北京生物医学工程

针对颅脑放疗规划的海马体自动勾画平台及其验证

Automated hippocampal contouring platform for craniocerebral radiotherapy planning and its verification

单位：中国科学院合肥物质科学研究院健康与医学技术研究所（合肥230031）中国科学技术大学（合肥 230026）中国科学院合肥肿瘤医院（合肥230031）

关键词：海马体; 放射性脑损伤; 危及器官; 自动勾画; 图像配准

分类号：R318

出版年·卷·期（页码）：2020·39·4（331-336）

摘要：

目的海马体是学习和记忆的神经生物基础,是头颈部放射治疗中需要重点保护的颅内危及器官。海马体轮廓通常由医生手动勾画,操作时间长且依赖医生经验。为提高海马体勾画的效率和可重复性,本文研发了一种海马体自动勾画平台(OAR AutoSketch),系统比较了基于中国人或欧美人大脑图谱配准的分割方法(scbt_Linear、scbt_Nonlinear、TT_linear、TT_Nonlinear),以及基于皮层准的分割方法(FreeSurfer)用于海马体勾画的可行性。方法选取 12 名鼻咽癌患者的数据,采用 OAR AutoSketch生成 5 种海马体轮廓和患者主治医生勾画的海马体轮廓混合呈现,招募 12 名医学影像部的医生进行随机双盲的主观准确性评分;邀请 1 名影像科专家在 20 名鼻咽癌患者的 MRI 图像上手动勾画海马体,作为海马体解剖标准,计算 5 种自动勾画方案的客观准确性。结果主观准确性评分结果显示,自动勾画的准确性普遍优于主治医生的手工勾画结果。和海马体解剖标准的空间相似性结果显示,FreeSurfer 方法准确度最高。结论海马体的自动勾画具备一定的可行性,皮层配准算法总体优于基于图谱的配准算法。

Objective The hippocampus, the neurobiological basis of learning and memory, is a crucial organ-at-risk that needs to be protected during head and neck radiotherapy Hippocampus is often contoured manually by doctors, which is laborious and expertise-dependent. To improve the efficiency and replicability in hippocampus contouring, we develop the OAR AutoSketch and systematically compared the segmentation technique based on Chinese or English brain atlas-based registration ( scbt_Linear, scbt_Nonlinear, TT_Linear, TT_Nonlinear) and the segmentation technique based on surface-based registration (FreeSurfer). Methods Image data of twelve patients of nasopharyngeal carcinoma were employed. The OAR AutoSketch was used to produce five solutions of hippocampus contouring, which was intermixed with the manual contouring solution by the doctor of the patient. To quantify subjective evaluation, twelve doctors from radiology and radiation oncology departments attended a randomized and double-blinded experiment that subjectively rated the contouring accuracy. With a sample of twenty nasopharyngeal carcinomas, the spatial similarity between the contouring solution from the automatic contouring solutions and the anatomical definition of the hippocampus from a radiology expert was computed. Results The subjective rating showed a general superiority of automatic contouring solutions over manual contouring. The analysis of spatial similarity with anatomical definition indicated that the FreeSurfer method was the most accurate among the five kinds of solutions. Conclusions Automatic hippocampus contouring was a feasible alternative to manual hippocampus contouring. The segmentation technique based on surface-based registration outperformed the atlas-based registration method in general.

参考文献：

[ 1 ] 中国放射性脑损伤多学科协作组, 中国医生协会神经内科分会脑与脊髓损害专业委员会. 放射性脑损伤诊治中国专家共识[J]. 中华神经医学杂志, 2019, 18(6):541-549.

[ 2 ] Makale M, Mcdonald CR, Hattangadi-Gluth JA, et al. Mechanisms of radiotherapy-associated cognitive disability in patients with brain tumours [ J ]. Nature Reviews Neurology, 2017, 13:52-64.

[ 3 ] Gondi V, Pugh SL, Tome WA, et al. Preservation of memory with conformal avoidance of the hippocampal neural stem-cell compartment during whole-brain radiotherapy for brain metastases (RTOG 0933):a phase II multi-institutional trial[ J]. Journal of Clinical Oncology, 2014, 32(34):3810-3816.

[ 4 ] Eekers DBP, In’t Ven L, Roelofs E, et al. The EPTN Consensus-based atlas for CT-and MR-based contouring in neuro-oncology [ J]. Radiotherapy and Oncology, 2018, 128 ( 1 ):37-43.

[ 5 ] 冉昭, 简俊明, 王蒙蒙, 等. 基于全卷积神经网络的直肠癌肿瘤磁共振影像自动分割方法[ J]. 北京生物医学工程, 2019, 38(5):465-471.

Ran Z, Jian JM, Wang MM, et al. Automatic segmentation method based on full convolution neural network for rectal cancer tumors in magnetic resonance image [ J ]. Beijing Biomedical Engineering, 2019, 38(5):465-471.

[ 6 ] 刘丰伟, 李汉军, 张逸鹤, 等. 人工智能在医学影像诊断中的应用[J]. 北京生物医学工程, 2019, 38(2):206-211.

Liu FW, Li HJ, Zhang YH, et al. Application of artificial intelligence in medical imaging diagnosis [J]. Beijing Biomedical Engineering, 2019, 38(2):206-211.

[ 7 ] 黄星奕, 丘子明, 许燕. 基于深度监督全卷积神经网络的 MRI 脑图像语义分割算法[J]. 北京生物医学工程, 2019, 38 (3):277-282.

Huang XY, Qiu ZM, Xu Y. Semantic segmentation algorithm for MRI brain image based on deeply supervised fully convolutional network[ J]. Beijing Biomedical Engineering, 2019, 38 ( 3): 277-282.

[ 8 ] Dill V, Franco AR, Pinho MS. Automated methods for hippocampus segmentation:the evolution and a review of the state of the art[J]. Neuroinformatics, 2015, 13:133-150.

[ 9 ] Mccarroll RE, Beadle BM, Balter PA, et al. Retrospective validation and clinical implementation of automated contouring of organs at risk in the head and neck: a step toward automated radiation treatment planning for low-and middle-income countries [J]. Journal of Global Oncology, 2018(4):1-11.

[10] Fischl B, Salat DH, Busa E, et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain[J]. Neuron, 2002, 33(3):341-355.

[11] Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages [ J]. Computers and Biomedical Research, 1996, 29(3):162-173.

[12] Fischl B. FreeSurfer[J]. Neuroimage, 2012, 62(2):774-781.

[13] Liang P, Shi L, Chen N, et al. Construction of brain atlases based on a multi?center MRI dataset of 2020 Chinese adults[ J]. Scientific Reports, 2015, 5:18216.

[14] Collins DL, Neelin P, Peters TM, et al. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space[ J]. Jounral of Computer Assisted Tomography, 1994, 18(2):192-205.

[15] Lin L, Dou Q, Jin YM, et al. Deep learning for automated contouring of primary tumor volumes by MRI for nasopharyngeal carcinoma[J]. Radiology, 2019, 291(3):677-686.

[16] Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al.Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain[J]. NeuroImage, 2002, 15(1):273-289.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录