CARDIO-NET: A DEEP LEARNING -BASED CLINICAL DECISION SUPPORT SYSTEM FOR AUTOMATED CARDIOMEGALY DETECTION FROM CHEST X-RAY IMAGES

Abstract

Cardiovascular diseases are among the leading causes of mortality worldwide, necessitating early and accurate diagnosis [21]. Cardiomegaly is a significant indicator of underlying cardiac disorders and is commonly evaluated using chest X-ray imaging [6], [7]. However, manual interpretation of chest X-rays is time-consuming and requires expert radiologists. This paper presents Cardio-Net, a deep learning-based clinical decision support system for automated cardiomegaly detection using Deep Learning techniques [8], [13]. The proposed system employs a DenseNet-based model for classification [1], combined with image preprocessing using CLAHE to enhance image quality. Additionally, the system integrates cardiothoracic ratio (CTR) estimation [6], [7] and Grad-CAM visualization [3] for improved clinical interpretability. Experimental results demonstrate the effectiveness of the proposed  system in assisting healthcare professionals in early diagnosis and decision-making.