Diagnostic accuracy of transthoracic ultrasound for detection of pneumothorax, hemothorax, and lung contusion in trauma patients: comparison with chest radiography Authors: Mohamed Fawzy Tantawy, Hady Saleh Abou-Ashour, Mahmoud Mousa Elhabashy, Ali Hamouda Ali, Rehab Mohammed Habib Published: 2026 ABSTRACT BackgroundThoracic trauma is a major cause of morbidity and mortality in trauma patients. Rapid detection of pneumothorax, hemothorax, and lung contusion is critical for timely management. This study aimed to evaluate the diagnostic accuracy of transthoracic ultrasound compared with chest radiography in thoracic trauma patients using chest computed tomography (CT) as the reference standard.MethodsThis prospective diagnostic accuracy study was conducted in the Emergency Department of Menoufia University Hospital between July 2023 and December 2024. Adult trauma patients with suspected thoracic injury were consecutively enrolled. All patients underwent transthoracic ultrasound, chest radiography, and chest CT. Diagnostic performance of ultrasound and chest radiography for detecting pneumothorax, hemothorax, and lung contusion was assessed using CT as the reference standard. Sensitivity, specificity, predictive values, accuracy, and Cohen’s kappa agreement were calculated.ResultsA total of 76 trauma patients were included in the final analysis (mean age 42.5 ± 12.7 years; 77.6% male), with blunt trauma accounting for 78.9% of injuries. CT identified pneumothorax in 45 patients (59.2%), hemothorax in 53 (69.7%), and lung contusion in 57 (75.0%). Transthoracic ultrasound detected pneumothorax in 41 patients (53.9%), hemothorax in 51 (67.1%), and lung contusion in 49 (64.5%), whereas chest radiography detected these injuries in 31 (40.8%), 33 (43.4%), and 14 patients (18.4%), respectively. Ultrasound demonstrated higher diagnostic performance than chest radiography across all injury types. For pneumothorax detection, ultrasound showed 91% sensitivity and 95% diagnostic accuracy compared with 69% sensitivity and 82% accuracy for chest radiography. For hemothorax detection, ultrasound achieved 96% sensitivity and 97% accuracy versus 62% sensitivity and 74% accuracy for chest radiography. For lung contusion detection, ultrasound demonstrated 86% sensitivity and 89% accuracy, whereas chest radiography showed only 25% sensitivity and 43% accuracy. Agreement with CT was strong for ultrasound (κ = 0.75–0.94) but lower for chest radiography (κ = 0.14–0.64).ConclusionTransthoracic ultrasound demonstrated high diagnostic performance and good agreement with chest CT for detecting traumatic thoracic injuries and outperformed chest radiography in this study. However, given the single-center design and relatively limited sample size, larger multicenter studies are warranted to validate these findings. Introduction Thoracic trauma remains a major cause of preventable morbidity and mortality in injured patients, largely due to rapidly evolving complications such as pneumothorax, hemothorax, and lung contusion. Failure to promptly identify these injuries may result in respiratory compromise, hemodynamic instability, and increased mortality. Consequently, international trauma guidelines emphasize rapid bedside assessment and timely imaging to facilitate early diagnosis while minimizing delays to definitive management and unnecessary radiation exposure [1]. Chest CT is widely regarded as the reference standard for diagnosing thoracic injuries because of its high sensitivity for both pleural and parenchymal abnormalities [2]. However, the routine use of CT during the initial trauma evaluation may be limited by patient instability, the need to transport critically ill patients away from the resuscitation area, and logistical constraints in resource-limited EDs [3]. As a result, supine chest radiography continues to be commonly used during early trauma assessment, despite its well-recognized limitations, including reduced sensitivity for occult pneumothorax, small hemothorax, and lung contusion [4]. Point-of-care transthoracic ultrasound, incorporated into the extended Focused Assessment with Sonography in Trauma (E-FAST), has become an integral component of modern trauma resuscitation [3, 5, 6]. The diagnostic performance of transthoracic ultrasound in traumatic pneumothorax has been extensively investigated. International evidence-based recommendations and critical care consensus statements support the use of lung ultrasound for early detection of pneumothorax and pleural effusions, particularly in hemodynamically unstable patients or when CT is not immediately available [7, 8]. Recent meta-analyses consistently demonstrate that lung ultrasound has significantly higher sensitivity than chest radiography for pneumothorax detection, while maintaining similarly high specificity [9, 10]. Despite the increasing integration of lung ultrasound into trauma protocols, comparative data evaluating its diagnostic performance against chest radiography across multiple thoracic injuries remain limited. Most previous studies have focused primarily on pneumothorax detection, whereas thoracic trauma often involves combined injuries such as hemothorax and lung contusion with differing diagnostic detectability. Therefore, this study aimed to evaluate the diagnostic accuracy of transthoracic ultrasound compared with chest radiography for detecting pneumothorax, hemothorax, and lung contusion in trauma patients, using chest CT as the reference standard. Methods Study design and setting This prospective diagnostic accuracy study was conducted in the Emergency Department of the Menoufia University Hospital, a tertiary care trauma center, between July 2023 and December 2024. The study aimed to evaluate the diagnostic performance of transthoracic ultrasound compared with chest radiography for detecting acute traumatic lung injuries, using chest CT as the reference standard. Ethical considerations Ethical approval was obtained from the Institutional Review Board of Menoufia University Faculty of Medicine (Approval No.: 2/2023 SURG 17). Written informed consent was secured from all participants or their legally authorized representatives prior to enrolment. All procedures were conducted in accordance with the principles of the Declaration of Helsinki, and patient confidentiality was strictly maintained throughout the study. Study population Consecutive adult trauma patients presenting with suspected thoracic injury were screened for eligibility to minimize selection bias. Patients were included if they were aged ≥ 18 years, presenting with blunt or penetrating thoracic trauma or had clinical suspicion of acute traumatic lung injury and underwent transthoracic ultrasound, chest radiography, and chest CT as part of their diagnostic evaluation. Patients were excluded if they required immediate life-saving intervention before completion of imaging, had prior chest tube placement before ED evaluation, had pre-existing chronic pleural or pulmonary disease that could confound imaging interpretation. Pregnant females were also excluded and any patients who declined participation in the study. Study procedure All included patients underwent comprehensive clinical assessment according to ATLS guidelines including airway, breathing, circulation, disability and exposure (ABCDE) approach, initial resuscitative interventions were performed simultaneously without delay. Clinical and demographic data, including age, gender, and mechanism of thoracic trauma (blunt or penetrating), were recorded for all patients. Bedside transthoracic ultrasound was performed during or immediately after the primary survey. Supine anteroposterior chest radiographs were obtained during the primary trauma evaluation whenever feasible according to institutional protocols. Chest CT examinations were performed once patients were clinically stable for transport using a 32-slice SOMATOM go.Up multidetector CT scanner (Siemens Healthineers, Germany) according to the institutional trauma imaging protocol. Non-contrast axial images extending from the thoracic inlet to the upper abdomen were acquired according to the institutional trauma protocol, with multiplanar coronal and sagittal reconstructions. Chest CT served as the reference standard for confirming or excluding pneumothorax, hemothorax, and lung contusion. All transthoracic ultrasound examinations were performed by a senior emergency physician with approximately 10 years of experience in point-of-care ultrasound and transthoracic ultrasound. The operator had completed specialized training in transthoracic ultrasound and regularly served as an instructor in accredited POCUS training. All ultrasound examinations were performed and documented before the patients underwent chest CT. Therefore, the ultrasound operator was blinded to the CT findings at the time of examination, minimizing interpretation bias. Chest radiographs and chest CT scans were interpreted independently by experienced radiologists who were blinded to the ultrasound findings. Transthoracic ultrasound Ultrasound examinations were performed using a SonoScape S30 ultrasound system (SonoScape Medical Corp., Shenzhen, China). Two probes were used, a high-frequency linear probe (7.5–10 MHz) for pleural evaluation and a curvilinear probe (2-5 MHz) for deeper thoracic and parenchymal assessment. Each hemithorax was examined systematically in the anterior, lateral, and posterior thoracic zones whenever patient positioning allowed. The anterior chest was primarily assessed for pneumothorax, whereas the dependent lateral and posterolateral regions were examined for pleural fluid and lung parenchymal abnormalities. Lung ultrasound in thoracic trauma relies on identifying alterations in the pleural line, pleural sliding, pleural fluid, and subpleural parenchymal lesions. Pneumothorax was diagnosed by the absence of lung sliding and B-lines, with additional assessment for lung pulse and lung point. The presence of lung point was considered confirmatory when identified. Current reviews describe this combination of signs as the standard sonographic approach for traumatic pneumothorax evaluation [7, 11]. Hemothorax was diagnosed by the presence of an anechoic or hypoechoic pleural fluid collection in the dependent thoracic recesses, with or without internal echoes suggestive of blood. Evaluation was focused on the costophrenic and pleural-dependent areas, where ultrasound is particularly sensitive for detecting pleural fluid in trauma patients [11, 12]. Lung contusion was diagnosed by the presence of focal or multifocal subpleural consolidations, focal B-lines or confluent B-lines, pleural line irregularity, and localized alveolar–interstitial syndrome adjacent to the pleural surface. Recent reviews and clinical studies of thoracic trauma describe these findings as characteristic of lung contusion on lung ultrasound [13]. Outcome measures The primary outcome was the diagnostic accuracy of transthoracic ultrasound for detecting pneumothorax, hemothorax and lung contusion using chest CT as the reference standard. The secondary outcomes included comparison of the diagnostic performance of transthoracic ultrasound and chest radiography, evaluation of agreement between each imaging modality and CT using Cohen’s kappa statistics, and determination of the detection rate of thoracic injuries by each imaging modality. Sample size calculation The sample size was calculated a priori based on the primary objective of evaluating the diagnostic accuracy of transthoracic ultrasound. A previous study by Sabri et al. (2018) reported a diagnostic accuracy of 93.4% for ultrasound in detecting pleural injuries in patients with blunt chest trauma [14]. Using a 95% confidence level (Z = 1.96) and an absolute precision (margin of error) of 5%, the minimum required sample size was calculated using the standard formula for estimating a single proportion, yielding a target sample size of 95 patients. Patients were enrolled consecutively during a predefined 18-month study period (July 2023 to December 2024). A total of 76 eligible patients were included (Fig. 1). The difference between the achieved and calculated sample sizes is acknowledged in the Strengths and Limitations section. Fig. 1STARD flow diagram illustrating patient enrollment and inclusion in the final analysis STARD flow diagram illustrating patient enrollment and inclusion in the final analysis Statistical analysis Statistical analysis was performed using IBM SPSS Statistics for Windows, version 28.0 (IBM Corp., Armonk, NY, USA; released in 2021). Continuous variables were presented as mean ± standard deviation (SD). Categorical variables were expressed as frequency and percentage. The diagnostic performance of transthoracic ultrasound and chest radiography was evaluated using chest CT as the reference standard. For each imaging modality, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall diagnostic accuracy were calculated for the detection of pneumothorax, hemothorax, and lung contusion, based on complete 2 × 2 contingency tables (true positive, false positive, false negative, and true negative counts) constructed separately for each injury and imaging modality. 95% confidence intervals (CIs) for these diagnostic accuracy measures were calculated using the exact (Clopper-Pearson) binomial method. Paired comparisons between transthoracic ultrasound and chest radiography were performed using McNemar’s chi-square test using paired binary outcomes. The test was applied separately for each injury to compare the paired diagnostic classifications of the two imaging modalities. A test could not be computed when specificity was 100% for both modalities, since no discordant pairs were available. Agreement between transthoracic ultrasound and CT, and between chest radiography and CT, was assessed using Cohen’s kappa (κ) coefficient, with corresponding 95% CIs calculated using the standard asymptotic (large-sample) method. Kappa values were interpreted as follows: < 0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.00, almost perfect agreement. A p-value < 0.05 was considered statistically significant. Results During the study period, 90 trauma patients were screened for eligibility. After applying the predefined exclusion criteria, 14 patients were excluded, and 76 patients were included in the final analysis. The patient selection process is illustrated in the STARD flow diagram (Fig. 1). The mean age of the patients was 42.5 ± 12.7 years (range: 24–64 years). The majority of participants were males (77.6%), while 22.4% were females. Blunt trauma represented the most common mechanism of injury, accounting for 78.9% of cases, whereas 21.1% sustained penetrating trauma. Regarding injury distribution, right-sided thoracic injuries were observed in 43.4% of patients, left-sided injuries in 40.8%, and bilateral injuries in 15.8%. The baseline demographic and clinical characteristics of the study population are summarized in Table 1. Table 1Baseline characteristics of the study population (n = 76)VariableValueAge, years (mean ± SD)42.5 ± 12.7Male sex, n (%)59 (77.6)Female sex, n (%)17 (22.4)Blunt trauma, n (%)60 (78.9)Penetrating trauma, n (%)16 (21.1)Right-sided injury, n (%)33 (43.4)Left-sided injury, n (%)31 (40.8)Bilateral injury, n (%)12 (15.8)SD = standard deviation Table 1Baseline characteristics of the study population (n = 76)VariableValueAge, years (mean ± SD)42.5 ± 12.7Male sex, n (%)59 (77.6)Female sex, n (%)17 (22.4)Blunt trauma, n (%)60 (78.9)Penetrating trauma, n (%)16 (21.1)Right-sided injury, n (%)33 (43.4)Left-sided injury, n (%)31 (40.8)Bilateral injury, n (%)12 (15.8)SD = standard deviation Baseline characteristics of the study population (n = 76) SD = standard deviation Imaging findings across diagnostic modalities Chest CT identified pneumothorax in 45 patients (59.2%), hemothorax in 53 patients (69.7%), and lung contusion in 57 patients (75.0%). Transthoracic ultrasound detected pneumothorax in 41 patients (53.9%), hemothorax in 51 patients (67.1%), and lung contusion in 49 patients (64.5%). In contrast, chest radiography identified pneumothorax in 31 patients (40.8%), hemothorax in 33 patients (43.4%), and lung contusion in only 14 patients (18.4%). No thoracic abnormalities were detected in 10 patients (13.2%) by ultrasound, 23 patients (30.3%) by chest radiography, and 5 patients (6.6%) by chest CT imaging. These findings demonstrate that transthoracic ultrasound detected thoracic injuries more frequently than chest radiography, particularly in the case of lung contusion. The distribution of thoracic injuries detected by transthoracic ultrasound, chest radiography, and chest CT is presented in Table 2. Table 2Diagnostic findings in transthoracic ultrasound, chest X-ray and chest CT in the studied patients (n = 76)FindingTransthoracic ultrasound n (%)Chest radiography n (%)Chest CT n (%)Pneumothorax41 (53.9)31 (40.8)45 (59.2)Hemothorax51 (67.1)33 (43.4)53 (69.7)Lung contusion49 (64.5)14 (18.4)57 (75.0)No abnormality10 (13.2)23 (30.3)5 (6.6)CT = computed tomography Table 2Diagnostic findings in transthoracic ultrasound, chest X-ray and chest CT in the studied patients (n = 76)FindingTransthoracic ultrasound n (%)Chest radiography n (%)Chest CT n (%)Pneumothorax41 (53.9)31 (40.8)45 (59.2)Hemothorax51 (67.1)33 (43.4)53 (69.7)Lung contusion49 (64.5)14 (18.4)57 (75.0)No abnormality10 (13.2)23 (30.3)5 (6.6)CT = computed tomography Diagnostic findings in transthoracic ultrasound, chest X-ray and chest CT in the studied patients (n = 76) CT = computed tomography Diagnostic accuracy by injury type The 2 × 2 contingency tables comparing transthoracic ultrasound and chest radiography against chest CT for each thoracic injury are presented in Tables 3 and 4. The diagnostic accuracy of transthoracic ultrasound and chest radiography for each thoracic injury using chest CT as the reference standard is summarized in Table 5. Table 32 × 2 contingency tables for Transthoracic ultrasound versus the chest CT as a reference standardTransthoracic ultrasound diagnosis of pneumothoraxChest CT diagnosis of pneumothorax in the studied patients (n = 76)Positive (n = 45)No (%)Negative (n = 31)No (%)Positive (n = 41)41 (91.1)0 (0.0)Negative (n = 35)4 (8.9)31 (100.0) Transthoracic ultrasound diagnosis of hemothorax Chest CT diagnosis of hemothorax in the studied patients ( n = 76) Positive (n = 53)No (%)Negative (n = 23)No (%)Positive (n = 51)51 (96.2)0 (0.0)Negative (n = 25)2 (3.8)23 (100.0) Transthoracic ultrasound diagnosis of lung contusion Chest CT diagnosis of lung contusion in the studied patients ( n = 76) Positive (n = 57)No (%)Negative (n = 19)No (%)Positive (n = 49)49 (86.0)0 (0.0)Negative (n = 27)8 (14.0)19 (100.0)CT = computed tomography Table 32 × 2 contingency tables for Transthoracic ultrasound versus the chest CT as a reference standardTransthoracic ultrasound diagnosis of pneumothoraxChest CT diagnosis of pneumothorax in the studied patients (n = 76)Positive (n = 45)No (%)Negative (n = 31)No (%)Positive (n = 41)41 (91.1)0 (0.0)Negative (n = 35)4 (8.9)31 (100.0) Transthoracic ultrasound diagnosis of hemothorax Chest CT diagnosis of hemothorax in the studied patients ( n = 76) Positive (n = 53)No (%)Negative (n = 23)No (%)Positive (n = 51)51 (96.2)0 (0.0)Negative (n = 25)2 (3.8)23 (100.0) Transthoracic ultrasound diagnosis of lung contusion Chest CT diagnosis of lung contusion in the studied patients ( n = 76) Positive (n = 57)No (%)Negative (n = 19)No (%)Positive (n = 49)49 (86.0)0 (0.0)Negative (n = 27)8 (14.0)19 (100.0)CT = computed tomography 2 × 2 contingency tables for Transthoracic ultrasound versus the chest CT as a reference standard CT = computed tomography Table 42 × 2 contingency tables for chest radiography versus the chest CT as a reference standardChest radiography diagnosis of pneumothoraxChest CT diagnosis of pneumothorax in the studied patients (n = 76)Positive (n = 45)No (%)Negative (n = 31)No (%)Positive (n = 31)31 (68.9)0 (0.0)Negative (n = 45)14 (31.1)31 (100.0) Chest radiography diagnosis of hemothorax Chest CT diagnosis of hemothorax in the studied patients ( n = 76) Positive (n = 53)No (%)Negative (n = 23)No (%)Positive (n = 33)33 (62.3)0 (0.0)Negative (n = 43)20 (37.7)23 (100.0) Chest radiography diagnosis of lung contusion Chest CT diagnosis of lung contusion in the studied patients ( n = 76) Positive (n = 57)No (%)Negative (n = 19)No (%)Positive (n = 14)14 (24.6)0 (0.0)Negative (n = 62)43 (75.4)19 (100.0)CT = computed tomography Table 42 × 2 contingency tables for chest radiography versus the chest CT as a reference standardChest radiography diagnosis of pneumothoraxChest CT diagnosis of pneumothorax in the studied patients (n = 76)Positive (n = 45)No (%)Negative (n = 31)No (%)Positive (n = 31)31 (68.9)0 (0.0)Negative (n = 45)14 (31.1)31 (100.0) Chest radiography diagnosis of hemothorax Chest CT diagnosis of hemothorax in the studied patients ( n = 76) Positive (n = 53)No (%)Negative (n = 23)No (%)Positive (n = 33)33 (62.3)0 (0.0)Negative (n = 43)20 (37.7)23 (100.0) Chest radiography diagnosis of lung contusion Chest CT diagnosis of lung contusion in the studied patients ( n = 76) Positive (n = 57)No (%)Negative (n = 19)No (%)Positive (n = 14)14 (24.6)0 (0.0)Negative (n = 62)43 (75.4)19 (100.0)CT = computed tomography 2 × 2 contingency tables for chest radiography versus the chest CT as a reference standard CT = computed tomography Table 5Diagnostic accuracy of transthoracic ultrasound and chest radiography using chest CT as reference standardInjuryModalitySensitivity (%)[95% CI]Specificity (%)[95% CI]PPV(%)[95% CI]NPV(%)[95% CI]Accuracy (%)[95% CI]PneumothoraxTransthoracic ultrasound91[0.78–0.97]100[0.86-1.00]100[0.89-1.00]89[0.72–0.96]95[0.86–0.98]Chest radiography69[0.53–0.81]100[0.86-1.00]100[0.86-1.00]69[0.53–0.81]82[0.71–0.89]HemothoraxTransthoracic ultrasound96[0.86–0.99]100[0.82-1.00]100[0.91-1.00]92[0.72–0.99]97[0.90-1.00]Chest radiography62[0.48–0.75]100[0.82-1.00]100[0.87-1.00]53[0.38–0.69]74[0.62–0.83]Lung contusionTransthoracic ultrasound86[0.74–0.93]100[0.79-1.00]100[0.91-1.00]70[0.50–0.86]89[0.80–0.95]Chest radiography25[0.15–0.38]100[0.79-1.00]100[0.73–0.99]31[0.20–0.44]43[0.32–0.55]CI= Confidence Interval, CT = computed tomography, PPV = positive predictive value, NPV = negative predictive value, NA = Not applicableChest CT was used as the reference standard Table 5Diagnostic accuracy of transthoracic ultrasound and chest radiography using chest CT as reference standardInjuryModalitySensitivity (%)[95% CI]Specificity (%)[95% CI]PPV(%)[95% CI]NPV(%)[95% CI]Accuracy (%)[95% CI]PneumothoraxTransthoracic ultrasound91[0.78–0.97]100[0.86-1.00]100[0.89-1.00]89[0.72–0.96]95[0.86–0.98]Chest radiography69[0.53–0.81]100[0.86-1.00]100[0.86-1.00]69[0.53–0.81]82[0.71–0.89]HemothoraxTransthoracic ultrasound96[0.86–0.99]100[0.82-1.00]100[0.91-1.00]92[0.72–0.99]97[0.90-1.00]Chest radiography62[0.48–0.75]100[0.82-1.00]100[0.87-1.00]53[0.38–0.69]74[0.62–0.83]Lung contusionTransthoracic ultrasound86[0.74–0.93]100[0.79-1.00]100[0.91-1.00]70[0.50–0.86]89[0.80–0.95]Chest radiography25[0.15–0.38]100[0.79-1.00]100[0.73–0.99]31[0.20–0.44]43[0.32–0.55]CI= Confidence Interval, CT = computed tomography, PPV = positive predictive value, NPV = negative predictive value, NA = Not applicableChest CT was used as the reference standard Diagnostic accuracy of transthoracic ultrasound and chest radiography using chest CT as reference standard 91 [0.78–0.97] 100 [0.86-1.00] 100 [0.89-1.00] 89 [0.72–0.96] 95 [0.86–0.98] 69 [0.53–0.81] 100 [0.86-1.00] 100 [0.86-1.00] 69 [0.53–0.81] 82 [0.71–0.89] 96 [0.86–0.99] 100 [0.82-1.00] 100 [0.91-1.00] 92 [0.72–0.99] 97 [0.90-1.00] 62 [0.48–0.75] 100 [0.82-1.00] 100 [0.87-1.00] 53 [0.38–0.69] 74 [0.62–0.83] 86 [0.74–0.93] 100 [0.79-1.00] 100 [0.91-1.00] 70 [0.50–0.86] 89 [0.80–0.95] 25 [0.15–0.38] 100 [0.79-1.00] 100 [0.73–0.99] 31 [0.20–0.44] 43 [0.32–0.55] CI= Confidence Interval, CT = computed tomography, PPV = positive predictive value, NPV = negative predictive value, NA = Not applicable Chest CT was used as the reference standard For pneumothorax detection, transthoracic ultrasound demonstrated 91% sensitivity, 100% specificity, and 95% diagnostic accuracy. Chest radiography showed lower sensitivity (69%) but maintained 100% specificity, resulting in an overall diagnostic accuracy of 82%. For hemothorax detection, transthoracic ultrasound demonstrated 96% sensitivity, 100% specificity, and 97% diagnostic accuracy. In comparison, chest radiography showed 62% sensitivity, 100% specificity, and 74% diagnostic accuracy. The greatest diagnostic difference between imaging modalities was observed for lung contusion detection. Transthoracic ultrasound demonstrated 86% sensitivity, 100% specificity, and 89% diagnostic accuracy. Chest radiography showed markedly lower sensitivity (25%), although specificity remained 100%, resulting in an overall diagnostic accuracy of 43%. Agreement analysis Agreement between imaging modalities and chest CT findings is summarized in Table 6. Transthoracic ultrasound demonstrated a high level of agreement with chest CT findings across all evaluated thoracic injuries. The strongest agreement was observed for hemothorax detection (κ = 0.94), indicating almost perfect agreement, followed by pneumothorax (κ = 0.89). For lung contusion, ultrasound maintained substantial agreement with chest CT (κ = 0.75). Table 6Agreement between imaging modalities and chest CT (Cohen’s kappa analysis)InjuryModalityKappa (κ)[95% CI]p-valueLevel of AgreementPneumothoraxTransthoracic ultrasound vs. chest CT0.89[0.79–0.99]< 0.001Almost perfectChest radiography vs. chest CT0.64[0.49–0.80]< 0.001SubstantialHemothoraxTransthoracic ultrasound vs. chest CT0.94[0.86–0.99]< 0.001Almost perfectChest radiography vs. chest CT0.50[0.34–0.66]< 0.001ModerateLung contusionTransthoracic ultrasound vs. chest CT0.75[0.59–0.91]< 0.001SubstantialChest radiography vs. chest CT0.14[-0.06-0.22]< 0.001SlightCI= Confidence Interval, CT = computed tomography; κ = Cohen’s kappa coefficient. Agreement levels were interpreted according to the Landis and Koch classification Table 6Agreement between imaging modalities and chest CT (Cohen’s kappa analysis)InjuryModalityKappa (κ)[95% CI]p-valueLevel of AgreementPneumothoraxTransthoracic ultrasound vs. chest CT0.89[0.79–0.99]< 0.001Almost perfectChest radiography vs. chest CT0.64[0.49–0.80]< 0.001SubstantialHemothoraxTransthoracic ultrasound vs. chest CT0.94[0.86–0.99]< 0.001Almost perfectChest radiography vs. chest CT0.50[0.34–0.66]< 0.001ModerateLung contusionTransthoracic ultrasound vs. chest CT0.75[0.59–0.91]< 0.001SubstantialChest radiography vs. chest CT0.14[-0.06-0.22]< 0.001SlightCI= Confidence Interval, CT = computed tomography; κ = Cohen’s kappa coefficient. Agreement levels were interpreted according to the Landis and Koch classification Agreement between imaging modalities and chest CT (Cohen’s kappa analysis) 0.89 [0.79–0.99] 0.64 [0.49–0.80] 0.94 [0.86–0.99] 0.50 [0.34–0.66] 0.75 [0.59–0.91] 0.14 [-0.06-0.22] CI= Confidence Interval, CT = computed tomography; κ = Cohen’s kappa coefficient. Agreement levels were interpreted according to the Landis and Koch classification In contrast, chest radiography showed lower agreement with chest CT across the studied injuries. Agreement between chest radiography and chest CT was substantial for pneumothorax (κ = 0.64), moderate for hemothorax (κ = 0.50), and only slight for lung contusion (κ = 0.14). All kappa statistics were statistically significant (p < 0.001). These findings indicate that transthoracic ultrasound demonstrates stronger agreement with chest CT compared with chest radiography. Paired comparison using McNemar’s test confirmed that the difference in sensitivity between transthoracic ultrasound and chest radiography was statistically significant for all three injuries (pneumothorax, hemothorax, and lung contusion; all p < 0.001). Because both modalities achieved 100% specificity across all injury types, the accuracy comparison was mathematically equivalent to the sensitivity comparison and is not reported separately; specificity itself could not be compared statistically, as no discordant pairs were available (Table 7). Table 7Paired comparison of diagnostic sensitivity between transthoracic ultrasound and chest radiography using McNemar’s testInjuryUltrasound sensitivity (%)Chest radiography sensitivity (%)McNemar χ²p-valuePneumothorax916915.13< 0.001Hemothorax966234.80< 0.001Lung contusion862575.33< 0.001McNemar’s test compares paired sensitivity between transthoracic ultrasound and chest radiography for each injury, based on discordant classifications among CT-positive patients. Because specificity was 100% for both modalities (Table 5), the accuracy comparison is mathematically equivalent to the sensitivity comparison and is not shown separately Table 7Paired comparison of diagnostic sensitivity between transthoracic ultrasound and chest radiography using McNemar’s testInjuryUltrasound sensitivity (%)Chest radiography sensitivity (%)McNemar χ²p-valuePneumothorax916915.13< 0.001Hemothorax966234.80< 0.001Lung contusion862575.33< 0.001McNemar’s test compares paired sensitivity between transthoracic ultrasound and chest radiography for each injury, based on discordant classifications among CT-positive patients. Because specificity was 100% for both modalities (Table 5), the accuracy comparison is mathematically equivalent to the sensitivity comparison and is not shown separately Paired comparison of diagnostic sensitivity between transthoracic ultrasound and chest radiography using McNemar’s test McNemar’s test compares paired sensitivity between transthoracic ultrasound and chest radiography for each injury, based on discordant classifications among CT-positive patients. Because specificity was 100% for both modalities (Table 5), the accuracy comparison is mathematically equivalent to the sensitivity comparison and is not shown separately Discussion In this prospective diagnostic accuracy study, transthoracic ultrasound demonstrated significantly superior performance compared with supine chest radiography for the detection of pneumothorax, hemothorax, and lung contusion in trauma patients, using CT as the reference standard. Ultrasound demonstrated higher sensitivity across all injury types (91% for pneumothorax, 96% for hemothorax, and 86% for lung contusion), while maintaining excellent specificity. In contrast, chest radiography demonstrated substantially lower sensitivity, particularly for lung contusion (25%) and hemothorax (62%). Furthermore, agreement with chest CT findings was consistently stronger for ultrasound than for chest radiography, highlighting the diagnostic reliability of point-of-care ultrasound in thoracic trauma assessment. Pneumothorax Our findings are consistent with large meta-analyses demonstrating that transthoracic ultrasound significantly outperforms supine chest radiography in detecting traumatic pneumothorax. A systematic review and meta-analysis by Wilkerson and Stone reported pooled sensitivity of 87% for ultrasound compared with 46% for supine chest radiography, with similarly high specificity for both modalities [15]. Likewise, Alrajhi et al. previously demonstrated pooled sensitivity and specificity of 90% and 98% for transthoracic ultrasound in traumatic pneumothorax, compared with substantially lower sensitivity for chest radiography [16]. A comprehensive meta-analysis by Ebrahimi et al. reported pooled sensitivities of 87–90% for ultrasound versus approximately 46–52% for supine chest radiography, with similarly high specificities for both modalities [17]. More recently, a meta-analysis by Sheng et al. confirmed significantly higher sensitivity of transthoracic ultrasound compared with chest radiography in trauma populations while maintaining comparable specificity [9]. Similarly, Bouillon-Minois et al. demonstrated pooled sensitivity of 79.4% for transthoracic ultrasound versus 48.1% for chest radiography with no statistically significant difference in specificity between the two modalities [10]. Recent study by Tantawy et al. aimed to assess the diagnostic performance of the bedside lung ultrasound in emergency (BLUE) and BLUE plus protocols across multiple respiratory conditions and reported high sensitivity and specificity for pneumothorax 90% and 100% respectively which further validates the robustness of ultrasound in detecting pneumothorax across different clinical settings in the ED [18]. The reduced sensitivity of chest radiography in this study and the previous literature is primarily attributable to the supine position used in trauma imaging, where intrapleural air accumulates anteriorly and may remain radiographically occult. In contrast, ultrasound can directly visualize pleural sliding abnormalities and the lung point sign, enabling earlier detection of occult pneumothorax. However, conflicting findings have been reported in some studies. Santorelli et al. reported that lung ultrasound may underdiagnose clinically significant pneumothorax, with sensitivity as low as 27.5% and a false-negative rate of 72% in their cohort [19]. These findings emphasize the importance of cautious interpretation of negative ultrasound examinations and the continued role of CT imaging when clinical suspicion remains high. Hemothorax For hemothorax detection, our study showed that transthoracic ultrasound was markedly more accurate than supine chest radiography. Using chest CT as the reference standard, transthoracic ultrasound showed very high sensitivity of 96% and perfect specificity, whereas chest radiography demonstrated substantially lower sensitivity (62%) with 100% specificity. In practical terms, transthoracic ultrasound detected almost all hemothoraces, whereas many smaller pleural collections were missed on initial chest radiographs. These findings are consistent with previous studies. Kithinji et al. reported E-FAST sensitivity of 96.1% compared with only 45.1% for chest radiography in detecting hemothorax [20]. Similarly, Aswin et al. found ultrasound sensitivity of 79% compared with 62.9% for chest radiography [21]. Vafaei et al. also reported superior diagnostic performance of ultrasound in the initial evaluation of thoracic trauma, with sensitivity of 75.9% compared with 58.6% for chest radiography [22]. The superior sensitivity of ultrasound can be explained by its ability to detect small volumes of pleural fluid. Transthoracic ultrasound may identify pleural collections as small as 20–100 mL, whereas chest radiography generally requires significantly larger volumes before costophrenic angle blunting becomes visible [23, 24]. This explains the marked discrepancy in sensitivity between the two modalities observed in our study. Lung contusion The most pronounced diagnostic difference in our study was observed in lung contusion detection. In our cohort, ultrasound sensitivity reached 86%, compared with only 25% for chest radiography, with diagnostic accuracies of 89% versus 43%. These findings are consistent with previous evidence demonstrating the superiority of lung ultrasound for detecting lung contusions. A systematic review and meta-analysis by Hosseini et al. reported pooled sensitivity of 92% for ultrasound versus only 44% for chest radiography in detecting lung contusion, while specificity remained high for both modalities (89% for ultrasound and 98% for chest radiography) [25]. Recent prospective studies also support these findings; for example, a comparative study by Kashani et al. for lung contusion assessment found lung ultrasound sensitivity 93.6% with high specificity when benchmarked against CT [26]. These results support the superior ability of ultrasound to detect parenchymal injury, particularly in the early post-trauma period when radiographic findings may be subtle or absent. Radiographic opacities associated with lung contusions may take several hours to become visible, whereas transthoracic ultrasound can detect early pleural line irregularities, B-lines, and small subpleural consolidations shortly after injury [27]. Nevertheless, chest CT remains the gold standard due to its ability to fully characterize the extent of lung injury and identify associated thoracic pathology. Agreement with CT Agreement analysis demonstrated consistently stronger concordance between transthoracic ultrasound and chest CT compared with chest radiography. Ultrasound showed almost perfect agreement with CT for pneumothorax (κ = 0.89) and hemothorax (κ = 0.94), and substantial agreement for lung contusion (κ = 0.75). In contrast, chest radiography demonstrated substantially weaker agreement with CT findings, particularly for lung contusion (κ = 0.14). These findings highlight the reliability of transthoracic ultrasound as a diagnostic tool in thoracic trauma when performed by experienced operators. Although ultrasound remains operator-dependent, its diagnostic performance in the present study was comparable to that reported in several previous trauma ultrasound investigations. Strengths and limitations This study has several strengths. First, the prospective design reduces potential bias associated with retrospective analyses. Second, chest CT was used as the reference standard, ensuring reliable confirmation of thoracic injuries. Third, the study simultaneously evaluated multiple thoracic injuries, allowing a comprehensive comparison of imaging modalities in trauma patients. However, several limitations should be acknowledged. Ultrasound is inherently operator-dependent, and diagnostic performance may vary according to operator experience. In addition, this was a single-center study, and the achieved sample of 76 patients was below the calculated target because of the predefined recruitment period and feasibility constraints, which may limit the generalizability of the findings and the precision of some diagnostic accuracy estimates, particularly for less common injury subgroups. Finally, the study did not evaluate time to diagnosis or the impact of ultrasound findings on clinical outcomes, which may further clarify the clinical utility of ultrasound in trauma care. Larger multicenter studies are warranted to validate these findings across different clinical settings and operator experience levels. Conclusion Transthoracic ultrasound showed high diagnostic accuracy and good agreement with chest CT for detecting traumatic thoracic injuries. Compared with chest radiography, ultrasound demonstrated higher diagnostic performance in this study. These findings suggest that transthoracic ultrasound may be a valuable bedside imaging modality for the early assessment of thoracic trauma. However, given the single-center design and relatively limited sample size, larger multicenter studies with variable ultrasound operator experience are warranted to validate these findings.