| Resumen |
This work proposes a method for an automatic radiographs classification of the pharyngeal airways using convolutional neural networks (CNNs). The objective of the study is to obtain an algorithm to classify between open or reduced pharyngeal airways. A database of 300 radiographs was used, each class has a cardinality of 150. The images are divided into two categories: normal and reduced pharyngeal airways. An analysis of five convolutional neural network algorithms were performed: LeNet5, AlexNet, ResNet, MobileNet and VGG16. The last one obtains the highest accuracy 89.64%. Additionally, the model was able to identify specific imaging features associated with airway obstruction. The results show that automatic classification of pharyngeal airway radiographs using convolutional neural networks can improve the accuracy of classification compared to traditional manual methods. This approach may have important clinical applications for early diagnosis and treatment monitoring of respiratory diseases such as sleep apnea. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. |
