Cardiomegaly Detection from Chest X-Rays Using InceptionV3: A Deep Learning Approach

This paper applies the deep learning method to automatically identify cardiomegaly in chest X-ray images based on the InceptionV3 architecture. A convolutional neural network (CNN) model has been trained and verified with balanced dataset, the accuracy of the model is 99.21% and the F1 score is 99% on the test set. Moreover, the model has also shown high generalization ability since the test loss is low with a value of 0.0242. These findings point out that the model under discussion is valid and efficient in clinical practice. Also, visual attention mechanisms were introduced in order to make it more interpretable so that clinicians could focus on the diagnostically meaningful areas. This study indicates the possibilities of the InceptionV3 in the analysis of medical images and helps move toward AI-aided medical diagnosis.