Search

Abstract Background Outside a screening program, early-stage lung cancer is generally diagnosed after the detection of incidental nodules in clinically ordered chest CT scans. Despite the advances in artificial intelligence (AI) systems for lung cancer detection, clinical validation of these systems is lacking in a non-screening setting. Method We developed a deep learning-based AI system and assessed its performance for the detection of actionable benign nodules (requiring follow-up), small lung cancers, and pulmonary metastases in CT scans acquired in two Dutch hospitals (internal and external validation). A panel of five thoracic radiologists labeled all nodules, and two additional radiologists verified the nodule malignancy status and searched for any missed cancers using data from the national Netherlands Cancer Registry. The detection performance was evaluated by measuring the sensitivity at predefined false positive rates on a free receiver operating characteristic curve and was compared with the panel of radiologists. Results On the external test set (100 scans from 100 patients), the sensitivity of the AI system for detecting benign nodules, primary lung cancers, and metastases is respectively 94.3% (82/87, 95% CI: 88.1–98.8%), 96.9% (31/32, 95% CI: 91.7–100%), and 92.0% (104/113, 95% CI: 88.5–95.5%) at a clinically acceptable operating point of 1 false positive per scan (FP/s). These sensitivities are comparable to or higher than the radiologists, albeit with a slightly higher FP/s (average difference of 0.6). Conclusions The AI system reliably detects benign and malignant pulmonary nodules in clinically indicated CT scans and can potentially assist radiologists in this setting.

In this study, we investigate the detection of cardiomegaly on frontal chest radiographs through two alternative deep-learning approaches - via anatomical segmentation and via image-level classification. We used the publicly available ChestX-ray14 dataset, and obtained heart and lung segmentation annotations for 778 chest radiographs for the development of the segmentation-based approach. The classification-based method was trained with 65k standard chest radiographs with image-level labels. For both approaches, the best models were found through hyperparameter searches where architectural, learning, and regularization related parameters were optimized systematically. The resulting models were tested on a set of 367 held-out images for which cardiomegaly annotations were hand-labeled by two independent expert radiologists. Sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve (AUC) were calculated. The performance of the segmentation-based system with an AUC of 0.977 is significantly better for classifying cardiomegaly than the classification-based model which achieved an AUC of 0.941. Only the segmentation-based model achieved comparable performance to an independent expert reader (AUC of 0.978). We conclude that the segmentation-based model requires 100 times fewer annotated chest radiographs to achieve a substantially better performance, while also producing more interpretable results.

Purpose One of the main pathophysiological mechanisms of chronic obstructive pulmonary disease is inflammation, which has been associated with lymphadenopathy. Intrapulmonary lymph nodes can be identified on CT as perifissural nodules (PFN). We investigated the association between the number and size of PFNs and measures of COPD severity. Materials and Methods CT images were obtained from COPDGene. 50 subjects were randomly selected per GOLD stage (0 to 4), GOLD-unclassified, and never-smoker groups and allocated to either “Healthy,” “Mild,” or “Moderate/severe” groups. 26/350 (7.4%) subjects had missing images and were excluded. Supported by computer-aided detection, a trained researcher prelocated non-calcified opacities larger than 3 mm in diameter. Included lung opacities were classified independently by two radiologists as either “PFN,” “not a PFN,” “calcified,” or “not a nodule”; disagreements were arbitrated by a third radiologist. Ordinal logistic regression was performed as the main statistical test. Results A total of 592 opacities were included in the observer study. A total of 163/592 classifications (27.5%) required arbitration. A total of 17/592 opacities (2.9%) were excluded from the analysis because they were not considered nodular, were calcified, or all three radiologists disagreed. A total of 366/575 accepted nodules (63.7%) were considered PFNs. A maximum of 10 PFNs were found in one image; 154/324 (47.5%) contained no PFNs. The number of PFNs per subject did not differ between COPD severity groups (p = 0.50). PFN short-axis diameter could significantly distinguish between the Mild and Moderate/severe groups, but not between the Healthy and Mild groups (p = 0.021). Conclusions There is no relationship between PFN count and COPD severity. There may be a weak trend of larger intrapulmonary lymph nodes among patients with more advanced stages of COPD.

We propose a deep learning system to automatically detect four explainable emphysema signs on frontal and lateral chest radiographs. Frontal and lateral chest radiographs from 3000 studies were retrospectively collected. Two radiologists annotated these with 4 radiological signs of pulmonary emphysema identified from the literature. A patient with ≥2 of these signs present is considered emphysema positive. Using separate deep learning systems for frontal and lateral images we predict the presence of each of the four visual signs and use these to determine emphysema positivity. The ROC and AUC results on a set of 422 held-out cases, labeled by both radiologists, are reported. Comparison with a black-box model which predicts emphysema without the use of explainable visual features is made on the annotations from both radiologists, as well as the subset that they agreed on. DeLong's test is used to compare with the black-box model ROC and McNemar's test to compare with radiologist performance. In 422 test cases, emphysema positivity was predicted with AUCs of 0.924 and 0.946 using the reference standard from each radiologist separately. Setting model sensitivity equivalent to that of the second radiologist, our model has a comparable specificity (p = 0.880 and p = 0.143 for each radiologist respectively). Our method is comparable with the black-box model with AUCs of 0.915 (p = 0.407) and 0.935 (p = 0.291), respectively. On the 370 cases where both radiologists agreed (53 positives), our model achieves an AUC of 0.981, again comparable to the black-box model AUC of 0.972 (p = 0.289). Our proposed method can predict emphysema positivity on chest radiographs as well as a radiologist or a comparable black-box method. It additionally produces labels for four visual signs to ensure the explainability of the result. The dataset is publicly available at https://doi.org/10.5281/zenodo.6373392.

To evaluate whether, and to which extent, experienced radiologists are able to visually correctly differentiate transient from persistent subsolid nodules from a single CT examination alone and to determine CT morphological features to make this differentiation.We selected 86 transient and 135 persistent subsolid nodules from the National Lung Screening Trial (NLST) database. Four experienced radiologists visually assessed a predefined list of morphological features and gave a final judgment on a continuous scale (0-100). To assess observer performance, area under the receiver operating characteristic (ROC) curve was calculated. Statistical differences of morphological features between transient and persistent lesions were calculated using Chi-square. Inter-observer agreement of morphological features was evaluated by percentage agreement.Forty-nine lesions were excluded by at least 2 observers, leaving 172 lesions for analysis. On average observers were able to differentiate transient from persistent subsolid nodules ≥ 10 mm with an area under the curve of 0.75 (95% CI 0.67-0.82). Nodule type, lesion margin, presence of a well-defined border, and pleural retraction showed significant differences between transient and persistent lesions in two observers. Average pair-wise percentage agreement for these features was 81%, 64%, 47% and 89% respectively. Agreement for other morphological features varied from 53% to 95%.The visual capacity of experienced radiologists to differentiate persistent and transient subsolid nodules is moderate in subsolid nodules larger than 10 mm. Performance of the visual assessment of CT morphology alone is not sufficient to generally abandon a short-term follow-up for subsolid nodules.