Artificial intelligence in lung cancer diagnosis and prognosis: Current application and future perspective.

Lung cancer is one of the malignant tumors with the highest incidence and mortality in the world. The overall five-year survival rate of lung cancer is relatively lower than many leading cancers. Early diagnosis and prognosis of lung cancer are essential to improve the patient's survival rate. With artificial intelligence (AI) approaches widely applied in lung cancer, early diagnosis and prediction have achieved excellent performance in recent years. This review summarizes various types of AI algorithm applications in lung cancer, including natural language processing (NLP), machine learning and deep learning, and reinforcement learning. In addition, we provides evidence regarding the application of AI in lung cancer diagnostic and clinical prognosis. This review aims to elucidate the value of AI in lung cancer diagnosis and prognosis as the novel screening decision-making for the precise treatment of lung cancer patients.

Fusion of laser-induced breakdown spectroscopy technology and deep learning: a new method to identify malignant and benign lung tumors with high accuracy.

Precisely distinguishing between malignant and benign lung tumors is pivotal for suggesting therapeutic strategies and enhancing prognosis, yet this differentiation remains a daunting task. The growth rates, metastatic potentials, and prognoses of benign and malignant tumors differ significantly. Developing specialized treatment protocols tailored to various tumor types is essential for enhancing patient survival outcomes. Employing laser-induced breakdown spectroscopy (LIBS) in conjunction with a deep learning methodology, we attained a high-precision differential diagnosis of malignant and benign lung tumors. First, LIBS spectra of malignant tumors, benign tumors, and normal tissues were collected. The spectra were preprocessed and Z score normalized. Then, the intensities of the Mg II 279.6, Mg I 285.2, Ca II 393.4, Cu II 518.3, and Na I 589.6 nm lines were analyzed in the spectra of the three tissues. The analytical results show that the elemental lines have different contents in the three tissues and can be used as a basis for distinguishing between the three tissues. Finally, the RF-1D ResNet model was constructed by combining the feature importance assessment method of random forest (RF) and one-dimensional residual network (1D ResNet). The classification accuracy, precision, sensitivity, and specificity of the RF-1D ResNet model were 91.1%, 91.6%, 91.3%, and 91.3%, respectively. And the model demonstrates superior performance with an area under the curve (AUC) value of 0.99. The above results show that combining LIBS with deep learning is an effective way to diagnose malignant and benign tumors.

Deep learning ensemble approach with explainable AI for lung and colon cancer classification using advanced hyperparameter tuning.

Lung and colon cancers are leading contributors to cancer-related fatalities globally, distinguished by unique histopathological traits discernible through medical imaging. Effective classification of these cancers is critical for accurate diagnosis and treatment. This study addresses critical challenges in the diagnostic imaging of lung and colon cancers, which are among the leading causes of cancer-related deaths worldwide. Recognizing the limitations of existing diagnostic methods, which often suffer from overfitting and poor generalizability, our research introduces a novel deep learning framework that synergistically combines the Xception and MobileNet architectures. This innovative ensemble model aims to enhance feature extraction, improve model robustness, and reduce overfitting.Our methodology involves training the hybrid model on a comprehensive dataset of histopathological images, followed by validation against a balanced test set. The results demonstrate an impressive classification accuracy of 99.44%, with perfect precision and recall in identifying certain cancerous and non-cancerous tissues, marking a significant improvement over traditional approach.The practical implications of these findings are profound. By integrating Gradient-weighted Class Activation Mapping (Grad-CAM), the model offers enhanced interpretability, allowing clinicians to visualize the diagnostic reasoning process. This transparency is vital for clinical acceptance and enables more personalized, accurate treatment planning. Our study not only pushes the boundaries of medical imaging technology but also sets the stage for future research aimed at expanding these techniques to other types of cancer diagnostics.

Accurate and Convenient Lung Cancer Diagnosis through Detection of Extracellular Vesicle Membrane Proteins via Förster Resonance Energy Transfer.

Tumor-derived extracellular vesicles (EVs) are promising to monitor early stage cancer. Unfortunately, isolating and analyzing EVs from a patient's liquid biopsy are challenging. For this, we devised an EV membrane proteins detection system (EV-MPDS) based on Förster resonance energy transfer (FRET) signals between aptamer quantum dots and AIEgen dye, which eliminated the EV extraction and purification to conveniently diagnose lung cancer. In a cohort of 80 clinical samples, this system showed enhanced accuracy (100% versus 65%) and sensitivity (100% versus 55%) in cancer diagnosis as compared to the ELISA detection method. Improved accuracy of early screening (from 96.4% to 100%) was achieved by comprehensively profiling five biomarkers using a machine learning analysis system. FRET-based tumor EV-MPDS is thus an isolation-free, low-volume (1 µL), and highly accurate approach, providing the potential to aid lung cancer diagnosis and early screening.

Promising Mass-Productive 4-Inch Commercial SERS Sensor with Particle in Micro-Nano Porous Ag/Si/Ag Structure Using in Auxiliary Diagnosis of Early Lung Cancer.

The construction of commercial surface enhanced Raman scattering (SERS) sensors suitable for clinical applications is a pending problem, which is heavily limited by the low production of high-performance SERS bases, because they usually require fine or complicated micro/nano structures. To solve this issue, herein, a promising mass-productive 4-inch ultrasensitive SERS substrate available for early lung cancer diagnosis is proposed, which is designed with a special architecture of particle in micro-nano porous structure. Benefitting from the effective cascaded electric field coupling inside the particle-in-cavity structure and efficient Knudsen diffusion of molecules within the nanohole, the substrate exhibits remarkable SERS performance for gaseous malignancy biomarker, with the limit of detection is 0.1 ppb and the average relative standard deviation value at different scales (from cm2 to µm2 ) is ≈16.5%. In practical application, this large-sized sensor can be further divided into small ones (1 × 1 cm2 ), and more than 65 chips will be obtained from just one 4-inch wafer, greatly increasing the output of commercial SERS sensor. Further, a medical breath bag composed of this small chip is designed and studied in detail here, which suggested high-specificity recognition for lung cancer biomarker in mixed mimetic exhalation tests.

Utility of needle biopsy in centrally located lung cancer for genome analysis: a retrospective cohort study.

BACKGROUND: It is essential to collect a sufficient amount of tumor tissue for successful next-generation sequencing (NGS) analysis. In this study, we investigated the clinical risk factors for avoiding re-biopsy for NGS analysis (re-genome biopsy) in cases where a sufficient amount of tumor tissue could not be collected by bronchoscopy. METHODS: We investigated the association between clinical factors and the risk of re-genome biopsy in patients who underwent transbronchial biopsy (TBB) or endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and required re-genome biopsy in cases enrolled in LC-SCRUM Asia, a prospective nationwide genome screening project in Japan. We also examined whether the frequency of re-genome biopsy decreased between the first and second halves of the enrolment period. RESULTS: Of the 572 eligible patients, 236 underwent TBB, and 134 underwent EBUS-TBNA. Twenty-four TBBs required re-genome biopsy, and multivariate analysis showed that the risk of re-genome biopsy was significantly increased in lesions where the tumor lesion was centrally located. In these cases, EBUS-TBNA should be utilized even if the lesion is a pulmonary lesion. However, it should be noted that even with EBUS-TBNA, lung field lesions are at a higher risk of re-canalization than mediastinal lymph node lesions. It was also found that even when tumor cells were detected in rapid on-site evaluation, a sufficient amount of tumor tissue was not always collected. CONCLUSIONS: For centrally located pulmonary mass lesions, EBUS-TBNA, rather than TBB, can be used to obtain tumor tissues that can be analyzed by NGS.

Lung Cancer Cell-Derived Exosome Detection Using Electrochemical Approach towards Early Cancer Screening.

Lung cancer is one of the deadliest cancers worldwide due to the inability of existing methods for early diagnosis. Tumor-derived exosomes are nano-scale vesicles released from tumor cells to the extracellular environment, and their investigation can be very useful in both biomarkers for early cancer screening and treatment assessment. This research detected the exosomes via an ultrasensitive electrochemical biosensor containing gold nano-islands (Au-NIs) structures. This way, a high surface-area-to-volume ratio of nanostructures was embellished on the FTO electrodes to increase the chance of immobilizing the CD-151 antibody. In this way, a layer of gold was first deposited on the electrode by physical vapor deposition (PVD), followed by thermal annealing to construct primary gold seeds on the surface of the electrode. Then, gold seeds were grown by electrochemical deposition through gold salt. The cell-derived exosomes were successfully immobilized on the FTO electrode through the CD-151 antibody, and cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) methods were used in this research. In the CV method, the change in the current passing through the working electrode is measured so that the connection of exosomes causes the current to decrease. In the EIS method, surface resistance changes were investigated so that the binding of exosomes increased the surface resistance. Various concentrations of exosomes in both cell culture and blood serum samples were measured to test the sensitivity of the biosensor, which makes our biosensor capable of detecting 20 exosomes per milliliter.

Racial difference in BMI and lung cancer diagnosis: analysis of the National Lung Screening Trial.

BACKGROUND: The inverse relationship between BMI and lung cancer diagnosis is well defined. However, few studies have examined the racial differences in these relationships. The purpose of this paper is to explore the relationships amongst race, BMI, and lung cancer diagnosis using the National Lung Screening Trial (NLST) data. METHODS: Multivariate regression analysis was used to analyze the BMI, race, and lung cancer diagnosis relationships. RESULTS: Among 53,452 participants in the NLST cohort, 3.9% were diagnosed with lung cancer, 43% were overweight, and 28% were obese. BMI was inversely related to lung cancer diagnosis among Whites: those overweight (aOR = .83, 95%CI = .75-.93), obese (aOR = .64, 95%CI = .56-.73) were less likely to develop lung cancer, compared to those with normal weight. These relationships were not found among African-Americans. CONCLUSION: Our findings indicate that the inverse relationship of BMI and lung cancer risk among Whites is consistent, whereas this relationship is not significant for African-Americans. In consideration of higher lung cancer incidence among African Americans, we need to explore other unknown mechanisms explaining this racial difference.

Optimal communication associated with lower risk of acute traumatic stress after lung cancer diagnosis.

PURPOSE: The aim of this study was to assess the role of the patient's background and perceived healthcare-related factors in symptoms of acute stress after lung cancer diagnosis. METHODS: The study population consisted of 89 individuals referred for diagnostic work-up at Landspitali National University Hospital in Iceland and subsequently diagnosed with lung cancer. Before diagnosis, the patients completed questionnaires on sociodemographic characteristics, pre-diagnostic distress (Hospital Anxiety and Depression Scale), social support, and resilience. At a median of 16 days after diagnosis, the patients reported symptoms of acute stress on the Impact of Event Scale-Revised (IES-R) and experience of communication and support from healthcare professionals and family during the diagnostic period. RESULTS: Patients were on average 68 years and 52% reported high levels of post-diagnostic acute stress (IES-R > 23) while 24% reported symptoms suggestive of clinical significance (IES-R > 32). Prior history of cancer (ß = 6.7, 95% CI: 0.1 to 13.3) and pre-diagnostic distress were associated with higher levels of post-diagnostic acute stress (ß = 8.8, 95% CI: 2.7 to 14.9), while high educational level (ß = - 7.9, 95% CI: - 14.8 to - 1.1) was associated with lower levels. Controlling for the abovementioned factors, the patients' perception of optimal doctor-patient (ß = - 9.1, 95% CI: - 14.9 to - 3.3) and family communication (ß = - 8.6, 95% CI: - 14.3 to - 2.9) was inversely associated with levels of post-diagnostic acute stress after lung cancer diagnosis. CONCLUSIONS: A high proportion of patients with newly diagnosed lung cancer experience high levels of acute traumatic stress of potential clinical significance. Efforts to improve doctor-patient and family communication may mitigate the risk of these adverse symptoms.

Focused Neck Ultrasound and Lymph Node Sampling by Respiratory Physicians in Suspected Lung Cancer.

BACKGROUND: Malignant cervical lymphadenopathy in the setting of lung cancer represents N3 disease, and neck ultrasound (NUS) with sampling is described in the Royal College of Radiologists ultrasound training curriculum for the non-radiologists. This study reviews the incorporation of NUS +/- biopsy in the routine practice of a lung cancer fast-track clinic in the UK. METHODS: We retrospectively assessed 29 months of activity of a lung cancer fast-track clinic. Systematic focused NUS was conducted in suspected thoracic malignancy, sampling nodes with a ≥5-mm short axis, under real-time US using a linear probe (5-12 Mhz). Fine-needle aspirations (FNAs) with or without 18 Ga core biopsies were taken. RESULTS: Between August 2017 and December 2019, of 152 peripheral lymph nodes (LNs)/deposits sampled, 98 (64.5%) were supraclavicular fossa LNs with median [IQR] size 12 [8-18] mm. Core biopsies were performed in 54/98 (55%) patients, while all patients had FNAs. No complications occurred. The representative yield was 90/95 (94.7%) in cases with suspected cancer. No difference was seen between FNA versus core biopsy (p = 0.44). Of the 5 non-diagnostic samples, one was FNA only. The commonest diagnosis was lung cancer in 66/98 (67.3%). PDL-1 was sufficient in 35/36 tested (97.2%). ALK-FISH was successful in 24/25 (96%) cases. EGFR mutation analysis was successful in 28/31 (90.3%) cases. Median time from clinic to initial diagnosis was 7 [5-10] days. Computed tomography (CT) scans reported no significant lymphadenopathy in 18/96 (18.7%) cases, yet 10/18 (55.5%) cases were positive for malignancy. CONCLUSION: Neck nodal sampling by respiratory physicians was safe, timely, with a high diagnostic yield and suitability for molecular testing. Neck US can provide a timely diagnosis in cases that may be missed by CT alone.

Intraoperative cytopathology of thoracic surgery (ICTS). A captivating, worthwhile, and rewarding service line.

Intraoperative cytopathology for thoracic surgeons is a service that has not been utilized to its full potential in most institutions. It has the advantage of a rapid turnaround time, low costs, high accuracy, real time communication with the surgeon, on-site visualization of the lesion before excision, simplicity, and safety. Our experience, common cytologic findings of the most frequent thoracic tumors encountered during ICTS, hints about the service, and models for implementation and maintenance are presented. This review is aimed to present our experience and perspective about intraoperative cytopathology for thoracic surgeons.

Receptor for Advanced Glycation End Products (RAGE) and Its Polymorphic Variants as Predictive Diagnostic and Prognostic Markers of NSCLCs: a Perspective.

PURPOSE OF REVIEW: Non-small cell lung cancers (NSCLCs) account for ~ 85% of all lung cancers, and 5-year survival in Europe and the USA is ~ 13-17%. In this review, we focus on the significance of Receptor for Advanced Glycation End products (RAGE) as a diagnostic or post-therapeutic prognostic marker for various forms of NSCLCs. RECENT FINDINGS: The lungs have the highest levels of basal RAGE expression in mammals. The physiologic RAGE in lungs may be involved in adhesion and spreading of AT-1 cells and maintenance of pulmonary homeostasis. However, high level expression of RAGE complicates various diseases including acute lung injury. In NSCLCs, while a number of studies report decreased RAGE expression, inferring a protective role, others suggest that RAGE expression may contribute to NSCLC pathogenesis. Genetic polymorphisms of RAGE are reportedly associated with NSCLC development and complications. RAGE and its polymorphic variants may be useful diagnostic or post-therapeutic prognostic markers of NSCLCs.

4D Electromagnetic Navigation Bronchoscopy for the Sampling of Pulmonary Lesions: First European Real-Life Experience.

PURPOSE: The use of Electromagnetic navigation bronchoscopy (ENB) for the diagnosis of pulmonary peripheral lesions is still debated due to its variable diagnostic yield; a new 4D ENB system, acquiring inspiratory and expiratory computed tomography (CT) scans, overcomes respiratory motion and uses tracked sampling instruments, reaching higher diagnostic yields. We aimed at evaluating diagnostic yield and accuracy of a 4D ENB system in sampling pulmonary lesions and at describing their influencing factors. METHODS: We conducted a three-year retrospective observational study including all patients with pulmonary lesions who underwent 4D ENB with diagnostic purposes; all the factors potentially influencing diagnosis were recorded. RESULTS: 103 ENB procedures were included; diagnostic yield and accuracy were, respectively, 55.3% and 66.3%. We reported a navigation success rate of 80.6% and a diagnosis with ENB was achieved in 68.3% of cases; sensitivity for malignancy was 61.8%. The majority of lesions had a bronchus sign on CT, but only the size of lesions influenced ENB diagnosis (p < 0.05). Transbronchial needle aspiration biopsy was the most used tool (93.2% of times) with the higher diagnostic rate (70.2%). We reported only one case of pneumothorax. CONCLUSION: The diagnostic performance of a 4D ENB system is lower than other previous navigation systems used in research settings. Several factors still influence the reachability of the lesion and therefore diagnostic yield. Patient selection, as well as the multimodality approach of the lesion, is strongly recommended to obtain higher diagnostic yield and accuracy, with a low rate of complications.

Black patients referred to a lung cancer screening program experience lower rates of screening and longer time to follow-up.

BACKGROUND: Racial disparities are well-documented in preventive cancer care, but they have not been fully explored in the context of lung cancer screening. We sought to explore racial differences in lung cancer screening outcomes within a lung cancer screening program (LCSP) at our urban academic medical center including differences in baseline low-dose computed tomography (LDCT) results, time to follow-up, adherence, as well as return to annual screening after additional imaging, loss to follow-up, and cancer diagnoses in patients with positive baseline scans. METHODS: A historical cohort study of patients referred to our LCSP was conducted to extract demographic and clinical characteristics, smoking history, and lung cancer screening outcomes. RESULTS: After referral to the LCSP, blacks had significantly lower odds of receiving LDCT compared to whites, even while controlling for individual lung cancer risk factors and neighborhood-level factors. Blacks also demonstrated a trend toward delayed follow-up, decreased adherence, and loss to follow-up across all Lung-RADS categories. CONCLUSIONS: Overall, lung cancer screening annual adherence rates were low, regardless of race, highlighting the need for increased patient education and outreach. Furthermore, the disparities in race we identified encourage further research with the purpose of creating culturally competent and inclusive LCSPs.

An Interpretable Deep Hierarchical Semantic Convolutional Neural Network for Lung Nodule Malignancy Classification.

While deep learning methods have demonstrated performance comparable to human readers in tasks such as computer-aided diagnosis, these models are difficult to interpret, do not incorporate prior domain knowledge, and are often considered as a "black-box." The lack of model interpretability hinders them from being fully understood by end users such as radiologists. In this paper, we present a novel interpretable deep hierarchical semantic convolutional neural network (HSCNN) to predict whether a given pulmonary nodule observed on a computed tomography (CT) scan is malignant. Our network provides two levels of output: 1) low-level semantic features; and 2) a high-level prediction of nodule malignancy. The low-level outputs reflect diagnostic features often reported by radiologists and serve to explain how the model interprets the images in an expert-interpretable manner. The information from these low-level outputs, along with the representations learned by the convolutional layers, are then combined and used to infer the high-level output. This unified architecture is trained by optimizing a global loss function including both low- and high-level tasks, thereby learning all the parameters within a joint framework. Our experimental results using the Lung Image Database Consortium (LIDC) show that the proposed method not only produces interpretable lung cancer predictions but also achieves significantly better results compared to using a 3D CNN alone.

Impact of rapid investigation clinic on timeliness of lung cancer diagnosis and treatment.

BACKGROUND: Guidelines recommend timely evaluation of patients with suspected lung cancer. We evaluated the impact of a Rapid Investigation Clinic (RIC) on timeliness of lung cancer diagnosis and treatment between February 2010 and December 2011. METHODS: Investigation within the RIC was conducted by a pulmonologist and a nurse clinician. Controls were patients with lung cancer, investigated outside the RIC at the same institution during the same time period. The primary outcome was time between first contact with a local physician for suspected lung cancer (T0) and first treatment. Factors associated with the delay from T0 to first treatment were examined using multivariate analysis. Completeness of lung cancer staging according to guidelines was assessed. RESULTS: A total of 195 patients were investigated within the RIC vs. 132 patients outside the RIC. The median delay between T0 and first treatment was 65 days (interquartile range [IQR] 46-92 days) in the RIC and 78 days (IQR 49-119 days) in the non-RIC patients (p ≤ 0.01). Time from T0 to pathological diagnosis was shorter in the RIC (median 26 days; IQR 14-42 days) vs. non-RIC patients (median 40 days; IQR 16-68 days). In multivariate analysis, investigation in the RIC was associated with a reduction in time to first treatment of 24 days (95% confidence interval [CI] 12-35 days) when adjusted for relevant confounders. Guideline-concordant investigation occurred more frequently in RIC patients, based on the quality indicators examined. CONCLUSIONS: A Rapid Investigation Clinic reduces delays to lung cancer diagnosis and treatment, and impacts quality of care.

Napsin A levels in epithelial lining fluid as a diagnostic biomarker of primary lung adenocarcinoma.

BACKGROUND: It is crucial to develop novel diagnostic approaches for determining if peripheral lung nodules are malignant, as such nodules are frequently detected due to the increased use of chest computed tomography scans. To this end, we evaluated levels of napsin A in epithelial lining fluid (ELF), since napsin A has been reported to be an immunohistochemical biomarker for histological diagnosis of primary lung adenocarcinoma. METHODS: In consecutive patients with indeterminate peripheral lung nodules, ELF samples were obtained using a bronchoscopic microsampling (BMS) technique. The levels of napsin A and carcinoembryonic antigen (CEA) in ELF at the nodule site were compared with those at the contralateral site. A final diagnosis of primary lung adenocarcinoma was established by surgical resection. RESULTS: We performed BMS in 43 consecutive patients. Among patients with primary lung adenocarcinoma, the napsin A levels in ELF at the nodule site were markedly higher than those at the contralateral site, while there were no significant differences in CEA levels. Furthermore, in 18 patients who were undiagnosed by bronchoscopy and finally diagnosed by surgery, the napsin A levels in ELF at the nodule site were identically significantly higher than those at the contralateral site. In patients with non-adenocarcinoma, there were no differences in napsin A levels in ELF. The area under the receiver operator characteristic curve for identifying primary lung adenocarcinoma was 0.840 for napsin A and 0.542 for CEA. CONCLUSION: Evaluation of napsin A levels in ELF may be useful for distinguishing primary lung adenocarcinoma.

A novel technique for CT-guided transthoracic biopsy of lung lesions: improved biopsy accuracy and safety.

OBJECTIVES: To present the diagnostic accuracy and safety of a novel technique for CT-guided transthoracic needle aspiration biopsy (TNAB) of lung lesions suspected of malignancy. METHODS: A novel technique for coaxial CT-guided TNAB is reported in this single-centre, retrospective study. A 22-gauge guide wire is used to accurately locate the lesion prior to biopsy. The technique enables penetration of lung lesions in various locations with less risk of harm to adjacent organs. Malignant and benign diagnoses were confirmed by histology or radiologic resolution. RESULTS: Clinical features of 181 patients included 59% men. Mean lesion size was 24 ± 14.9 mm with a mean depth of 13.6 ± 18.3 mm. Among 160 (88.4%) confirmed malignancies, 151 (94.4%) were diagnosed with TNAB. Among the 13 (7.2%) confirmed benign diagnoses, 11 (84.6%) received a specific, benign diagnosis with TNAB. The overall diagnostic accuracy of CT-TNAB was 93.6% among all confirmed diagnoses (173/181). Complications included 48 (26.5%) with pneumothorax, of which 77.8% resolved spontaneously, 20% by aspiration and 2.2% required chest drain insertion. Intrapulmonary haemorrhage was observed in 3.9% and hemoptysis in 6.0% without clinical significance. CONCLUSION: The guide wire technique provides a novel method for needle biopsy of lung lesions with improved accuracy and safety. KEY POINTS: Lung cancer screening has increased the detection of lung lesions. The guide wire technique is a novel method to biopsy lung lesions. The guide wire technique for lung biopsy demonstrates improved accuracy and safety. The chest tube insertion rate is reduced with aspiration during the procedure.

Accuracy of transthoracic ultrasound for the prediction of chest wall infiltration by lung cancer and of lung infiltration by chest wall tumours.

BACKGROUND: We wanted to determine the accuracy of transthoracic ultrasound in the prediction of chest wall infiltration by lung cancer or lung infiltration by chest wall tumours. METHODS: Patients having preoperative CT-scan suspect for lung/chest wall infiltration were prospectively enrolled. Inclusion criteria for lung cancer were: obliteration of extrapleural fat, obtuse angle between tumour and chest wall, associated pleural thickening. The criteria for chest wall tumours were: rib destruction and intercostal muscles infiltration with extrapleural fat obliteration and intrathoracic extension. Lung cancer patients with evident chest wall infiltration were excluded. Transthoracic ultrasound was preoperatively performed. Predictions were checked during surgical intervention. RESULTS: Twenty-three patients were preoperatively examined. Sensitivity, specificity, positive and negative predictive values of transthoracic ultrasound were 88.89%, 100%, 100% and 93.3%, respectively. Youden index was used to determine the best cut-off for tumour size in predicting lung/chest wall infiltration: 4.5cm. At univariate logistic regression, tumour size (<4.5 vs ≥ 4.5cm) (p=0.0072) was significantly associated with infiltration. CONCLUSIONS: Transthoracic ultrasound is a useful instrument for predicting neoplastic lung or chest wall infiltration in cases of suspect CT-scans and could be used as part of the preoperative workup to assess tumour staging and to plan the best surgical approach.