Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Advanced computer simulations of photon-counting spectral CT for crosstalk artifacts evaluation
KTH, School of Engineering Sciences (SCI), Physics.
2025 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Avancerade datorsimuleringar av fotonräknande spektral CT för utvärdering av krosstalförvrängningar (Swedish)
Abstract [en]

Photon-counting computed tomography (PCCT) represents a significant advancement in medical imaging, as it promises higher spatial resolution, spectral imaging capabilities, and reduced radiation dose compared to conventional CT. GE HealthCare is developing an innovative photon-counting detector made of silicon, called the Deep-Silicon Detector, which employs a new edge-on geometry that ensures sufficient detection efficiency. Photoncounting detectors are direct conversion and in general there are no physical barriers between pixels, which allows for crosstalk due to charge sharing and X-ray scattering inside the detector, producing so-called crosstalk artifacts. The Deep-Silicon Detector is equipped with tungsten sheets between the sensors that block cross talk in the slice direction of the detector, but thereare no barriers in the pixel direction. Hence, cross talk remains a major challenge in achieving high-quality reconstructed images and, if not corrected for, poses diffculties from a clinical perspective as well. This thesis focuses on characterizing and explaining the origins of these artifacts, simulating crosstalk inside the detector, and validating the accuracy of the CatSim simulation software, developed by GE HealthCare, in reproducing crosstalk artifacts in clinical images through direct comparison with real images. The work is closely related to the development of a correction algorithm for crosstalk artifacts, as the ability to simulate a phenomenon makes it possible to solve the inverse problem and build such an algorithm. For the simulations, KTH’s supercomputer Dardel was employed, enabling the exploration ofparallelization strategies to optimize the computational effciency of CatSim. The findings confirm that we are able to simulate crosstalk inside the detector, but CatSim can reproduce crosstalk artifacts only to a certain extent. This highlights the need to incorporate more complex phenomena into the simulation before drawing definitive conclusions regarding CatSim’s accuracy and obtaining more realistic artifact representations.

Abstract [sv]

Fotonräknande datortomografi (PCCT) utgör ett betydande framsteg inom medicinsk bildbehandling med löften om högre spatiell och spektral upplösning samt minskad stråldos jämfört med konventionell CT. GE HealthCare utvecklar en ny typ av fotonräknande detektor gjord av kisel med beteckningen Deep Silicon. Detektorn använder sig av ”edge-on-geometri” för att säkerställa absorptionseffektivitet. Fotonräknande detektorer saknar i allmänhet fysiska barriärer mellan pixlarna, vilket leder till växelverkan (så kallad crosstalk) på grund av laddningsdelning och röntgenspridning inuti detektorn och följaktligen artefakter i bilden. Deep Silicon-detektorn har visserligen tunna wolframplåtar mellan sensorerna som reducerar växelverkan i djupled, men problemet kvarstår, framförallt i pixelled. Denna avhandling fokuserar på att karaktärisera och förklara ursprunget till dessa artefakter. Dessutom är målet med avhandlingen att validera noggrannheten i simuleringar av växelverkan inuti detektorn med CatSim, en programvara som har utvecklats av GE HealthCare, genom jämförelser med verkliga bilder. Arbetet är nära relaterat till utvecklingen av enkorrektionsalgoritm för crosstalkartefakter, eftersom förmågan att simulera ett fenomen gör det möjligt att lösa det omvända problemet och bygga en sådan algoritm. För simuleringarna användes KTH:s superdator Dardel och paralleliseringsstrategier för att optimera CatSims beräkningskapacitet utforskades dessutom. Resultaten bekräftar att vi kan simulera crosstalk inuti detektorn, men CatSim kan endast reproducera artefakterna som syns i verkliga bilder till viss del. Detta belyser behovet av att införliva mer komplexa fenomen i simuleringen innan definitiva slutsatser om CatSims noggrannhet kan dras.

 

Place, publisher, year, edition, pages
2025.
Series
TRITA-SCI-GRU ; 2025:022
Keywords [en]
Photon Counting Detector, Crosstalk artifacts, CT, CatSim, Software parallelization
Keywords [sv]
Fotonräknande detektor, Krosstalförvrängningar, CT, CatSim, Programvaru parallellisering
National Category
Physical Sciences
Identifiers
URN: urn:nbn:se:kth:diva-361016OAI: oai:DiVA.org:kth-361016DiVA, id: diva2:1943354
Subject / course
Physics
Educational program
Master of Science - Engineering Physics
Supervisors
Examiners
Available from: 2025-03-10 Created: 2025-03-10 Last updated: 2025-03-10Bibliographically approved

Open Access in DiVA

fulltext(11467 kB)201 downloads
File information
File name FULLTEXT01.pdfFile size 11467 kBChecksum SHA-512
e5edb41c41bbe20e0c97e8f99f7b336fe3ac1e722f21dcc3bce9af10251f3de95ef1717a7b3c8756946d288521aa36a4075d3e22c359d01a507e731a8f5290c1
Type fulltextMimetype application/pdf

By organisation
Physics
Physical Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 201 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 442 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf