MS/MRI Research Group (Division of Neurology), University of British Columbia, Vancouver, BC, Canada.
MS/MRI Research Group (Division of Neurology), University of British Columbia, Vancouver, BC, Canada; Department of Radiology, University of British Columbia, Vancouver, BC, Canada.
Neuroimage Clin. 2019;23:101918. doi: 10.1016/j.nicl.2019.101918. Epub 2019 Jul 5.
Accurate segmentation of MS lesions on MRI is difficult and, if performed manually, time consuming. Automatic segmentations rely strongly on the image contrast and signal-to-noise ratio. Literature examining segmentation tool performances in real-world multi-site data acquisition settings is scarce.
FLAIR, a combination of T-weighted and fluid attenuated inversion recovery (FLAIR) images, improves tissue contrast while suppressing CSF. We compared the use of FLAIR and FLAIR in LesionTOADS, OASIS and the lesion segmentation toolbox (LST) when applied to non-homogenized, multi-center 2D-imaging data.
Lesions were segmented on 47 MS patient data sets obtained from 34 sites using LesionTOADS, OASIS and LST, and compared to a semi-automatically generated reference. The performance of FLAIR and FLAIR was assessed using the relative lesion volume difference (LVD), Dice coefficient (DSC), sensitivity (SEN) and symmetric surface distance (SSD). Performance improvements related to lesion volumes (LVs) were evaluated for all tools. For comparison, LesionTOADS was also used to segment lesions from 3 T single-center MR data of 40 clinically isolated syndrome (CIS) patients.
Compared to FLAIR, the use of FLAIR in LesionTOADS led to improvements of 31.6% (LVD), 14.0% (DSC), 25.1% (SEN), and 47.0% (SSD) in the multi-center study. DSC and SSD significantly improved for larger LVs, while LVD and SEN were enhanced independent of LV. OASIS showed little difference between FLAIR and FLAIR, likely due to its inherent use of Tw and FLAIR. LST replicated the benefits of FLAIR only in part, indicating that further optimization, particularly at low LVs is needed. In the CIS study, LesionTOADS did not benefit from the use of FLAIR as the segmentation performance for both FLAIR and FLAIR was heterogeneous.
In this real-world, multi-center experiment, FLAIR outperformed FLAIR in its ability to segment MS lesions with LesionTOADS. The computation of FLAIR enhanced lesion detection, at minimally increased computational time or cost, even retrospectively. Further work is needed to determine how LesionTOADS and other tools, such as LST, can optimally benefit from the improved FLAIR contrast.
磁共振成像（MRI）上 MS 病变的精确分割较为困难，且手动分割非常耗时。自动分割严重依赖于图像对比度和信噪比。目前，文献中关于在真实世界多站点数据采集环境下分割工具性能的研究较少。
FLAIR 是 T 加权和液体衰减反转恢复（FLAIR）图像的组合，可在抑制脑脊液的同时改善组织对比度。我们比较了 FLAIR 和 FLAIR 在 LesionTOADS、OASIS 和病变分割工具箱（LST）中的应用，这些工具应用于非均匀、多中心 2D 成像数据。
在 34 个站点的 47 例 MS 患者数据集中，使用 LesionTOADS、OASIS 和 LST 对病变进行分割，并与半自动生成的参考进行比较。使用相对病变体积差异（LVD）、Dice 系数（DSC）、灵敏度（SEN）和对称表面距离（SSD）评估 FLAIR 和 FLAIR 的性能。评估了所有工具与病变体积（LV）相关的性能改进。为了比较，LesionTOADS 还用于分割 40 例临床孤立综合征（CIS）患者的 3T 单中心 MR 数据中的病变。
与 FLAIR 相比，在多中心研究中，LesionTOADS 中使用 FLAIR 可分别提高 31.6%（LVD）、14.0%（DSC）、25.1%（SEN）和 47.0%（SSD）。较大 LV 的 DSC 和 SSD 显著提高，而 LVD 和 SEN 则独立于 LV 而增强。OASIS 中 FLAIR 和 FLAIR 之间差异较小，可能是由于其固有地使用了 Tw 和 FLAIR。LST 仅部分复制了 FLAIR 的益处，表明需要进一步优化，尤其是在低 LV 时。在 CIS 研究中，LesionTOADS 并未受益于 FLAIR 的使用，因为 FLAIR 和 FLAIR 的分割性能存在异质性。
在这项真实世界的多中心实验中，与 FLAIR 相比，LesionTOADS 更能准确分割 MS 病变。FLAIR 计算增强了病变检测能力，在计算时间或成本仅略有增加的情况下，甚至可以进行回顾性检测。需要进一步研究以确定如何使 LesionTOADS 和其他工具（如 LST）能够从改进的 FLAIR 对比度中受益。
