A deep-learning model for rapid quantification of cardiothoracic ratio (CTR)

A deep-learning model for rapid quantification of cardiothoracic ratio (CTR)

Introduction : The cardiothoracic ratio (CTR) is the ratio of the diameter of the heart to the diameter of the thorax. A CTR greater than 0.55 may suggest cardiomegaly, which is an indicator of multiple conditions, including hypertension, coronary artery disease, cardiomyopathies, and valvular heart disease. The accurate prediction of an abnormal CTR from posteroanterior chest X- rays (CXRs) aids in the early diagnosis of clinical conditions.


We developed a deep learning-based AI model for automatic CTR calculation that can assist radiologists with the diagnosis of cardiomegaly and optimize the radiology flow.



Results : 

● The model had a sensitivity of 80%, a precision of 99%, F1 score of 88%, and a sensitivity of 100%

● The model performed extremely well in calculating CTR with an MAE of 0.0254 ± 0.06 and an MSE of 0.0016 ± 0.014.

● The sensitivity of the reviewing radiologist in detecting cardiomegaly increased from 40.50% to 88.4% when assisted by the AI-generated CTR.



Conclusion : Our segmentation-based AI model demonstrated high specificity, sensitivity, F1 score, and precision for CTR calculation. The performance of the radiologist on the observer performance test improved significantly with AI assistance.

Share this

Request A Whitepaper

Related Posts