Enomoto Yukiko, Yamauchi Keita, Asano Takahiko, Otani Katharina, Iwama Toru
1 Department of Neurosurgery, Graduate School of Medicine, Gifu University, Gifu, Japan.
2 Department of Radiology, Graduate School of Medicine, Gifu University, Gifu, Japan.
Interv Neuroradiol. 2018 Jun;24(3):303-308. doi: 10.1177/1591019917754039. Epub 2018 Feb 21.
Background and purpose C-arm cone-beam computed tomography (CBCT) has the drawback that image quality is degraded by artifacts caused by implanted metal objects. We evaluated whether metal artifact reduction (MAR) prototype software can improve the subjective image quality of CBCT images of patients with intracranial aneurysms treated with coils or clips. Materials and methods Forty-four patients with intracranial aneurysms implanted with coils (40 patients) or clips (four patients) underwent one CBCT scan from which uncorrected and MAR-corrected CBCT image datasets were reconstructed. Three blinded readers evaluated the image quality of the image sets using a four-point scale (1: Excellent, 2: Good, 3: Poor, 4: Bad). The median scores of the three readers of uncorrected and MAR-corrected images were compared with the paired Wilcoxon signed-rank and inter-reader agreement of change scores was assessed by weighted kappa statistics. The readers also recorded new clinical findings, such as intracranial hemorrhage, air, or surrounding anatomical structures on MAR-corrected images. Results The image quality of MAR-corrected CBCT images was significantly improved compared with the uncorrected CBCT image ( p < 0.001). Additional clinical findings were seen on CBCT images of 70.4% of patients after MAR correction. Conclusion MAR software improved image quality of CBCT images degraded by metal artifacts.
背景与目的 C型臂锥形束计算机断层扫描（CBCT）存在的缺点是，植入的金属物体会导致图像质量因伪影而下降。我们评估了金属伪影减少（MAR）原型软件是否能改善接受线圈或夹子治疗的颅内动脉瘤患者CBCT图像的主观图像质量。材料与方法 44例植入线圈（40例）或夹子（4例）的颅内动脉瘤患者接受了一次CBCT扫描，从中重建了未校正和MAR校正的CBCT图像数据集。三位不知情的阅片者使用四点量表（1：优秀，2：良好，3：差，4：糟糕）评估图像集的图像质量。将三位阅片者对未校正图像和MAR校正图像的中位数评分进行配对Wilcoxon符号秩检验比较，并通过加权kappa统计评估阅片者间变化评分的一致性。阅片者还记录了MAR校正图像上的新临床发现，如颅内出血、空气或周围解剖结构。结果 与未校正的CBCT图像相比，MAR校正的CBCT图像的图像质量有显著改善（p < 0.001）。MAR校正后，70.4%的患者的CBCT图像上发现了额外的临床发现。结论 MAR软件改善了因金属伪影而退化的CBCT图像的质量。
