Scattered radiation in cone beam computed tomography [Elektronische Ressource] : analysis, quantification and compensation / vorgelegt von Jens Wiegert

Sujets

• Gesundheit

Informations

• Publié par: rheinisch-westfalischen_technischen_hochschule_-rwth-_aachen
• Publié le: 01 janvier 2007
• Nombre de lectures: 24
• Langue: English
• Poids de l'ouvrage: 4 Mo

Extrait

Scattered radiation in cone-beam
computed tomography: analysis,
quantification and compensation
Von der Fakult¨at fu¨r Elektrotechnik und Informationstechnik
der Rheinisch-Westf¨alischen Technischen Hochschule Aachen
zur Erlangung des akademischen Grades eines Doktors der
Ingenieurwissenschaften genehmigte Dissertation
vorgelegt von
Diplom-Ingenieur Jens Wiegert
aus Dortmund
Berichter: Universitatsprofessor Dr.-Ing. T. Aach¨
Universitatsprofessor Dr.-Ing. J. R. Ohm¨
Tag der mundlichen Prufung: 2. Mai 2007¨ ¨
Diese Dissertation ist auf den Internetseiten der
Hochschulbibliothek online verfu¨gbar.2Preface
This thesis summarizes the work carried out during my PhD thesis studies
at the Philips Research Laboratories in Aachen, Germany. Many results
and achievements described in this work were only possible because of the
support of a lot of people and I would like use this opportunity to express
my gratitude to these persons.
First, I would like to thank Prof. Dr.-Ing. Til Aach for giving me the
opportunity to do my PhD thesis in cooperation with an potent industrial
partner, and under his professional guidance, first at Lubeck University,¨
lateratRWTHAachenUniversity. IamalsogratefultoProf.Dr.-Ing.Jens-
Rainer Ohm for the interest in my work and for beeing the co-examiner in
the jury.
I want to thank Falko Busse and Michael Overdick for their continued
interest in my work and the various types of support they provided.
I am indebted to my current and former project colleagues, Matthias
Bertram, Steffen Hohmann, Christoph Neukirchen, Anne-Katrin Kienap-
pel,NorbertConrads,GeorgRose,andDirkSch¨afer. Youcontributedwith
countless fruitful discussions and you surprisingly never complained about
the endless texts to proofread. I also would like to thank our internship
and diploma thesis workers Timo Sattel, J¨org Wulff, Christian Steinhaus,
and Achim Kanert for their fruitful collaboration in the project work their
good and fresh ideas.
I am grateful for the support of our project partners and funders at
Philips Medical Systems, Hein Haas, Jan Timmer, Niels Noordhoek, Peter
van de Haar and Thijs Elenbaas. Your input is always very important
to come up with algorithms that are not only interesting from a scientific
point of view but also have a perspective for real world problems.
I want to thank our colleagues at Philips Research Hamburg, Hermann
Schomberg, Rainer Koppe, Michael Grass, Volker Rasche, and Babak
Movassaghi for providing reconstruction insight and software where ever
necessary and for being reliable partners in doing good research.
AndthenthereisamultitudeofcolleaguesIwanttothankatPhilipsRe-
search Aachen: Olivier Ecabert and Klaus Fiedler, for the pleaseant inter-
iPreface
action while sharing an office with me and making work a really enjoyable
experience, Klaus Jurgen Engel for working together on the development¨
of DiPhoS, Jorg Bredno, Andreas Godicke, Bernd Menser, Walter Rutten,¨ ¨ ¨
Kai Eck, Sabine Mollus, Gereon Vogtmeier, and Jurgen Weese and many¨
many more, for many fruitful discussions and good ideas, Rainer Pietig
and Astrid Lewalter, who were never tired of answering another question
regarding X-ray tubes, and Christian Baumer, Heinrich von Busch, and¨
BarbaraMartinLeungfortheirvarioussupportsin gettingeverythingcor-
rectly on paper.
I also would like to thank Thomas Stehle for his help in preparing my
public PhD presentation.
My very special thanks go to my family, on whose support I always can
and could rely on. Especially I am very grateful to my parents - without
your love, support and education I would never have come to this point.
Finally, I want to express my very special gratitude to my dear wife Marta
who has endured me in this laborious time and has always supported me
with her love. Thank you!
iiAbstract
For imaging during minimally-invasive treatment in the so-called catheter
laboratory conventional X-ray projection imaging is classically used. Par-
ticularly in cardio-vascular angiography and neuroradiology, so-called C-
arm systems are used, enabling a flexible positioning of X-ray tube and
detector. In the last years, these systems experienced the most important
technical innovation with the introduction of 3D imaging functionality by
means of cone-beam computed tomography (CBCT). With this technique
a large number of projections is acquired during a rotation of the C-arm
around the patient. Afterwards, these projections are reconstructed to
volumetric images using algorithms similar to those used in classical com-
puted tomography. The objective of current research is to improve the 3D
image qualityin order to extendthe imaging capabilityto high qualitylow
contrast imaging with C-arm X-ray systems.
In this context, this thesis addresses the problem of scattered radiation.
BecauseinCBCTwithlargeareaX-raydetectorstheirradiatedpatientvol-
umeissubstantiallylargerthaninclassicalcomputedtomography,alsothe
amount of scattered radiation reaching the detector is significantly larger
and can even be superior to the amount of primary radiation. Therefore,
scatteredradiationisamajorsourceofimagedegradationandnonlinearity
inflat-detectorbasedCBCTandisthemostseverecauseofinhomogeneity
artifacts in reconstructed images.
Theprimaryobjectivesofthisthesisarethedetailedquantitativeanaly-
sis of scattered radiation, the assessment of existing scatter compensation
methods as well as the development of new effective methods for the re-
duction of scatter induced artifacts.
AfteranintroductiontothephysicalandalgorithmicprinciplesofCBCT
in the first part of the thesis, at first a detailed quantitative analysis of the
characteristicsofscatteredradiationinprojectionsofCBCTisundertaken.
This analysis is based on the advancements of a Monte-Carlo CBCT sim-
ulator allowing to study realistic and clinically relevant patient geometries
obtained from real data sets of conventional computed tomography. With
this method practically noise free reference data sets for typical measure-
iiiAbstract
ment objects such as the head, thorax and pelvis region are generatedthat
allow to exactly study the influence of scattered radiation and that are
used in the course of the thesis for the assessment of the various methods
for scatter compensation.
Subsequently, the impact of scattered radiation on the reconstructed
volume is quantitatively studied. For this purpose, and as one of the key
contributions of this thesis, a mathematical description of the propagation
of the most relevant image quality characteristics, signal, contrast, and
noise from the projections into the reconstructed volume is derived.
Based on this description and based on the well known Feldkamp-
algorithm, new reconstruction algorithms are developed that – instead of
the usual CT Hounsfield values – allow for reconstruction of the respective
image quality feature, i.e., voxel-wise inhomogeneities, voxel-wise decrease
of object contrast, and voxel-wise standard deviations of the noise.
Using the developedanalysis method and based on the created reference
datasetsacomprehensivestudyofanti-scattergridsastheclassicalmethod
ofscattersuppressionrevealsthatthequalityofanti-scattergridsavailable
for X-ray flat-detectors is not sufficient in order to effectively suppress
scatter induced artifacts. Additionally, the investigation shows that usage
of strongly scatter reducing anti-scatter grids has a negative impact on the
signal-to-noise ratio.
Therefore, in order to provide the desired image quality in low-contrast
CBCT, it is essential to correct for scatter contained in the projections by
means of software-based a-posteriori methods. In literature, however, so
far no practical methods can be found.
Therefore – as second important contribution – in this thesis a number
of new scatter compensation methods have been developed. These can be
grouped in four different classes: post-processing techniques performed in
3D reconstructed images, methods using model based Monte-Carlo simu-
lations, methods based on single scatter estimation schemes, and iterative
methods using artifact evaluation and feedback schemes.
All correction methods are comparatively validated using the clinical
referencedatasets. Itisshownthatespeciallyexploitationofbothavailable
datadomains,theplanarprojectiondataandthe3Dinformation,allowsfor
combating the large scatter background present in this application and to
meet the demanding accuracy requirements to achieve the expected image
quality in CBCT.
ivKurzfassung
Zur Bildgebung w¨ahrend minimal-invasiver Eingriffe im sogenannten
Katheterlabor wird klassisch konventionelle Ro¨ntgendurchleuchtung mit
fl¨achigen Ro¨ntgendetektoren eingesetzt. Hierbei finden, vor allem
in der kardio-vaskul¨aren Angiographie und der Neuroradiologie, so-
genannte C-Bogen-Systeme Anwendung, mit denen die Position von
Rontgenrohre und -detektor flexibel eingestellt werden kann. Die¨ ¨
wichtigste Neuerung erfuhren diese Systeme in den letzten Jahren mit
der Einfuhrung dreidimensionaler Bildgebungsfunktionalitat mittels der¨ ¨
Kegelstrahl-Computertomographie (KSCT). Dazu werden wahrend einer¨
kontinuierlichen Rotation des C-Bogens um d