影像数据联合临床-实验室数据构建融合模型预测自发性脑出血预后  

Fusion model of imaging data combined with clinical and laboratory data can predict the prognosis of spontaneous intracerebral hemorrhage

作　　者：彭成远 曾春意 吴宗山 周坤生 PENG Chengyuan;ZENG Chunyi;WU Zongshan;ZHOU Kunsheng(Department of Medical Imaging,Lu'an Hospital of Anhui Medical University,Lu'an 237000,China;Department of Magnetic Resonance,Wanxi Health Vocational College Affiliated Hospital,Lu'an 237008,China)

机构地区：[1]安徽医科大学附属六安医院(六安市人民医院)医学影像科,安徽六安237000 [2]皖西卫生职业学院附属医院磁共振室,安徽六安237008

出　　处：《分子影像学杂志》2025年第1期91-96,共6页Journal of Molecular Imaging

基　　金：安徽省高等学校科学研究项目(2023AH053358)。

摘　　要：目的分析影像数据、临床-实验室数据构建融合模型预测自发性脑出血患者预后的价值。方法回顾性收集2021年9月~2023年3月六安市人民医院头颅CT诊断为自发性脑出血的患者363例,使用改良Rankin量表对患者出院6月后的功能恢复情况进行评价,根据结果分为预后良好组(n=175)和预后不良组(n=188)。记录患者年龄、性别、入院中性粒细胞、淋巴细胞计数及其各百分比、D-二聚体以及首次头颅CT平扫中勾画并提取的血肿影像数据资料,采用多元Logistic回归分别构建临床-实验室数据模型、影像数据模型和融合模型。结果两组GCS评分、中性粒细胞数量及其百分数、淋巴细胞百分数的差异有统计学意义(P<0.001);两组脑室有无积血、血肿球形度、表面积以及费雷特直径等影像学资料的差异有统计学意义(P<0.001)。临床-实验室数据模型、影像数据模型以及融合模型的AUC分别为0.82(95%CI:0.78~0.86)、0.80(95%CI:0.75~0.84)和0.86(95%CI:0.82~0.89)。Delong检验显示融合模型性能与单一临床-实验室数据模型、影像数据模型的差异有统计学意义(P<0.05)。结论影像数据联合临床、实验室数据构建的融合模型对预测自发性脑出血的不良预后有较显著价值。Objective To analyze the value of imaging data,clinical-laboratory data model and fusion model in predicting prognosis of patients with spontaneous intracerebral hemorrhage.Methods A total of 363 patients diagnosed with spontaneous cerebral hemorrhage by head CT in Lu'an People's Hospital from September 2021 to March 2023 were retrospectively collected.The functional recovery of patients 6 months after discharge was evaluated using the modified Rankin scale.According to the results,the patients were divided into two groups:the good outcome(n=175)and the poor outcome(n=188).Clinical data such as age and gender,neutrophil and lymphocyte counts and their percentages,D-dimer,and hematoma image data outlined and extracted from the first head CT scan on admission were recorded.Multiple logistic regression method was used to construct clinical-laboratory data model,image data model and fusion model,respectively.Results Clinical and laboratory data of statistical significance between the two groups included GCS score,number and percentage of neutrophils,and percentage of lymphocytes(P<0.001).Imaging data included ventricle presence or absence of hematoma,hematoma sphericity,surface area,and feret diameter(P<0.001).The AUC of clinical-laboratory data model,image data model and fusion model were 0.82(95%CI:0.78-0.86),0.80(95%CI:0.75-0.84)and 0.86(95%CI:0.82-0.89).Delong test showed that the performance of fusion model was significantly different from that of single clinical-laboratory data model and imaging data model(P<0.05).Conclusion The fusion model of imaging data combined with clinical and laboratory data has significant value in predicting the prognosis of spontaneous intracerebral hemorrhage.

关 键 词：横断面研究 头颅CT 自发性脑出血 预后 

分 类 号：R743.34[医药卫生—神经病学与精神病学] R816.1[医药卫生—临床医学]