论文部分内容阅读
本文提出一种运用兆伏锥形束CT(MVCBCT)校正千伏CT(kVCT)图像中假牙金属伪影的新方法。该方法分别用kVCT和MVCBCT扫描佩戴假牙的患者,得到两种CT图像。在kVCT中阈值分割得到金属图像,运用MVCBCT和kVCT融合得到先验图像,对先验图像前投影来替代原始金属区投影,最后通过滤波反投影(FBP)重建图像。将本文方法校正后效果与归一化金属伪影校正法(NMAR)、以MVCBCT为先验图像的归一化金属伪影校正法(NMAR-MV),以及线性插值法(LIMAR)这三种常用伪影校正方法进行比较,计算其归一化均方根偏差(NRMSD)和平均绝对偏差(MAD)。实验结果显示本文方法去除了严重的金属伪影且没有引入其他伪影,基于参考图像计算的NRMSD值和MAD值最小。NMAR、NMAR-MV、LIMAR以及本文方法的NRMSD值分别为21.0%、22.1%、41.9%、17.0%;MAD值分别为232、235、553、205 HU。本文提出的伪影校正方法能较好地去除假牙的金属伪影,大幅改善CT图像质量。
In this paper, we propose a new method to correct denture metal artifacts in kV CT images using megavoltage cone beam computed tomography (MVCBCT). The method of kVCT and MVCBCT were worn wearing dentures patients to obtain two kinds of CT images. A metal image was obtained by threshold segmentation in kVCT. A priori image was obtained by fusion of MVCBCT and kVCT. Projection of the original metal area was replaced by prior projection of the prior image. Finally, the image was reconstructed by FBP. The results of this paper are compared with normalized metal artifact correction (NMAR), normalized metal artifact correction (NMAR-MV) based on MVCBCT, and linear interpolation (LIMAR) Common artifact correction methods were compared to calculate the normalized root mean square deviation (NRMSD) and mean absolute deviation (MAD). Experimental results show that the proposed method removes severe metal artifacts and introduces no other artifacts. The NRMSD and MAD values calculated based on the reference image are the least. The NRMSD values of NMAR, NMAR-MV, LIMAR and the proposed method were 21.0%, 22.1%, 41.9% and 17.0%, respectively. The MAD values were 232, 235, 553 and 205 HU, respectively. The artifact correction method proposed in this paper can remove the metal artifacts of dentures and improve the CT image quality significantly.