Ajmera Pranav, Kharat Amit, Gupte Tanveer, Pant Richa, Kulkarni Viraj, Duddalwar Vinay, Lamghare Purnachandra
Department of Radiodiagnosis, Dr DY Patil Medical College, Hospital and Research Center, DY Patil Vidyapeeth, DPU, Pune, India.
DeepTek Medical Imaging Pvt. Ltd, Pune, India.
Acta Radiol Open. 2022 Jul 21;11(7):20584601221107345. doi: 10.1177/20584601221107345. eCollection 2022 Jul.
Cardiothoracic ratio (CTR) is the ratio of the diameter of the heart to the diameter of the thorax. An abnormal CTR (>0.55) is often an indicator of an underlying pathological condition. The accurate prediction of an abnormal CTR chest X-rays (CXRs) aids in the early diagnosis of clinical conditions.
We propose a deep learning (DL)-based model for automatic CTR calculation to assist radiologists with rapid diagnosis of cardiomegaly and thus optimise the radiology flow.
The study population included 1012 posteroanterior CXRs from a single institution. The Attention U-Net DL architecture was used for the automatic calculation of CTR. An observer performance test was conducted to assess the radiologist's performance in diagnosing cardiomegaly with and without artificial intelligence assistance.
U-Net model exhibited a sensitivity of 0.80 [95% CI: 0.75, 0.85], specificity >99%, precision of 0.99 [95% CI: 0.98, 1], and a F1 score of 0.88 [95% CI: 0.85, 0.91]. Furthermore, the sensitivity of the reviewing radiologist in identifying cardiomegaly increased from 40.50% to 88.4% when aided by the AI-generated CTR.
Our segmentation-based AI model demonstrated high specificity (>99%) and sensitivity (80%) for CTR calculation. The performance of the radiologist on the observer performance test improved significantly with provision of AI assistance. A DL-based segmentation model for rapid quantification of CTR can therefore have significant potential to be used in clinical workflows by reducing radiologists' burden and alerting to an abnormal enlarged heart early on.
心胸比率（CTR）是心脏直径与胸廓直径之比。异常的CTR（>0.55）通常是潜在病理状况的一个指标。准确预测胸部X光片（CXR）上的异常CTR有助于临床疾病的早期诊断。
我们提出一种基于深度学习（DL）的模型用于自动计算CTR，以协助放射科医生快速诊断心脏肥大，从而优化放射科流程。
研究人群包括来自单一机构的1012张后前位胸部X光片。使用注意力U-Net深度学习架构自动计算CTR。进行了一项观察者性能测试，以评估放射科医生在有无人工智能辅助下诊断心脏肥大的表现。
U-Net模型的灵敏度为0.80 [95%置信区间：0.75, 0.85]，特异性>99%，精确度为0.99 [95%置信区间：0.98, 1]，F1分数为0.88 [95%置信区间：0.85, 0.91]。此外，在人工智能生成的CTR辅助下，阅片放射科医生识别心脏肥大的灵敏度从40.50%提高到了88.4%。
我们基于分割的人工智能模型在CTR计算方面表现出高特异性（>99%）和灵敏度（80%）。在提供人工智能辅助后，放射科医生在观察者性能测试中的表现有了显著改善。因此，一种基于深度学习的CTR快速量化分割模型有很大潜力应用于临床工作流程，可减轻放射科医生的负担并尽早提醒心脏异常增大。
