Mohamad sobri, Nurul Fathin (2021) Dual-thresholding based sinogram interpolation To reduce metal artefact In Computed tomography imaging. Masters thesis, Universiti Sains Malaysia.
![[img]](http://eprints.usm.my/style/images/fileicons/application_pdf.png) |
PDF
Download (602kB) |
Abstract
Metal artefacts can degrade the image quality of computed tomography(CT)images which lead to difficulties and errors in diagnostic analysis. Metal artefactsappear as bright and dark streaks around the metal object. It is caused bybeamhardening effects, photons starvation, non-linear partial volume, photonscatter, undersampling and inappropriate correction algorithm during reconstructionofimage. The aim of this study is to improve the sinogram interpolation techniqueformetal artefacts reduction (MAR) in CT images. The MAR algorithmof this studywas developed using MATLAB platform. The first step of the algorithmwas theacquisition of virtual sinogram from the CT image using Radon transformfunction. Then, the thresholding technique is used to detect and isolate the metal part withinthe sinogram of the CT image. Two interpolation methods, cubic spline (CS) andLaplace interpolation (LI) were applied to replace the missing sinogramdata. Theinterpolated sinogram was then reconstructed using backprojection (byinverseRadon function) and finally the thresholded and interpolated images were fusedtoproduce corrected image. A total of seven phantom and six clinical CTimages withdifferent artefacts severity were selected for MAR correction. Qualitativeandquantitative evaluation of the corrected CT images was performed to evaluatetheeffectiveness of the proposed MAR technique. The findings showed metal artefact was significantly reduced by LI method (p-value = 0.02) as compared to CSmethod(p-value = 0.17). From quantitative analysis, the artefact index (AI) for LI-correctedimages is lower (mean AI = 66.07), but CS-corrected images yielded higher AI values (mean AI = 141.35) than the original images (mean AI = 99.31). The resultsfrom qualitative analysis showed LI-corrected images received higher scores (meanscore ± SD of 1.75 ± 1.22) from radiologist compared to original image (1.08±1.08)and CS-corrected images (1.33 ± 1.23). In conclusion, LI method workmoreeffectively in artefact reduction on both phantom and clinical images of different degrees of artefact severity.
| Item Type: | Thesis (Masters) |
|---|---|
| Subjects: | R Medicine |
| Divisions: | Institut Perubatan & Pergigian Termaju (Advanced Medical & Dental Institute (AMDI)) > Thesis |
| Depositing User: | NOR HASHIMY BIN ABDULLAH |
| Date Deposited: | 22 Apr 2026 02:10 |
| Last Modified: | 22 Apr 2026 02:13 |
| URI: | http://eprints.usm.my/id/eprint/63948 |
Actions (login required)
 |
View Item |
