先天性肺异常的术前可视化:人工智能与虚拟现实技术的融合
本研究对5例无症状未手术的先天性肺异常(CLA)病例,采用半自动分割支气管血管结构与病灶,并用虚拟现实(VR)与人工智能(AI)技术可视化。两位专家独立评估常规CT与三维VR图像的信息价值。基于CT图像,仅1例的肺段定位在专家间一致;基于三维VR图像,5例中3例定位一致。若手术,加用三维VR工具将在4例中改变手术策略(肺叶切除与肺段切除的选择)。
要点
- AI-VR融合可视化在CLA患者肺段水平解剖中技术可行,5例均成功完成支气管血管分割与容积计算。
- 基于CT图像,专家间肺段定位仅1例一致;基于三维VR图像,5例中3例一致。
- 若手术,加用三维VR工具将在4例中改变手术策略(肺叶切除与肺段切除的选择)。
核心结论
- 本研究证明了对CLA患者进行AI-VR融合的肺段水平解剖可视化在技术上是可行的。
- 因此,有必要进一步探索三维VR在识别适合解剖性肺段切除病例中的价值。
研究设计与样本
- 研究纳入5例无症状、未手术的先天性肺异常病例,采用半自动分割支气管血管结构与病灶,并用虚拟现实(VR)和人工智能(AI)技术可视化。
- 两位专家通过问卷独立评估常规计算机断层扫描(CT)与三维VR图像的信息价值。
主要发现
- 在全部5例病例中,VR环境下的支气管血管分割、容积计算和图像可视化均成功完成。
- 基于CT图像, consulted specialists之间对CLA病灶所属特定肺段的定位仅在1例中一致。
- 基于三维VR图像,5例中3例的定位一致。
- 若患者接受手术,将三维VR工具加入术前检查将在4例中改变手术策略(即肺叶切除与肺段切除的选择)。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文结论指出有必要进一步探索三维VR在识别适合解剖性肺段切除病例中的价值。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVES: When surgical resection is indicated for a congenital lung abnormality (CLA), lobectomy is often preferred over segmentectomy, mostly because the latter is associated with more residual disease. Presumably, this occurs in children because sublobar surgery often does not adhere to anatomical borders (wedge resection instead of segmentectomy), thus increasing the risk of residual disease. This study investigated the feasibility of identifying eligible cases for anatomical segmentectomy by combining virtual reality (VR) and artificial intelligence (AI). METHODS: Semi-automated segmentation of bronchovascular structures and lesions were visualized with VR and AI technology. Two specialists independently evaluated via a questionnaire the informative value of regular computed tomography versus three-dimensional (3D) VR images. RESULTS: Five asymptomatic, non-operated cases were selected. Bronchovascular segmentation, volume calculation and image visualization in the VR environment were successful in all cases. Based on the computed tomography images, assignment of the CLA lesion to specific lung segments matched between the consulted specialists in only 1 out of the cases. Based on the three 3D VR images, however, the localization matched in 3 of the 5 cases. If the patients would have been operated, adding the 3D VR tool to the preoperative workup would have resulted in changing the surgical strategy (i.e. lobectomy versus segmentectomy) in 4 cases. CONCLUSIONS: This study demonstrated the technical feasibility of a hybridized AI-VR visualization of segment-level lung anatomy in patients with CLA. Further exploration of the value of 3D VR in identifying eligible cases for anatomical segmentectomy is therefore warranted.
引用
- Bakhuis W, Kersten CM, Sadeghi AH, et al. Preoperative visualization of congenital lung abnormalities: hybridizing artificial intelligence and virtual reality. Eur J Cardiothorac Surg. 2022. PMID: 36645240. doi:10.1093/ejcts/ezad014
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。