Deep Learning Model for Prediction of Bronchopulmonary Dysplasia in Preterm Infants Using Chest Radiographs

• DOI: 10.1007/s10278-024-01050-9
• Publication Date: 2024-03-18T18:01:47Z
• Authors: Hao-Yang Chou; Yung-Chieh Lin; Sun-Yuan Hsieh; Hsin-Hung Chou; Cheng-Shih Lai; Bow Wang; Yi-Shan Tsai
• Keywords: Bronchopulmonary Dysplasia; Transfer of learning; Predictive modelling

Abstract Bronchopulmonary dysplasia (BPD) is common in preterm infants and may result pulmonary vascular disease, compromising lung function. This study aimed to employ artificial intelligence (AI) techniques help physicians accurately diagnose BPD a timely efficient manner. retrospective involves two datasets: region segmentation dataset comprising 1491 chest radiographs of infants, prediction 1021 infants. Transfer learning pre-trained machine model was employed for image fusion enhance the performance AI model. The uses transfer achieve dice score 0.960 with $$\le$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 168 h postnatal age. exhibited superior diagnostic compared that experts demonstrated consistent obtained at 24 age, those 25 first use deep on develop an early detection time less than h. Additionally, this model’s according both NICHD Jensen criteria BPD. Results demonstrate surpasses accuracy predicting development