Pulmonary hypertension (PH) is a progressive cardiovascular pathology that is typified by high morbidity and mortality rates, especially when it has not been identified. Invasion or resource-consuming modalities such as right heart catheterization and echocardiography are conventional diagnostic modalities but accurate. The recent developments of deep learning and more specifically, convolutional neural networks (CNNs) have been found to be promising as techniques of detecting diseases through medical imaging in an undeterred, fast, and dependable manner. The current paper suggests an automated diagnostic model based on CNN structures that will be used to detect PH on the basis of the radiographs of the chest. In particular, the model uses a DenseNet -121 backbone, which was finetuned on a carefully curated dataset, resulting in a high level of diagnostic accuracy. Compared experiments prove the high effectiveness of the suggested system compared to the traditional machine-learning methods. The article attempts to eliminate the gap between AI-based healthcare applications and clinical diagnostic needs.