TY - JOUR

T1 - Metal artefact reduction for accurate tumour delineation in radiotherapy

AU - Kovacs, David Gergely

AU - Rechner, Laura A.

AU - Appelt, Ane L.

AU - Berthelsen, Anne K.

AU - Costa, Junia C.

AU - Friborg, Jeppe

AU - Persson, Gitte F.

AU - Bangsgaard, Jens Peter

AU - Specht, Lena

AU - Aznar, Marianne C.

PY - 2018

Y1 - 2018

N2 - Background and purpose: Two techniques for metal artefact reduction for computed tomography were studied in order to identify their impact on tumour delineation in radiotherapy. Materials and methods: Using specially designed phantoms containing metal implants (dental, spine and hip) as well as patient images, we investigated the impact of two methods for metal artefact reduction on (A) the size and severity of metal artefacts and the accuracy of Hounsfield Unit (HU) representation, (B) the visual impact of metal artefacts on image quality and (C) delineation accuracy. A metal artefact reduction algorithm (MAR) and two types of dual energy virtual monochromatic (DECT VM) reconstructions were used separately and in combination to identify the optimal technique for each implant site. Results: The artefact area and severity was reduced (by 48-76% and 58-79%, MAR and DECT VM respectively) and accurate Hounsfield-value representation was increased by 22-82%. For each energy, the observers preferred MAR over non-MAR reconstructions (p<0.01 for dental and hip cases, p<0.05 for the spine case). In addition, DECT VM was preferred for spine implants (p<0.01). In all cases, techniques that improved target delineation significantly (p<0.05) were identified. Conclusions: DECT VM and MAR techniques improve delineation accuracy and the optimal of reconstruction technique depends on the type of metal implant.

AB - Background and purpose: Two techniques for metal artefact reduction for computed tomography were studied in order to identify their impact on tumour delineation in radiotherapy. Materials and methods: Using specially designed phantoms containing metal implants (dental, spine and hip) as well as patient images, we investigated the impact of two methods for metal artefact reduction on (A) the size and severity of metal artefacts and the accuracy of Hounsfield Unit (HU) representation, (B) the visual impact of metal artefacts on image quality and (C) delineation accuracy. A metal artefact reduction algorithm (MAR) and two types of dual energy virtual monochromatic (DECT VM) reconstructions were used separately and in combination to identify the optimal technique for each implant site. Results: The artefact area and severity was reduced (by 48-76% and 58-79%, MAR and DECT VM respectively) and accurate Hounsfield-value representation was increased by 22-82%. For each energy, the observers preferred MAR over non-MAR reconstructions (p<0.01 for dental and hip cases, p<0.05 for the spine case). In addition, DECT VM was preferred for spine implants (p<0.01). In all cases, techniques that improved target delineation significantly (p<0.05) were identified. Conclusions: DECT VM and MAR techniques improve delineation accuracy and the optimal of reconstruction technique depends on the type of metal implant.

KW - Delineation uncertainty

KW - Dual energy CT

KW - IGRT

KW - Iterative metal artefact reduction

U2 - 10.1016/j.radonc.2017.09.029

DO - 10.1016/j.radonc.2017.09.029

M3 - Journal article

C2 - 29050958

AN - SCOPUS:85031499162

VL - 126

SP - 479

EP - 486

JO - Radiotherapy & Oncology

JF - Radiotherapy & Oncology

SN - 0167-8140

IS - 3

ER -