This study aims to assess the impact of two different image preprocessing methods on the performance of a YOLOv12-based object detection approach for calculating the cardiothoracic index ( \(C_i\) ) on chest X-ray. A total of 470 radiographic images were used to train and evaluate models capable of detecting anatomical structures relevant to the \(C_i\) , specifically measurements AB (heart width) and C (thoracic width). The application of two types of contrast adjustment as a preprocessing technique proved effective. In particular, Model C achieved a mean Average Precision (mAP) of 0.951 when adaptive equalization was applied. The results demonstrate that structure-level detection offers advantages over conventional classification approaches by allowing radiologists to visually validate predictions. Furthermore, this method enables the automatic calculation of the \(C_i\) , facilitating the early identification of cardiomegaly. These findings support the integration of AI-assisted tools into radiological workflows to enhance diagnostic precision and efficiency.