Chronic lung allograft dysfunction (CLAD) is the principal cause of graft failure in lung transplant recipients and prognosis depends on CLAD phenotype. We used a machine learning computed tomography (CT) lung texture analysis tool at CLAD diagnosis for phenotyping and prognostication compared with radiologist scoring.

This retrospective study included all adult first double lung transplant patients (January 2010–December 2015) with CLAD (censored December 2019) and inspiratory CT near CLAD diagnosis. The machine learning tool quantified ground-glass opacity, reticulation, hyperlucent lung and pulmonary vessel volume (PVV). Two radiologists scored for ground-glass opacity, reticulation, consolidation, pleural effusion, air trapping and bronchiectasis. Receiver operating characteristic curve analysis was used to evaluate the diagnostic performance of machine learning and radiologist for CLAD phenotype. Multivariable Cox proportional hazards regression analysis for allograft survival controlled for age, sex, native lung disease, cytomegalovirus serostatus and CLAD phenotype.

88 patients were included (57 bronchiolitis obliterans syndrome (BOS), 20 restrictive allograft syndrome (RAS)/mixed and 11 unclassified/undefined) with CT a median 9.5 days from CLAD onset. Radiologist and machine learning parameters phenotyped RAS/mixed with PVV as the strongest indicator (area under the curve (AUC) 0.85). Machine learning hyperlucent lung phenotyped BOS using only inspiratory CT (AUC 0.76). Radiologist and machine learning parameters predicted graft failure in the multivariable analysis, best with PVV (hazard ratio 1.23, 95% CI 1.05–1.44; p=0.01).

Machine learning discriminated between CLAD phenotypes on CT. Both radiologist and machine learning scoring were associated with graft failure, independent of CLAD phenotype. PVV, unique to machine learning, was the strongest in phenotyping and prognostication.

慢性肺移植物功能障碍 (CLAD) 是肺移植受者移植失败的主要原因，预后取决于 CLAD 表型。与放射科医师评分相比，我们在 CLAD 诊断中使用机器学习计算机断层扫描 (CT) 肺纹理分析工具进行表型分析和预后分析。

这项回顾性研究包括所有患有 CLAD（2019 年 12 月审查）和近 CLAD 诊断的吸气 CT 的成年首例双肺移植患者（2010 年 1 月至 2015 年 12 月）。机器学习工具量化了毛玻璃不透明度、网状结构、超透明肺和肺血管容积 (PVV)。两名放射科医生对磨玻璃影、网状、实变、胸腔积液、空气潴留和支气管扩张进行评分。接受者操作特征曲线分析用于评估机器学习和放射科医师对 CLAD 表型的诊断性能。针对年龄、性别、天然肺病、巨细胞病毒血清状态和 CLAD 表型控制的同种异体移植物存活的多变量 Cox 比例风险回归分析。

包括 88 名患者（57 名闭塞性细支气管炎综合征 (BOS)、20 名限制性同种异体移植综合征 (RAS)/混合和 11 名未分类/未定义），CT 中位时间为 CLAD 发病后 9.5 天。放射科医生和机器学习参数表型 RAS/混合 PVV 作为最强指标（曲线下面积 (AUC) 0.85）。仅使用吸气 CT (AUC 0.76) 的机器学习超透明肺表型 BOS。放射科医生和机器学习参数在多变量分析中预测移植失败，PVV 最好（风险比 1.23，95% CI 1.05–1.44；p=0.01）。

机器学习在 CT 上区分 CLAD 表型。放射科医师和机器学习评分均与移植失败相关，与 CLAD 表型无关。PVV 是机器学习独有的，在表型和预测方面最强。