智能自动配准:应用于肝脏超声融合成像的可行性与效率评估
前瞻性纳入2017年5月至12月30例患者30个肝脏肿瘤,随机比较基于肝血管与基于肝表面的两种智能自动配准及手动配准的配准成功率和效率;基于肝血管的自动配准成功率为80%,低于基于肝表面及手动配准的96.67%,但差异无统计学意义(P=0.125);基于肝血管的自动配准效率优于其他两种(P<0.05),一步成功率(53.33%,16/30)高于其他两种(P<0.05);肝血管显示为3级的病灶中,基于血管的自动配准成功率(0%)低于基于肝表面及手动配准(100%)(P=0.031)。
要点
- 基于肝血管的智能自动配准对肝血管显示清晰的肝脏肿瘤超声-CT/MR融合成像可行且高效。
- 肝血管显示为3级的病灶中,基于血管的自动配准成功率为0%,低于基于肝表面及手动配准的100%(P=0.031)。
- 肝血管显示不佳时,基于肝表面的自动配准和手动配准可作为有效补充。
核心结论
- 对于肝血管显示清晰的患者,基于肝血管的智能自动配准是超声-CT/MR融合成像中可行且高效的配准方法;基于肝表面和手动配准可作为肝血管显示不佳病例的有效补充。
研究设计与样本
- 前瞻性研究,2017年5月至2017年12月纳入30例患者30个肝脏肿瘤,随机进行基于肝血管的自动配准、基于肝表面的自动配准和手动配准,比较配准成功率和效率。
诊断标准与关键数值
- 基于肝血管的自动配准成功率80%,基于肝表面及手动配准成功率96.67%,差异无统计学意义(P=0.125)。
- 基于肝血管的自动配准效率优于基于肝表面及手动配准(P<0.05)。
- 基于肝血管的自动配准一步成功率53.33%(16/30),高于其他两种配准(P<0.05)。
- 肝血管显示为3级的病灶中,基于血管的自动配准成功率0%,低于基于肝表面及手动配准的100%(P=0.031)。
局限性
- 原文未报告局限性。
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)。
英文原文(摘要)· 点击展开
PURPOSE: To evaluate the feasibility and efficiency between the two intelligent auto-registrations (based on hepatic vessels or based on liver surface) and manual registration for US-CT/MR fusion imaging of liver tumours.
METHODS: From May 2017 to December 2017, 30 patients with 30 liver tumours were enrolled in this prospectively study. Two intelligent auto-registrations (based on hepatic vessels or based on liver surface) and manual registration were randomly performed, the registration success rate and efficiency were compared.
RESULTS: In terms of success rate, auto-registrations based on the hepatic vessels (80%) was lower than auto-registration base on liver surface and manual registration (96.67%), but with no statistical difference (P = 0.125). In comparison of the registration efficiency, the efficiency of the auto-registration based on the hepatic vessels was superior to auto-registration based on liver surface and manual registration (P < 0.05). The one-step success rate of auto-registration based on the hepatic vessels (53.33%, 16/30) was higher than that of other two registrations (P < 0.05). Stratified analysis showed that, for the lesion with display of hepatic vessels in grade 3, the success rate of auto-registration based on vessels (0%) was lower than that of auto-registration based on liver surface and manual registration (100%) (P = 0.031).
CONCLUSION: Intelligent auto-registration based on hepatic vessels is a feasible and efficient registration method for US-CT/MR fusion imaging of liver tumours for the patients with clear hepatic vessels. The auto-registration based on liver surface and manual registration can be an effective supplement for cases with poor hepatic vessels display.
引用
- Zeng Q, Yan R, Zhang L, et al. Intelligent automatic registration: is it feasible and efficient for application of ultrasound fusion imaging in liver? Abdom Radiol (NY). 2024. PMID: 39613872. doi:10.1007/s00261-024-04724-8
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。