| 魏静,熊曾,易平,李会霞,郭晨旭,郭兴,曹义君,张利华.上中肺叶融合策略CT Delta影像组学评估轻-中度慢性阻塞性肺疾病[J].中国医学影像技术,2026,42(5):716~721 |
| 上中肺叶融合策略CT Delta影像组学评估轻-中度慢性阻塞性肺疾病 |
| CT Delta radiomics based on upper- and middle-lobe fusion strategy for evaluating mild-to-moderate chronic obstructive pulmonary disease |
| 投稿时间：2026-01-05 修订日期：2026-03-23 |
| DOI：10.13929/j.issn.1003-3289.2026.05.013 |
| 中文关键词: 肺疾病,慢性阻塞性 体层摄影术,X线计算机 影像组学 |
| 英文关键词:pulmonary disease, chronic obstructive tomography, X-ray computed radiomics |
| 基金项目:国家重点研发计划项目(2022YFC2010000、2022YFC2010006)。 |
|
| 摘要点击次数: 107 |
| 全文下载次数: 47 |
| 中文摘要: |
| 目的 观察基于上中肺叶融合策略CT Delta影像组学评估轻-中度慢性阻塞性肺疾病(COPD)的价值。方法 前瞻性纳入80例接受低剂量双气相胸部CT的轻-中度COPD患者及63名健康人,按7∶3比例划分训练集与验证集。分别于全肺及5个肺叶水平提取、筛选影像组学特征;以不同分类器基于全肺吸气相和呼气相特征构建模型,以最优分类器构建吸气相、呼气相和Delta影像组学模型,比较其在不同肺区的效能;基于优势肺叶融合和全部肺叶融合策略构建优势气相模型,并与全肺模型进行比较。最后基于最优影像组学模型及临床特征构建联合模型,比较其与最优影像组学模型效能。结果 支持向量机为最优分类器;Delta影像组学模型为优势气相模型;优势肺叶包括左上、右上及右中肺叶。上中肺叶融合Delta影像组学模型评估验证集轻-中度COPD的曲线下面积(AUC)为0.872,高于全部肺叶融合Delta影像组学模型的0.788(P=0.043)和全肺Delta影像组学模型的0.732(P=0.025),而与联合模型差异无统计学意义(AUC=0.881,P＞0.05)。结论 基于上中肺叶融合策略的CT Delta影像组学模型评估轻-中度COPD效能良好。 |
| 英文摘要: |
| Objective To investigate the value of CT Delta radiomics based on upper- and middle-lobe fusion strategy for evaluating mild-to-moderate chronic obstructive pulmonary disease (COPD). Methods Eighty patients with mild-to-moderate COPD and 63 healthy controls who would undergo low-dose dual-phase chest CT were prospectively enrolled and divided into training and validation sets at a ratio of 7∶3. Radiomics features were extracted and selected at levels of the whole-lung and five lung lobes. Models were constructed using different classifiers based on whole-lung inspiratory- and expiratory-phase features. Inspiratory-phase, expiratory-phase and Delta radiomics models were established using the optimal classifier, and their efficiency were compared among different lung regions. The optimal respiratory-phase model was further constructed using dominant-lobe fusion and all-lobe fusion strategies, and its performance was compared with that of whole-lung model. Finally a combined model incorporating the optimal radiomics model and clinical features was developed, and its performance was compared with that of optimal radiomics model. Results Support vector machine was the optimal classifier, Delta radiomics model was the optimal respiratory-phase model. The dominant lobes included left upper, right upper and right middle lobes. The area under the curve (AUC) of upper- and middle-lobe fusion Delta radiomics model in validation set for evaluating mild-to-moderate COPD was 0.872, higher than that of all-lobe fusion Delta radiomics model (AUC: 0.788, P=0.043) and whole-lung Delta radiomics model (AUC: 0.732, P=0.025), but not significantly different with that of the combined model (AUC: 0.881, P＞0.05). Conclusion CT Delta radiomics model based on upper- and middle-lobe fusion strategy had good efficiency for evaluating mild-to-moderate COPD. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
