Artificial Intelligence-Based Detection of Achalasia on Plain Chest Radiography.

Achalasia is generally considered a progressive condition where esophageal deformity worsens over time; diagnostic delay results in the overdilated and tortuous esophagus.1 Major established tests for achalasia include high-resolution manometry (HRM), esophagogastroduodenoscopy (EGD), and barium esophagogram. However, few noninvasive screening tests are available for this condition, possibly contributing to diagnostic delay.2,3 Plain chest radiographs can capture several characteristic signs of achalasia, such as air-fluid level, opacity, and air esophagogram around the mediastinum, representing the dilated esophagus with food and liquid retention.4-6 These findings suggest the utility of chest radiographs as a noninvasive screening test for achalasia. In recent years, artificial intelligence (AI), developed through deep learning, has been applied to image diagnostics and demonstrates high diagnostic accuracy in several medical conditions.7 Therefore, this retrospective study was aimed to develop and validate a deep learning-based AI model using chest radiographs for achalasia diagnosis.