Metal Artifact Reduction in Sinograms of Dental Computed Tomography
Us, Defne (2013)
Us, Defne
2013
Master's Degree Programme in Biomedical Engineering
Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2013-06-05
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201306121199
https://urn.fi/URN:NBN:fi:tty-201306121199
Tiivistelmä
Use of metal objects such as dental implants, fillings, crowns, screws, nails, prosthesis and plates have increased in dentistry over the past 20 years, which raised a need for new methods for reducing the metal artifacts in medical images. Although there are several algorithms for metal artifact reduction, none of these algorithms are efficient enough to recover the original image free of all artifacts.
This thesis presents two approaches for reducing metal artifacts through accurate segmentation of metal objects on dental computed tomography images. First approach was based on construction and tilting of a 3D jaw phantom, aiming to obtain fewer metals on each slice. 3D jaw phantom included the main anatomical structures of a jaw, and multiple metal fillings inserted on the teeth. Each jaw slice on the 3D phantom was tilted in order to mimic the (1) nodding movement, and (2) mouth opening/closing. Second approach was to segment the metals on an experimental dataset, consisting of a Cone-Beam Computed Tomography image, by using different segmentation algorithms. K-means clustering, Otsu’s thresholding method and logarithmic enhancement were used for extracting the metals from a real dental CT slice. Once the metal fillings on the jaw phantom were segmented out from the image, they were compensated by gap filling methods; Discrete Cosine Domain Gap Filling and inpainting.
Qualitative and quantitative analyses were carried out for evaluating the performance of implemented segmentation methods. Efficiency of tilting alternatives on the segmentation of metal fillings was compared. In conclusion, jaw opening/closing movement between 24º-30º suggested a significant enhancement in segmentation, thus, metal artifact reduction on the jaw phantom. Inpainting method showed a better performance for both simulated and experimental dataset over the DCT domain gap filling method. Moreover, merging the logarithmic enhancement and inpainting showed superior results over other metal artifact reduction alternatives.
This thesis presents two approaches for reducing metal artifacts through accurate segmentation of metal objects on dental computed tomography images. First approach was based on construction and tilting of a 3D jaw phantom, aiming to obtain fewer metals on each slice. 3D jaw phantom included the main anatomical structures of a jaw, and multiple metal fillings inserted on the teeth. Each jaw slice on the 3D phantom was tilted in order to mimic the (1) nodding movement, and (2) mouth opening/closing. Second approach was to segment the metals on an experimental dataset, consisting of a Cone-Beam Computed Tomography image, by using different segmentation algorithms. K-means clustering, Otsu’s thresholding method and logarithmic enhancement were used for extracting the metals from a real dental CT slice. Once the metal fillings on the jaw phantom were segmented out from the image, they were compensated by gap filling methods; Discrete Cosine Domain Gap Filling and inpainting.
Qualitative and quantitative analyses were carried out for evaluating the performance of implemented segmentation methods. Efficiency of tilting alternatives on the segmentation of metal fillings was compared. In conclusion, jaw opening/closing movement between 24º-30º suggested a significant enhancement in segmentation, thus, metal artifact reduction on the jaw phantom. Inpainting method showed a better performance for both simulated and experimental dataset over the DCT domain gap filling method. Moreover, merging the logarithmic enhancement and inpainting showed superior results over other metal artifact reduction alternatives.