深度学习图像重建检测肝脏病灶的最低临床可达到剂量:体模与患者研究
本研究通过体模与前瞻性单中心患者研究,比较约4.5 mGy超低剂量深度学习图像重建(DLIR)与标准剂量(10-15 mGy)自适应统计迭代重建-V(ASIR-V)检测局灶性肝脏病灶(FLL)的性能。体模研究中DLIR在4.5 mGy下所有定量指标均优于10 mGy ASIR-V;患者研究中DLIR图像质量更优、FLL检出率更高(93.0% vs 77.5%),敏感度90.1%、特异度75.0%。结论认为该方案在大幅降低辐射剂量的同时提供更优的FLL检测性能。
要点
- 4.5 mGy DLIR较10-15 mGy ASIR-V降低辐射剂量50%-70%,且FLL检出率更高(93.0% vs 77.5%)。
- DLIR的FLL检测敏感度90.1%、特异度75.0%,均高于ASIR-V(P<0.001)。
- 1.25 mm层厚DLIR主观图像质量评分3.9±0.6,优于ASIR-V的2.2±0.4(P<0.001)。
核心结论
- 4.5 mGy DLIR在显著降低辐射剂量的同时,FLL检测性能优于10-15 mGy ASIR-V,该方案为FLL筛查和随访提供了安全且准确的选择。
研究设计与样本
- 体模研究使用Gammex CT体模(模拟碘浓度2.0-20 mg/mL的FLL及正常肝密度1.06 g/cm3),分别以DLIR(4.5 mGy)和ASIR-V(10及15 mGy)扫描;前瞻性单中心研究纳入84名受试者(平均年龄64±12岁,IQR 60-69岁;男性48名),行三期上腹部CT(目标剂量4.5 mGy/期),图像以DLIR和ASIR-V重建,由两名放射科医师盲法评估。
关键影像表现/征象
- 体模研究中,DLIR在4.5 mGy下所有定量指标均优于ASIR-V 10 mGy(P<0.001)。
- 体模研究中,DLIR在4.5 mGy下噪声和NPSpeak优于ASIR-V 15 mGy(P<0.05)。
- 患者研究中,DLIR的1.25 mm层厚主观图像质量评分为3.9±0.6,优于ASIR-V的2.2±0.4(P<0.001)。
诊断标准与关键数值
- 患者研究共检出71个FLL(平均大小12.8±10 mm;良性55个,恶性16个),DLIR的FLL检出率为93.0%,ASIR-V为77.5%(P<0.001)。
- DLIR检测FLL的敏感度为90.1%、特异度为75.0%,两者均高于ASIR-V(P<0.001)。
- 可检出性参考标准包括组织病理学、3个月标准剂量CT或MRI。
- 中位CTDIvol为4.64 mGy,较标准剂量降低50%至70%。
临床应用与阅片要点
- 作者认为,该4.5 mGy DLIR方案为FLL筛查和随访提供了安全且准确的选择。
- DLIR在超低剂量下图像噪声和NPSpeak可优于15 mGy ASIR-V,提示其可用于降低上腹部CT辐射剂量。
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
RATIONALE AND OBJECTIVES: To investigate the performance of deep learning image reconstruction (DLIR) at an ultra-low dose of approximately 4.5 mGy for detecting focal liver lesions (FLLs), in comparison with adaptive statistical iterative reconstruction-V (ASIR-V) at standard doses (10-15 mGy), through both phantom and prospective patient studies.
MATERIALS AND METHODS: A Gammex CT phantom (simulating FLLs with iodine concentrations 2.0-20 mg/mL and normal liver density 1.06 g/cm3) was scanned using DLIR (4.5 mGy) and ASIR-V (10 and 15 mGy). Quantitative metrics included image noise, signal-to-noise ratio, contrast-to-noise ratio, noise power spectrum peak (NPSpeak), and detectability index. In a prospective single-center study, 84 participants (mean age 64 ± 12 years, IQR 60-69 years; 48 males) underwent triple-phase upper-abdominal CT (target dose: 4.5 mGy/phase). Images were reconstructed with DLIR and ASIR-V. Two radiologists blindly evaluated image quality (5-point scale), FLL detectability, sensitivity, and specificity. The reference standard for detectability included histopathology, 3-month standard-dose CT, or MRI.
RESULTS: In the phantom study, DLIR at 4.5 mGy outperformed ASIR-V at 10 mGy across all quantitative metrics (P < 0.001) and exceeded ASIR-V at 15 mGy in noise and NPSpeak (P < 0.05). Clinically, 71 FLLs (mean size 12.8 ± 10 mm; 55 benign, 16 malignant) were identified. The median CTDIvol was 4.64 mGy (50% to 70% lower than standard doses). DLIR showed superior qualitative image quality vs. ASIR-V (1.25 mm slices: 3.9 ± 0.6 vs. 2.2 ± 0.4, P < 0.001) and higher FLL detection rate (93.0% vs. 77.5%, P < 0.001), with sensitivity 90.1% and specificity 75.0% (both higher than ASIR-V, P < 0.001).
CONCLUSION: DLIR at 4.5 mGy achieves substantial radiation dose reduction while providing superior FLL detection performance compared to ASIR-V at 10-15 mGy. This protocol offers a safe and accurate option for FLL screening and follow-up.
引用
- Pan Z, Xu M, Qu T, et al. Minimum Clinically Achievable Dose for Detecting Liver Lesions Using Deep Learning Image Reconstruction: A Phantom and Patient Study. Acad Radiol. 2026. PMID: 41781262. doi:10.1016/j.acra.2026.02.022
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。