No abstract

PurposeDespite the importance of radial endobronchial ultrasound (rEBUS) in transbronchial biopsy, researchers have yet to apply artificial intelligence to the analysis of rEBUS images.Materials and methodsThis study developed a convolutional neural network (CNN) to differentiate between malignant and benign tumours in rEBUS images. This study retrospectively collected rEBUS images from medical centres in Taiwan, including 769 from National Taiwan University Hospital Hsin-Chu Branch, Hsinchu Hospital for model training (615 images) and internal validation (154 images) as well as 300 from National Taiwan University Hospital (NTUH-TPE) and 92 images were obtained from National Taiwan University Hospital Hsin-Chu Branch, Biomedical Park Hospital (NTUH-BIO) for external validation. Further assessments of the model were performed using image augmentation in the training phase and test-time augmentation (TTA).ResultsUsing the internal validation dataset, the results were as follows: area under the curve (AUC) (0.88 (95% CI 0.83 to 0.92)), sensitivity (0.80 (95% CI 0.73 to 0.88)), specificity (0.75 (95% CI 0.66 to 0.83)). Using the NTUH-TPE external validation dataset, the results were as follows: AUC (0.76 (95% CI 0.71 to 0.80)), sensitivity (0.58 (95% CI 0.50 to 0.65)), specificity (0.92 (95% CI 0.88 to 0.97)). Using the NTUH-BIO external validation dataset, the results were as follows: AUC (0.72 (95% CI 0.64 to 0.82)), sensitivity (0.71 (95% CI 0.55 to 0.86)), specificity (0.76 (95% CI 0.64 to 0.87)). After fine-tuning, the AUC values for the external validation cohorts were as follows: NTUH-TPE (0.78) and NTUH-BIO (0.82). Our findings also demonstrated the feasibility of the model in differentiating between lung cancer subtypes, as indicated by the following AUC values: adenocarcinoma (0.70; 95% CI 0.64 to 0.76), squamous cell carcinoma (0.64; 95% CI 0.54 to 0.74) and small cell lung cancer (0.52; 95% CI 0.32 to 0.72).ConclusionsOur results demonstrate the feasibility of the proposed CNN-based algorithm in differentiating between malignant and benign lesions in rEBUS images.

BACKGROUND Electronic nose (eNose) can differentiate between healthy individuals and patients with specific respiratory diseases. However, its application in discriminating among various respiratory diseases is limited. OBJECTIVE We aimed to investigate the feasibility of using a novel eNose to differentiate between patients with three respiratory disease entities, including lung cancer, pneumonia, and structural lung disease, and healthy control group. Additionally, we validated the results using gas chromatography mass spectrometry validation (GC-MS) quantitative analysis. METHODS Patients with lung cancer, pneumonia, and structural lung diseases, along with healthy participants were recruited between May 2019 to July 2022. Exhaled breath samples were collected for eNose and GC-MS analysis. Breathprint features from eNose were analyzed using support vector machine model and leave-one-out cross-validation was performed. RESULTS A total of 263 participants (including 95 lung cancer, 59 pneumonia, 71 structural lung disease, and 38 healthy participants) were included. Three-dimensional linear discriminant analysis (LDA) showed a clear distribution of breathprints. The overall accuracy of eNose for four groups was 0.738 (194/263). The accuracy was 0.86 (61/71), 0.81 (77/95), 0.53 (31/59), and 0.66 (25/38) for structural lung disease, lung cancer, pneumonia, and control groups respectively. Pair-wise diagnostic performance comparison revealed excellent discriminant power (AUC:1-0.813) among four groups. The best performance was between structural lung disease and healthy controls (AUC:1), followed by lung cancer and structural lung disease (AUC:0.958). Volatile organic compounds revealed a high individual occurrence rate of cyclohexanone and N,N-dimethylacetamide in pneumonic patients, ethyl acetate in structural lung disease, and 2,3,4-trimethylhexane in lung cancer patients. CONCLUSIONS Our study demonstrated that the novel eNose can differentiate different respiratory diseases. The convenience, short turn-around time, and noninvasiveness of eNose make it a potential point-of-care test in clinical practice.

No abstract

No abstract