Multi-institutional Validation Study of Commercially Available Deformable Image Registration Software for Thoracic Images

详细信息    查看全文

• 作者：Noriyuki Kadoya ; PhD^&lowast ; ; ^{kadoya.n@rad.med.tohoku.ac.jp" class="auth_mail" title="E-mail the corresponding author} ; Yujiro Nakajima ; RTT^&lowast ; ; Masahide Saito ; RTT^&lowast ; ; Yuki Miyabe ; MS^&dagger ; ; Masahiko Kurooka ; MS^&Dagger ; ; Satoshi Kito ; MS^§ ; ; Yukio Fujita ; PhD^‖ ; Motoharu Sasaki ; RTT^¶ ; ; Kazuhiro Arai ; MS^# ; Kensuke Tani ; MS^&lowast ; &lowast ; ; Masashi Yagi ; PhD^&dagger ; &dagger ; ; Akihisa Wakita ; MS^&Dagger ; &Dagger ; ; Naoki Tohyama ; MS^§ ; § ; ; Keiichi Jingu ; MD ; PhD^&lowast ;
• 刊名：International Journal of Radiation Oncology*Biology*Physics
• 出版年：2016
• 出版时间：1 October 2016
• 年：2016
• 卷：96
• 期：2
• 页码：422-431
• 全文大小：3065 K

文摘

To assess the accuracy of the commercially available deformable image registration (DIR) software for thoracic images at multiple institutions.

Methods and Materials

Thoracic 4-dimensional (4D) CT images of 10 patients with esophageal or lung cancer were used. Datasets for these patients were provided by DIR-lab (dir-lab.com) and included a coordinate list of anatomic landmarks (300 bronchial bifurcations) that had been manually identified. Deformable image registration was performed between the peak-inhale and -exhale images. Deformable image registration error was determined by calculating the difference at each landmark point between the displacement calculated by DIR software and that calculated by the landmark.

Results

Eleven institutions participated in this study: 4 used RayStation (RaySearch Laboratories, Stockholm, Sweden), 5 used MIM Software (Cleveland, OH), and 3 used Velocity (Varian Medical Systems, Palo Alto, CA). The ranges of the average absolute registration errors over all cases were as follows: 0.48 to 1.51 mm (right-left), 0.53 to 2.86 mm (anterior-posterior), 0.85 to 4.46 mm (superior-inferior), and 1.26 to 6.20 mm (3-dimensional). For each DIR software package, the average 3-dimensional registration error (range) was as follows: RayStation, 3.28 mm (1.26-3.91 mm); MIM Software, 3.29 mm (2.17-3.61 mm); and Velocity, 5.01 mm (4.02-6.20 mm). These results demonstrate that there was moderate variation among institutions, although the DIR software was the same.

Conclusions

We evaluated the commercially available DIR software using thoracic 4D-CT images from multiple centers. Our results demonstrated that DIR accuracy differed among institutions because it was dependent on both the DIR software and procedure. Our results could be helpful for establishing prospective clinical trials and for the widespread use of DIR software. In addition, for clinical care, we should try to find the optimal DIR procedure using thoracic 4D-CT data.