个性化肾上腺体积参考范围及全自动深度学习筛查工具的研发
本研究基于低剂量非增强胸部CT开发全自动深度学习工具筛查肾上腺体积异常,并建立个性化参考范围。18,538名成人中健康者7,907名(42.65%),肾上腺体积右侧3.24 cm³(2.49-4.06)、左侧4.51 cm³(3.72-5.39)。分割模型Dice相似系数0.926、ICC 0.954(95%CI:0.946-0.961),在糖尿病和高血压验证集(n=3,266)中检出62.31%异常病例,肾上腺增生(n=240)中检出率77.09%。
要点
- 低剂量胸部CT可全自动分割测量肾上腺体积,Dice相似系数0.926、ICC 0.954(95%CI:0.946-0.961)。
- 肾上腺体积右侧3.24 cm³(2.49-4.06)、左侧4.51 cm³(3.72-5.39),并随年龄、性别、血压、血糖等因素变化。
- 基于高斯混合模型的异常检测在糖尿病和高血压验证集(n=3,266)检出62.31%异常,肾上腺增生(n=240)检出77.09%。
核心结论
- 本研究建立了基于低剂量CT的个性化肾上腺体积参考范围和全自动深度学习测量工具,筛查工具显示出辅助识别肾上腺异常的潜力,并可能为未来临床评估和应用提供方法学基础。
研究设计与样本
- 研究纳入常规健康体检中接受低剂量非增强胸部CT的≥18岁受试者,按特定标准提取健康参考、高血压/糖尿病验证和肾上腺异常验证三个数据集。从数据集中随机抽取400幅低剂量胸部CT图像训练基于nnU-Net的深度学习模型,550幅用于验证。采用多变量回归和限制性立方样条评估年龄、性别、体表面积和血液标志物等因素对肾上腺体积的影响,用分位数回归模型建立个体化参考范围,并开发基于高斯混合模型的异常检测系统用于异常体积筛查,在高血压、糖尿病和肾上腺增生数据集上测试。
诊断标准与关键数值
- 肾上腺体积范围为右侧3.24 cm³(2.49-4.06)、左侧4.51 cm³(3.72-5.39)。参数调整后的参考范围下限=-2.739+0.016×年龄+0.431×性别+0.032×腰围(WC)+0.874×体表面积(BSA)+0.007×舒张压(DBP)+0.081×甘油三酯(TG)-0.445×高密度脂蛋白胆固醇(HDL)+0.306×空腹血糖(FPG);上限=-3.549+0.031×年龄+0.339×性别+0.030×腰围+2.721×BSA+0.020×DBP+0.004×TG-1.244×HDL+0.447×FPG。分割模型Dice相似系数0.926、ICC 0.954(95%CI:0.946-0.961)。在糖尿病和高血压验证集(n=3,266)中筛查检出62.31%异常病例,在肾上腺增生(n=240)中检出率77.09%。
临床应用与阅片要点
- 作者认为该筛查工具显示出辅助识别肾上腺异常的潜力,并可能为未来临床评估和应用提供方法学基础。
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
OBJECTIVE: This study aims to develop a low-dose CT-based, fully automated deep learning tool for screening adrenal gland volume abnormalities and establishing personalized reference ranges to assist in diagnosing adrenal diseases.
METHODS: This study included subjects (≥18 years) who underwent low-dose non-contrast chest CT during routine health check-ups, and three datasets were extracted based on specific criteria: healthy reference, hypertension/diabetes validation, and adrenal abnormality validation. Randomly sampled from these datasets 400 low-dose chest CT images were used to train the nnU-Net-based deep learning model, and 550 images were used for validation. Multivariable regression and restricted cubic splines (RCS) assessed the effects of factors like age, sex, body surface area, and blood markers on adrenal volume. A quantile regression model was used to create individualized reference ranges. A GMM-based anomaly detection system was developed for abnormal volume screening, tested on datasets for hypertension, diabetes, and adrenal hyperplasia.
RESULTS: Among the 18,538 adults, 7,907 (42.65 %) were healthy. Adrenal volume ranges ere 3.24 cm3 (2.49-4.06) on the right and 4.51 cm3 (3.72-5.39) on the left. Adrenal volume correlated with age, sex, blood pressure, and glucose (p < 0.05). The parameter-adjusted reference ranges were defined as: Lower limit = -2.739 + 0.016 × Age + 0.431 × Sex + 0.032 × Waist circumference (WC) + 0.874 × Body surface area (BSA) + 0.007 × Diastolic blood pressure (DBP) + 0.081 × Triglycerides (TG) - 0.445 × High-density lipoprotein cholesterol (HDL) + 0.306 × Fasting plasma glucose (FPG);Upper limit = - 3.549 + 0.031 × Age + 0.339 × Sex + 0.030 × Waist + 2.721 × BSA + 0.020 × DBP + 0.004 × TG - 1.244 × HDL + 0.447 × FPG.The segmentation model achieved a Dice Similarity Coefficient of 0.926 and an ICC of 0.954(95 % CI: 0.946-0.961).In validation sets for diabetes and hypertension (n = 3,266), the screening detected 62.31 % abnormal cases, while in adrenal hyperplasia (n = 240), the rate was 77.09 %.
CONCLUSION: This study developed individualized adrenal gland volume reference ranges and a low-dose CT-based deep learning tool for automated measurement. The screening tool shows potential to assist in identifying adrenal abnormalities and may provide a methodological basis for future clinical evaluation and application.
引用
- Li Y, Zhao Y, Li C, et al. Personalized adrenal gland volume reference ranges and development of a fully automated deep learning screening tool. Eur J Radiol. 2025. PMID: 41352231. doi:10.1016/j.ejrad.2025.112591
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。