Background: Lung cancer remains a leading cause of cancer-related deaths. Early-stage detection is pivotal for curative interventions, yet most cases are diagnosed at advanced stages. Noninvasive point-of-care diagnostic strategies are needed to reduce lung cancer mortality rates. We assess the potential of electronic nose (eNose) devices as a tool for lung cancer detection.Study design and methods: All online available databases were searched. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we reviewed studies using eNose forin vivoexhaled air analysis in lung cancer patients. Included patients had pathologically confirmed lung cancer, both non-small cell lung cancer and small cell lung cancer, in various stages. Primary outcomes were sensitivity, specificity, negative predictive value and area under the curve. Secondary outcomes evaluated eNose performance in distinguishing non-small cell lung cancer subtypes and small cell lung cancer. Additionally, a subgroup analysis was conducted to evaluate the efficacy of various commonly used sensors in this differentiation. The Quality Assessment of Diagnostic Accuracy Studies 2 tool was used for risk-of-bias assessment.Results: We included 35 studies, comprising 4483 patients. Pooled sensitivity and specificity were 0.90 (95% CI 0.87–0.93) and 0.89 (95% CI 0.85–0.93), respectively. The median negative predictive value was 0.93 and the median area under the curve was 0.91. Both the likelihood of bias and concerns regarding applicability were minimal.Interpretation: eNose devices hold promise as noninvasive, point-of-care tools for lung cancer diagnosis. Owing to the high sensitivity and negative predictive values and great ease of use, they could be valuable additions to the diagnostic toolbox. Further research and validation efforts are critical to maximise their impact on early lung cancer detection.

Purpose To evaluate the pooled diagnostic accuracy of externally tested AI models for malignancy classification of lung nodules on chest CT. Materials and Methods A systematic search of PubMed, Embase, Web of Science, CINAHL, and the Cochrane Library was performed in January 2025 to identify studies evaluating AI models for malignancy classification of lung nodules on chest CT using pathology and/or at least 2-year follow-up as reference standards. Risk of bias was assessed using QUADAS-2, and pooled sensitivity and specificity were estimated using bivariate random-effects models. Results Twenty-one studies including 7,454 nodules were analyzed, with lung cancer prevalence ranging from 5.7% (17/297) to 91.5% (214/234). All models were based on deep learning; 17 studies (81%) involved Asian populations, 15 (71%) used non-screening populations, 14 (67%) reported 2D or 3D CNN architectures, and eight (38%) specified predefined malignancy thresholds. High risk of bias was identified in five studies for patient selection and two for index testing. Pooled sensitivity was 88%, specificity 75%, positive likelihood ratio 3.55, negative likelihood ratio 0.16, area under the receiver operating characteristic curve 0.89, and diagnostic odds ratio 22.4. Heterogeneity was high (I ² > 90%). Model architecture was associated with specificity, with higher values in studies reporting 2D or 3D CNNs compared with those without reported architecture (82–83% vs 58%, P = .03; meta-regression P = .02); other subgroup analyses showed no evidence of differences. Conclusion Externally tested AI models demonstrated high sensitivity but moderate specificity for malignancy classification of lung nodules on chest CT, supporting a potential role in rule-out strategies. However, substantial heterogeneity, inconsistent reporting, and risk of bias limit interpretation. ©RSNA, 2026

Purpose: Worldwide, colorectal carcinoma (CRC) has a high incidence and a substantial cancer-related mortality. The recurrence risk is 30–50% and lung metastases are common. Treatment of lung metastases with stereotactic ablative radiotherapy (SABR) or metastasectomy may increase survival. The best modality for thoracic screening in the follow-up, however, remains controversial. In this study, we aimed to unravel the additional value of routine chest X-ray (CXR) for detecting lung metastases during the follow-up of CRC patients treated with curative surgery. Methods: Between 2013 and 2017, 668 CRC patients were treated with curative intent, of whom 633 patients were included in follow-up, which consisted of CXR, serum Carcino-Embryonic Antigen (CEA) and ultrasound of the liver. Patients who developed lung metastases, diagnosed with CXR and characterised by a normal concomitant serum CEA level, were identified. Number, size and treatment of lung metastases were described. Results: Thirty-four (5.4%) patients developed lung metastases. Seventeen (50%) were detected by CXR without pathological CEA levels. Eleven (65%) of these patients were treated with curative intent, whereas 21% of patients with lung metastases and elevated CEA levels were treated with curative intent (p = 0.049). Higher numbers of lung metastases were associated with a lower chance of curative treatment. Conclusions: More than 50% of patients with lung metastases on CXR in the follow-up would not have been detected with CEA-triggered imaging only. In addition, patients with colorectal lung metastases without elevated CEA levels were often suitable for curative treatment and, therefore, CXR seems to have additional value within the follow-up of CRC.

Introduction: Lung cancer remains a leading cause of cancer-related deaths. Early-stage detection is pivotal for curative interventions, yet most cases are diagnosed at advanced stages. Non-invasive point-of-care diagnostic strategies are needed to reduce lung cancer mortality. We assess the potential of electronic nose (eNose) devices as a non-invasive tool for lung cancer detection. Methods: All online available databases were searched. Following PRISMA guidelines, we reviewed studies using eNose for in vivo exhaled air analysis in lung cancer patients. All patients had pathological confirmed lung cancer, both non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC) in various stages. Primary outcomes were sensitivity, specificity, negative predictive value and area under the curve. Secondary outcomes evaluated eNose performance in distinguishing NSCLC subtypes and SCLC. Additionally, subgroup analysis of efficacy of commonly used sensors was performed. The QUADAS-2 tool was used for risk-of-bias assessment. Results: Thirty studies, comprising 4001 patients, were included. Pooled sensitivity and specificity were 0.90 (CI 0.86-0.93) and 0.89 (CI 0.83-0.92), respectively. Median negative predictive value was 0.93 and median area under the curve was 0.87. In general, both the likelihood of bias and concerns regarding applicability were minimal. Conclusion: eNose devices hold promise as non-invasive, point-of-care tools for lung cancer diagnosis. Due to high sensitivity and negative predictive value, and great ease of use they might be valuable additions to the diagnostic toolbox. Further research and validation efforts are critical to maximise their impact on early lung cancer detection.

Background Lung cancer remains a leading cause of cancer-related mortality. Early-stage non-small cell lung cancer (NSCLC) can be treated curatively. However, recurrence is common and detecting residual disease post-treatment is of utmost importance. Metabolic profiling of exhaled air through volatile organic compound (VOC) analysis has shown promise as a diagnostic tool for detection of lung cancer. While the origins of the VOC alterations are not fully understood, they are thought to be influenced by both the tumour and the host’s response. It is unclear whether VOC profiles change after tumour resection and whether breath analysis can be used to assess residual disease. This study explores the metabolic profile of exhaled air in NSCLC patients after surgery. Patients and methods A prospective multicentre cohort study was conducted with patients undergoing curative surgery. Preoperative and postoperative breath samples were analysed using an Aeonose™, which employs an externally validated algorithm to detect NSCLC with a sensitivity and specificity of 95% and 51%, respectively. The primary endpoint was the specificity of postoperative breath patterns. Results Forty-eight patients were included in the analysis. Postoperative breath patterns remained positive according to the lung cancer algorithm. Analysis classified 3 (6%) patients as negative, 2 (4%) of whom had already been classified as negative prior to surgery. Conclusion Postoperative VOC profiles did not differ significantly from preoperative patterns, suggesting that factors beyond tumour presence influence VOC composition. Further studies are needed to better understand the biological processes behind VOCs and their potential role in postoperative lung cancer prognostication.