| 侯哲,王敏红,李志鸿,吴树剑,殷鹏展,周运锋.垂体相关临床及MRI影像组学特征联合列线图鉴别特发性中枢性性早熟与单纯乳房早发育[J].中国医学影像技术,2023,39(8):1133~1138 |
| 垂体相关临床及MRI影像组学特征联合列线图鉴别特发性中枢性性早熟与单纯乳房早发育 |
| Nomogram based on pituitary-related clinical and MRI radiomics features for differential diagnosis of idiopathic central precocious puberty and premature thelarche |
| 投稿时间：2023-04-10 修订日期：2023-06-14 |
| DOI：10.13929/j.issn.1003-3289.2023.08.004 |
| 中文关键词: 垂体 青春期,早熟 磁共振成像 影像组学 |
| 英文关键词:pituitary gland puberty, precocious magnetic resonance imaging radiomics |
| 基金项目: |
|
| 摘要点击次数: 2359 |
| 全文下载次数: 497 |
| 中文摘要: |
| 目的 观察垂体相关临床及MRI影像组学特征联合列线图鉴别特发性中枢性性早熟(ICPP)与单纯乳房早发育(PT)的价值。方法 纳入67例ICPP及51例PT共118例患儿,按照7 ∶ 3比例随机分为训练集(n=83)和验证集(n=35),记录其垂体相关临床资料,以多因素logistic回归分析筛选并建立临床模型。采集垂体MRI,基于矢状位T1WI提取垂体影像组学特征,以最大相关最小冗余、最小绝对收缩和选择算子及多因素logistic回归筛选最佳影像组学特征,构建影像组学模型。联合应用临床、MRI及影像组学特征构建列线图模型。绘制受试者工作特征曲线,评估模型鉴别诊断效能;以决策曲线分析(DCA)观察临床获益度。结果 训练集ICPP与PT患儿年龄、骨龄、体质量、黄体生成素(LH)基础值、卵泡刺激素基础值及垂体高度差异均有统计学意义(P均＜0.05)。骨龄及LH基础值是鉴别ICPP与PT的独立因素(OR=1.807、1.422,P均＜0.05),以之建立的临床模型鉴别训练集、验证集ICPP与PT的曲线下面积(AUC)分别为0.849和0.812。共提取垂体1 781个影像组学特征,于其中筛选出1个形态特征、1个一阶特征及1个灰度区域大小矩阵特征建立影像组学模型,其鉴别训练集和验证集ICPP与PT的AUC分别为0.956和0.947。基于最终得出的2个临床及3个垂体MRI影像组学特征构建的列线图模型鉴别训练集、验证集ICPP与PT的AUC分别为0.981、0.977,均优于临床模型(P均＜0.05),而与影像组学模型差异无统计学意义(P均＞0.05)。一定危险阈值范围内,列线图模型净收益最大。结论 基于垂体相关临床及MRI影像组学特征建立的联合列线图模型用于鉴别ICPP与PT具有较高价值。 |
| 英文摘要: |
| Objective To observe the nomogram based on combined pituitary-related clinical and MRI radiomics features for differential diagnosis of idiopathic central precocious puberty (ICPP) and premature thelarche (PT). Methods Totally 67 children with ICPP and 51 children with PT were enrolled and randomly divided into training set (n=83) or validation set (n=35) at the ratio of 7 ∶ 3. Pituitary-related clinical data were recorded, then a clinical model was established using multivariate logistic regression analysis. Based on sagittal pituitary T1WI, radiomics features were screened, and radiomics model was constructed using maximum relevance minimum redundancy, least absolute shrinkage and selection operator, as well as multivariate logistic regression. Finally nomogram model was constructed combining clinical and MRI radiomics features. Receiver operating characteristic curve was drawn, and decision curve analysis (DCA) was performed to evaluate the differential diagnosis efficacy and clinical benefit of the models. Results In training set, significant differences of age, bone age, body mass, basic luteinizing hormone (LH), basic follicle-stimulating hormone and pituitary height were found between ICPP and PT children (all P＜0.05). Bone age and basic LH were both independent factors for differential diagnosis of ICPP and PT (OR=1.807, 1.422, both P＜0.05), and area under the curve (AUC) of clinical model for differential diagnosis of ICPP and PT was 0.849 and 0.812 in training set and validation set, respectively. Based on 1 781 pituitary radiomics features extracted from T1WI, 1 morphological feature, 1 first-order feature and 1 gray level size zone matrix feature were finally screened out, and AUC of the radiomics model for differential diagnosis of ICPP and PT in training set and validation set was 0.956 and 0.947, respectively. AUC of the nomogram model constructed based on the final screened 2 clinical and 3 radiomics features for differential diagnosis of ICPP and PT in training set and validation set was 0.981 and 0.977, respectively, which were better than those of the clinical model (both P＜0.05) but not significant different from those of the radiomics model (both P＞0.05). DCA showed that within a certain hazard threshold, the nomogram model had the largest net benefit in both training set and validation set. Conclusion Nomogram model based on combined pituitary-related clinical and MRI radiomics features had high application value for differential diagnosis of ICPP and PT. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
