Image quality of dual-energy cone-beam CT with total nuclear variation regularization. (4th February 2022)
- Record Type:
- Journal Article
- Title:
- Image quality of dual-energy cone-beam CT with total nuclear variation regularization. (4th February 2022)
- Main Title:
- Image quality of dual-energy cone-beam CT with total nuclear variation regularization
- Authors:
- Schröder, Lukas
Stankovic, Uros
Rit, Simon
Sonke, Jan-Jakob - Abstract:
- Abstract: Despite the improvements in image quality of cone beam computed tomography (CBCT) scans, application remains limited to patient positioning. In this study, we propose to improve image quality by dual energy (DE) imaging and iterative reconstruction using least squares fitting with total variation (TV) regularization. The generalization of TV called total nuclear variation (TNV) was used to generate DE images. We acquired single energy (SE) and DE scans of an image quality phantom (IQP) and of an anthropomorphic human male phantom (HMP). The DE scans were dual arc acquisitions of 70 kV and 130 kV with a variable dose partitioning between low energy (LE) and high energy (HE) arcs. To investigate potential benefits from a larger spectral separation between LE and HE, DE scans with an additional 2 mm copper beam filtration in the HE arc were acquired for the IQP. The DE TNV scans were compared to SE scans reconstructed with FDK and iterative TV with varying parameters. The contrast-to-noise ratio (CNR), spatial frequency, and structural similarity (SSIM) were used as image quality metrics. Results showed largely improved image quality for DE TNV over FDK for both phantoms. DE TNV with the highest dose allocation in the LE arm yielded the highest CNR. Compared to SE TV, these DE TNV results had a slightly lower CNR with similar spatial resolution for the IQP. A decrease in the dose allocated to the LE arm improved the spatial resolution with a trade-off against CNR. ForAbstract: Despite the improvements in image quality of cone beam computed tomography (CBCT) scans, application remains limited to patient positioning. In this study, we propose to improve image quality by dual energy (DE) imaging and iterative reconstruction using least squares fitting with total variation (TV) regularization. The generalization of TV called total nuclear variation (TNV) was used to generate DE images. We acquired single energy (SE) and DE scans of an image quality phantom (IQP) and of an anthropomorphic human male phantom (HMP). The DE scans were dual arc acquisitions of 70 kV and 130 kV with a variable dose partitioning between low energy (LE) and high energy (HE) arcs. To investigate potential benefits from a larger spectral separation between LE and HE, DE scans with an additional 2 mm copper beam filtration in the HE arc were acquired for the IQP. The DE TNV scans were compared to SE scans reconstructed with FDK and iterative TV with varying parameters. The contrast-to-noise ratio (CNR), spatial frequency, and structural similarity (SSIM) were used as image quality metrics. Results showed largely improved image quality for DE TNV over FDK for both phantoms. DE TNV with the highest dose allocation in the LE arm yielded the highest CNR. Compared to SE TV, these DE TNV results had a slightly lower CNR with similar spatial resolution for the IQP. A decrease in the dose allocated to the LE arm improved the spatial resolution with a trade-off against CNR. For the HMP, DE TNV displayed a lower CNR and/or lower spatial resolution depending on the reconstruction parameters. Regarding the SSIM, DE TNV was superior to FDK and SE TV for both phantoms. The additional beam filtration for the IQP led to improved image quality in all metrics, surpassing the SE TV results in CNR and spatial resolution. … (more)
- Is Part Of:
- Biomedical physics & engineering express. Volume 8:Number 2(2022)
- Journal:
- Biomedical physics & engineering express
- Issue:
- Volume 8:Number 2(2022)
- Issue Display:
- Volume 8, Issue 2 (2022)
- Year:
- 2022
- Volume:
- 8
- Issue:
- 2
- Issue Sort Value:
- 2022-0008-0002-0000
- Page Start:
- Page End:
- Publication Date:
- 2022-02-04
- Subjects:
- cone beam CT -- image reconstruction -- dual-energy -- total nuclear variation -- image quality
Medical physics -- Periodicals
Biophysics -- Periodicals
Biomedical engineering -- Periodicals
Medical sciences -- Periodicals
610.153 - Journal URLs:
- http://iopscience.iop.org/2057-1976/ ↗
http://www.iop.org/ ↗ - DOI:
- 10.1088/2057-1976/ac4e2e ↗
- Languages:
- English
- ISSNs:
- 2057-1976
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - BLDSS-3PM
British Library STI - ELD Digital store - Ingest File:
- 20686.xml