This diagnostic image is a posteroanterior (PA) chest X-ray of a healthy adult, serving as an anatomical diagram of the thorax. The radiograph displays key structures labeled for educational purposes: the trachea (1) appears as a vertical radiolucent (dark) air column in the midline; the clavicle (2) is a dense radiopaque bone at the superior margin; and the posterior portion of the fourth rib (3) is visible in the upper lung field. Lower down, the right main bronchus (4) branches from the carina. Soft tissue features include the lower margin of the right breast (5). The abdomen-thorax interface highlights the gastric air bubble (6) beneath the left diaphragmatic dome (7). The cardiac silhouette is represented by the left ventricle (8), while the descending aorta (9) and left pulmonary artery (10) are seen in the mediastinum and hilum. The left lung (11) exhibits normal radiolucency. This image demonstrates standard radiographic anatomy, spatial relationships of mediastinal structures, and common landmarks used in clinical imaging and diagnostic triage.

A multi-panel diagnostic image comparison displaying three clinical cases (a, b, c) of posterior-anterior (PA) chest X-rays (CXRs). The image is organized into three columns for each case: 'Original CXR', 'Bounding boxes (GT)' representing the ground truth expert annotations, and 'Bounding boxes (YOLOX)' showing deep learning model predictions with confidence scores. Case (a) demonstrates bilateral apical pneumothorax and basilar pleural effusions. Case (b) illustrates complex pathology including emphysema, aortic arch calcification, a rib fracture, fibrosis, and pleural thickening. Case (c) shows a large pulmonary mass/consolidation in the right lower lobe, emphysema, and calcification. The color-coded bounding boxes categorize 14 different chest abnormalities, including effusion, pneumothorax, consolidation, and mass. This comparison serves as an educational tool for artificial intelligence in radiology, highlighting the model's ability to localize and identify multi-focal pathology while also demonstrating discrepancies such as missed small calcifications or additional diagnoses not in the ground truth.

This diagnostic image is a comparison chart consisting of a 3x4 grid of posteroanterior (PA) chest x-ray (CXR) radiographs, each labeled with a specific clinical finding or diagnosis. The collection serves as a visual guide for thoracic pathology in adult and pediatric patients. Conditions depicted include Atelectasis (linear opacities), Bacterial Pneumonia (localized consolidation), Cardiomegaly (enlarged cardiac silhouette), and Consolidation. The second row demonstrates COVID-19 (bilateral patchy opacities), Pleural Effusion (blunting of the costophrenic angle), a Healthy chest for baseline comparison, and Infiltration. The final row shows Mass, Nodule (small focal opacity), Pneumothorax (characterized by hyperlucency, absence of lung markings, and a visible pleural line indicating lung collapse), and Viral Pneumonia (diffuse, interstitial infiltrates). The educational focus is on differentiating between normal pulmonary anatomy and various respiratory pathologies based on radiodensity, distribution of opacities, and anatomical distortion, making it a valuable tool for radiology training and computer-aided diagnosis (CAD) classification tasks.

This composite image evaluates AI performance in respiratory disease diagnosis using CT scans and Chest X-rays (CXR). Section (a) presents two ROC curves comparing diagnostic accuracy for eight conditions: bronchiectasis, COPD, ILD, lung cancer, pleural effusion, pneumonia, pneumothorax, and tuberculosis. The CT cohort shows a slightly higher mean AUC (0.889) compared to the CXR cohort (0.866). Section (b) demonstrates a case of lung cancer. The axial CT slice shows a hyperdense pulmonary nodule in the right lung (highlighted with a red circle), while the corresponding CXR and AI heatmap show the system's ability to localize the lesion. Section (c) illustrates pneumonia; the axial CT scan displays bilateral patchy infiltrates and ground-glass opacities (indicated by red arrows) in the lower lobes. The accompanying CXR and heatmap visualize the infectious consolidation. This comparison highlights the superior sensitivity of CT for detecting subtle lesions, such as early-stage lung cancer and localized infectious shadows, compared to standard radiography, and demonstrates the utility of AI heatmaps in clinical localization.