CNN-Based Detection of Cardiomegaly in Chest X-Rays

Within the framework of this study, an automated system for detecting cardiomegaly using convolutional neural networks (CNNs) has been developed. The goal is to improve the accuracy and objectivity of chest X-ray analysis. Cardiomegaly is a serious condition that requires early diagnosis. However, traditional methods, such as visual examination of X-ray images, have significant drawbacks. They depend on the physician's experience and are prone to subjective interpretation, which can lead to diagnostic errors. The study presents a CNN-based algorithm trained on a large dataset, including X-rays from the NIH Chest X-rays database. To enhance analysis quality, data augmentation and z-normalization techniques are applied, improving the model's robustness to variations in input images. The results demonstrate the effectiveness of the proposed approach, with the model achieving an accuracy of 82%. The advantages of the algorithm include process automation, reduced human factor influence, and the potential for integration into clinical practice—particularly in settings with limited access to specialized studies such as MRI. However, the method has limitations related to its dependence on input data quality. Low-resolution images or artifacts may reduce diagnostic accuracy. The model was trained on over 4,000 annotated images, ensuring the representativeness of the results. The algorithm highlights the potential of artificial intelligence in diagnosing cardiomegaly. The developed system demonstrates high clinical significance, providing consistent results across various types of X-rays. This research contributes to the advancement of automated medical imaging systems and offers a solution to improve the accessibility and accuracy of diagnostics.