 85
5.2.1 Experimental setup ................... 85