在放射学中基于图像的搜索:在数据库中使用基于MRI的放射性特征识别脑瘤亚型
Marc von Reppert1, Saahil Chadha1, Klara Willms1
1From the University of Leipzig, Leipzig, Germany (M.V.R., K.L., K.T.F.); Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, CT, USA (S.C., A.A.); Center for Outcomes Research and Evaluation, Yale School of Medicine, New Haven, CT, USA (S.C., S.A.); Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA (A.A.); Department of Radiology, Children's Hospital of Philadelphia, Philadelphia, PA, USA (N.M., M.S.A.); Center for Translational Imaging Analysis and Machine Learning, Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, CT, USA (T.Z.); Department of Neurosurgery, Heinrich-Heine-University, Moorenstrasse 5, 40225 Duesseldorf, Germany (J.L.); Department of Diagnostic and Interventional Radiology, Medical Faculty, University Dusseldorf, 40225, Dusseldorf, Germany (N.T.); University of Duisburg-Essen, 45147 Essen, Germany (L.J.); DKFZ Division of Translational Neuro-oncology at the WTZ, German Cancer Consortium, DKTK Partner Site, University Hospital Essen, 45147 Essen, Germany (L.J.); University of Ulm, Ulm, Germany (S.M.); Visage Imaging, Inc., San Diego, CA, USA (M.D.L); Department of Therapeutic Radiology, Yale School of Medicine, New Haven, CT, USA (S.A.); IBSR consortium: Manpreet Kaur (Ludwig Maximilian University, Munich, Germany), Divya Ramakrishnan (Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, CT, USA), Khaled Bousabarah (Visage Imaging GmbH, Berlin, Germany), Nicolas Linder (University of Leipzig, Leipzig, Germany), Sarah M. Jacobs (Center for Image Sciences, University Medical Center Utrecht, Utrecht, The Netherlands), Francesca Conti (Università di Roma -La Sapienza), Gabriel Cassinelli Petersen (University of Göttingen, Göttingen, Germany), Ujjwal Baid (Indiana University School of Medicine, Indianapolis, IN, USA), Ichiro Ikuta (Department of Radiology, Mayo Clinic Comprehensive Cancer Center, Phoenix, AZ), Fatima Memon (Carolina Radiology, Myrtle Beach, SC, USA), Claudia Kirsch (Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, CT, USA), Melissa Davis (Department of Radiology and Biomedical Imaging, Yale School of Medicine, New Haven, CT, USA), Spyridon Bakas (Indiana University School of Medicine, Indianapolis, IN, USA).
这项研究引入了对原发性脑瘤的基于图像的搜索,使得能够检索类似的历史MRI病例. 该方法有效地识别用于神经放射学研究的视觉上相似的参考MRI.
科学领域:
- 神经辐射学神经辐射学
- 医学成像分析 医学成像分析
- 人工智能在医学中的应用
背景情况:
- 现有的神经放射学参考材料缺乏全面的初级脑瘤表现.
- 基于文本的医学图像搜索受到不一致的放射学报告结构的阻碍.
研究的目的:
- 开发和验证一种基于图像的初级脑瘤搜索方法.
- 利用一个机构数据库来寻找视觉上相似的参考MRI.
主要方法:
- 对295名原发性脑瘤患者 (2000-2021) 的回顾性研究.
- 基于半自动卷积神经网络 (CNN) 的瘤细分和放射性特征提取.
- 减小尺寸 (t-分布式静态邻居嵌入) 和最近邻居检索以获得图像相似性.
主要成果:
- 采用6个组件的t-分布式静态邻居嵌入 (t-SNE) 实现了高性能 (根据瘤类型的平均精度在5从78%到100%).
- 顶部检索的参考病例显示高视觉相似性 (76%"相似"或"非常相似"),根据专家神经放射学家的评价.
结论:
- 开发了一种有效的基于图像的初级脑瘤MRI搜索方法.
- 该方法有助于检索类似于查询瘤的参考病例.
- 专家评估验证了该技术在神经放射学中的有效性.
相关概念视频
Magnetic Resonance Imaging
Brain Imaging
These technologies include computerized axial tomography (CAT or CT scans), positron-emission tomography (PET scans), magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), and Transcranial Magnetic...


