Surgical Outcomes in Non-Small Cell Lung Cancer with Distant Metastasis: The Prognostic Significance of Delayed Metastasis Diagnosis.

BACKGROUND: The diagnosis of distant metastasis on preoperative examinations for non-small cell lung cancer (NSCLC) can be challenging, leading to surgery for some patients with uncertain metastasis. This study evaluated the prognostic impact of delayed diagnosis of metastasis on patients who underwent upfront surgery. METHODS: The study enrolled patients who underwent lobectomy or pneumonectomy for NSCLC between June 2010 and December 2017 and evaluated the presence of distant metastasis before surgery. Overall survival (OS) for patients with stage IV cancer was compared with that for patients without metastasis, and the prognostic factors were analyzed. RESULTS: Of 3046 patients (mean age, 63 years; 1770 men), 100 (3.3 %) had distant metastasis, diagnosed preoperatively in 1.4 % (42/3046) and postoperatively in 1.9 % (58/3046) of the patients. The two most common metastasis sites diagnosed after surgery were contralateral lung (22/58, 37.9 %) and ipsilateral pleura (16/58, 27.6 %). The OS (median, 42.7 months) for the patients with stage IV cancer diagnosed postoperatively was comparable with that for the patients with stage IIIB cancer (P = 0.865), whereas the OS (median OS, 91.7 months) for the patients with stage IV cancer diagnosed preoperatively was better than for the patients with stage IIIB cancer (P = 0.001). Among the patients with distant metastasis, squamous cell type (hazard ratio [HR], 3.15; P = 0.002) and systemic treatment for metastasis (HR, 2.42; P = 0.002) were independent predictors of worse OS. CONCLUSIONS: Among NSCLC patients undergoing upfront surgery, the OS for the patients with stage IV cancer diagnosed postoperatively was comparable with that for the patients with stage IIIB cancer. For patients with stage IV disease, squamous cell type and systemic treatment for metastasis were prognostic factors for poorer OS.

Approximating Intermediate Feature Maps of Self-Supervised Convolution Neural Network to Learn Hard Positive Representations in Chest Radiography.

Recent advances in contrastive learning have significantly improved the performance of deep learning models. In contrastive learning of medical images, dealing with positive representation is sometimes difficult because some strong augmentation techniques can disrupt contrastive learning owing to the subtle differences between other standardized CXRs compared to augmented positive pairs; therefore, additional efforts are required. In this study, we propose intermediate feature approximation (IFA) loss, which improves the performance of contrastive convolutional neural networks by focusing more on positive representations of CXRs without additional augmentations. The IFA loss encourages the feature maps of a query image and its positive pair to resemble each other by maximizing the cosine similarity between the intermediate feature outputs of the original data and the positive pairs. Therefore, we used the InfoNCE loss, which is commonly used loss to address negative representations, and the IFA loss, which addresses positive representations, together to improve the contrastive network. We evaluated the performance of the network using various downstream tasks, including classification, object detection, and a generative adversarial network (GAN) inversion task. The downstream task results demonstrated that IFA loss can improve the performance of effectively overcoming data imbalance and data scarcity; furthermore, it can serve as a perceptual loss encoder for GAN inversion. In addition, we have made our model publicly available to facilitate access and encourage further research and collaboration in the field.

CT-guided percutaneous transthoracic needle biopsy for anterior mediastinal lymphoma: the role of PET/CT.

BACKGROUND: Computed tomography (CT)-guided percutaneous transthoracic needle biopsy (PTNB) is not recommended as the diagnostic modality of choice for anterior mediastinal lymphoma, despite its advantages of minimal invasiveness and easy accessibility. PURPOSE: To identify the modifiable risk factors for non-diagnostic results from CT-guided PTNB for anterior mediastinal lymphoma. MATERIAL AND METHODS: This retrospective study identified CT-guided PTNB for anterior mediastinal lesions diagnosed as lymphoma between May 2007 and December 2021. The diagnostic sensitivity and complications were investigated. The appropriateness of PTNB targeting was evaluated using positron emission tomography (PET)/CT and images from intra-procedural CT-guided PTNB. Targeting was considered inappropriate when the supposed trajectory of the cutting needle was within a region of abnormally low metabolism. The risk factors for non-diagnostic results were determined using logistic regression analysis. RESULTS: A total of 67 PTNBs in 60 patients were included. The diagnostic sensitivity for lymphoma was 76.1% (51/67), with an immediate complication rate of 4.5% (3/67). According to the PET/CT images, PTNB targeting was inappropriate in 10/14 (71.4%) of the non-diagnostic PTNBs but appropriate in all diagnostic PTNBs (P <0.001). Inappropriate targeting was the only significant risk factor for non-diagnostic results (odds ratio = 203.69; 95% confidence interval = 8.17-999.99; P = 0.001). The number of specimen acquisitions was not associated with non-diagnostic results (P = 0.40). CONCLUSIONS: Only inappropriate targeting of the non-viable portion according to PET/CT was an independent risk factor for non-diagnostic results. Acquiring PET/CT scans before biopsy and targeting the viable portion on PET/CT may help improve the diagnostic sensitivity of PTNB.

Evaluation of retrieval accuracy and visual similarity in content-based image retrieval of chest CT for obstructive lung disease.

The aim of our study was to assess the performance of content-based image retrieval (CBIR) for similar chest computed tomography (CT) in obstructive lung disease. This retrospective study included patients with obstructive lung disease who underwent volumetric chest CT scans. The CBIR database included 600 chest CT scans from 541 patients. To assess the system performance, follow-up chest CT scans of 50 patients were evaluated as query cases, which showed the stability of the CT findings between baseline and follow-up chest CT, as confirmed by thoracic radiologists. The CBIR system retrieved the top five similar CT scans for each query case from the database by quantifying and comparing emphysema extent and size, airway wall thickness, and peripheral pulmonary vasculatures in descending order from the database. The rates of retrieval of the same pairs of query CT scans in the top 1-5 retrievals were assessed. Two expert chest radiologists evaluated the visual similarities between the query and retrieved CT scans using a five-point scale grading system. The rates of retrieving the same pairs of query CTs were 60.0% (30/50) and 68.0% (34/50) for top-three and top-five retrievals. Radiologists rated 64.8% (95% confidence interval 58.8-70.4) of the retrieved CT scans with a visual similarity score of four or five and at least one case scored five points in 74% (74/100) of all query cases. The proposed CBIR system for obstructive lung disease integrating quantitative CT measures demonstrated potential for retrieving chest CT scans with similar imaging phenotypes. Further refinement and validation in this field would be valuable.

Deep Learning-Based CT Reconstruction Kernel Conversion in the Quantification of Interstitial Lung Disease: Effect on Reproducibility.

RATIONALE AND OBJECTIVES: The effect of different computed tomography (CT) reconstruction kernels on the quantification of interstitial lung disease (ILD) has not been clearly demonstrated. The study aimed to investigate the effect of reconstruction kernels on the quantification of ILD on CT and determine whether deep learning-based kernel conversion can reduce the variability of automated quantification results between different CT kernels. MATERIALS AND METHODS: Patients with ILD or interstitial lung abnormality who underwent noncontrast high-resolution CT between June 2022 and September 2022 were retrospectively included. Images were reconstructed with three different kernels: B30f, B50f, and B60f. B60f was regarded as the reference standard for quantification, and B30f and B50f images were converted to B60f images using a deep learning-based algorithm. Each disease pattern of ILD and the fibrotic score were quantified using commercial software. The effect of kernel conversion on measurement variability was estimated using intraclass correlation coefficient (ICC) and Bland-Altman method. RESULTS: A total of 194 patients were included in the study. Application of different kernels induced differences in the quantified extent of each pattern. Reticular opacity and honeycombing were underestimated on B30f images and overestimated on B50f images. After kernel conversion, measurement variability was reduced (mean difference, from -2.0 to 3.9 to -0.3 to 0.4%, and 95% limits of agreement [LOA], from [-5.0, 12.7] to [-2.7, 2.1]). The fibrotic score for converted B60f from B50f images was almost equivalent to the original B60f (ICC, 1.000; mean difference, 0.0; and 95% LOA [-0.4, 0.4]). CONCLUSION: Quantitative CT analysis of ILD was affected by the application of different kernels, but deep learning-based kernel conversion effectively reduced measurement variability, improving the reproducibility of quantification.

Deep Learning for Automated Triaging of Stable Chest Radiographs in a Follow-up Setting.

Background Most artificial intelligence algorithms that interpret chest radiographs are restricted to an image from a single time point. However, in clinical practice, multiple radiographs are used for longitudinal follow-up, especially in intensive care units (ICUs). Purpose To develop and validate a deep learning algorithm using thoracic cage registration and subtraction to triage pairs of chest radiographs showing no change by using longitudinal follow-up data. Materials and Methods A deep learning algorithm was retrospectively developed using baseline and follow-up chest radiographs in adults from January 2011 to December 2018 at a tertiary referral hospital. Two thoracic radiologists reviewed randomly selected pairs of "change" and "no change" images to establish the ground truth, including normal or abnormal status. Algorithm performance was evaluated using area under the receiver operating characteristic curve (AUC) analysis in a validation set and temporally separated internal test sets (January 2019 to August 2021) from the emergency department (ED) and ICU. Threshold calibration for the test sets was conducted, and performance with 40% and 60% triage thresholds was assessed. Results This study included 3 304 996 chest radiographs in 329 036 patients (mean age, 59 years ± 14 [SD]; 170 433 male patients). The training set included 550 779 pairs of radiographs. The validation set included 1620 pairs (810 no change, 810 change). The test sets included 533 pairs (ED; 265 no change, 268 change) and 600 pairs (ICU; 310 no change, 290 change). The algorithm had AUCs of 0.77 (validation), 0.80 (ED), and 0.80 (ICU). With a 40% triage threshold, specificity was 88.4% (237 of 268 pairs) and 90.0% (261 of 290 pairs) in the ED and ICU, respectively. With a 60% triage threshold, specificity was 79.9% (214 of 268 pairs) and 79.3% (230 of 290 pairs) in the ED and ICU, respectively. For urgent findings (consolidation, pleural effusion, pneumothorax), specificity was 78.6%-100% (ED) and 85.5%-93.9% (ICU) with a 40% triage threshold. Conclusion The deep learning algorithm could triage pairs of chest radiographs showing no change while detecting urgent interval changes during longitudinal follow-up. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Czum in this issue.

Overcoming the Challenges in the Development and Implementation of Artificial Intelligence in Radiology: A Comprehensive Review of Solutions Beyond Supervised Learning.

Artificial intelligence (AI) in radiology is a rapidly developing field with several prospective clinical studies demonstrating its benefits in clinical practice. In 2022, the Korean Society of Radiology held a forum to discuss the challenges and drawbacks in AI development and implementation. Various barriers hinder the successful application and widespread adoption of AI in radiology, such as limited annotated data, data privacy and security, data heterogeneity, imbalanced data, model interpretability, overfitting, and integration with clinical workflows. In this review, some of the various possible solutions to these challenges are presented and discussed; these include training with longitudinal and multimodal datasets, dense training with multitask learning and multimodal learning, self-supervised contrastive learning, various image modifications and syntheses using generative models, explainable AI, causal learning, federated learning with large data models, and digital twins.

Long-Term Follow-Up of Interstitial Lung Abnormality: Implication in Follow-Up Strategy and Risk Thresholds.

Rationale: The optimal follow-up computed tomography (CT) interval for detecting the progression of interstitial lung abnormality (ILA) is unknown. Objectives: To identify optimal follow-up strategies and extent thresholds on CT relevant to outcomes. Methods: This retrospective study included self-referred screening participants aged 50 years or older, including nonsmokers, who had imaging findings relevant to ILA on chest CT scans. Consecutive CT scans were evaluated to determine the dates of the initial CT showing ILA and the CT showing progression. Deep learning-based ILA quantification was performed. Cox regression was used to identify risk factors for the time to ILA progression and progression to usual interstitial pneumonia (UIP). Measurements and Main Results: Of the 305 participants with a median follow-up duration of 11.3 years (interquartile range, 8.4-14.3 yr), 239 (78.4%) had ILA on at least one CT scan. In participants with serial follow-up CT studies, ILA progression was observed in 80.5% (161 of 200), and progression to UIP was observed in 17.3% (31 of 179), with median times to progression of 3.2 years (95% confidence interval [CI], 3.0-3.4 yr) and 11.8 years (95% CI, 10.8-13.0 yr), respectively. The extent of fibrosis on CT was an independent risk factor for ILA progression (hazard ratio, 1.12 [95% CI, 1.02-1.23]) and progression to UIP (hazard ratio, 1.39 [95% CI, 1.07-1.80]). Risk groups based on honeycombing and extent of fibrosis (1% in the whole lung or 5% per lung zone) showed significant differences in 10-year overall survival (P = 0.02). Conclusions: For individuals with initially detected ILA, follow-up CT at 3-year intervals may be appropriate to monitor radiologic progression; however, those at high risk of adverse outcomes on the basis of the quantified extent of fibrotic ILA and the presence of honeycombing may benefit from shortening the interval for follow-up scans.

Sublobar resection in non-small cell lung cancer: patient selection criteria and risk factors for recurrence.

OBJECTIVE: To validate selection criteria for sublobar resection in patients with lung cancer with respect to recurrence, and to investigate predictors for recurrence in patients for whom the criteria are not suitable. METHODS: Patients who underwent sublobar resection for lung cancer between July 2010 and December 2018 were retrospectively included. The criteria for curative sublobar resection were consolidation-to-tumor ratio ≤0.50 and size ≤3.0 cm in tumors with a ground-glass opacity (GGO) component (GGO group), and size of ≤2.0 cm and volume doubling time ≥400 days in solid tumors (solid group). Cox regression was used to identify predictors for time-to-recurrence (TTR) in tumors outside of these criteria (non-curative group). RESULTS: Out of 530 patients, 353 were classified into the GGO group and 177 into the solid group. In the GGO group, the 2-year recurrence rates in curative and non-curative groups were 2.1 and 7.7%, respectively (p = 0.054). In the solid group, the 2-year recurrence rates in curative and non-curative groups were 0.0 and 28.6%, respectively (p = 0.03). Predictors of 2-year TTR after non-curative sublobar resection were pathological nodal metastasis (hazard ratio [HR], 6.63; p = 0.02) and lymphovascular invasion (LVI; HR, 3.28; p = 0.03) in the GGO group, and LVI (HR, 4.37; p < 0.001) and fibrosis (HR, 3.18; p = 0.006) in the solid group. CONCLUSION: The current patient selection criteria for sublobar resection are satisfactory. LVI was a predictor for recurrence after non-curative resection. ADVANCES IN KNOWLEDGE: This result supports selection criteria of patients for sublobar resection. LVI may help predict recurrence after non-curative sublobar resection.

MuSiC-ViT: A multi-task Siamese convolutional vision transformer for differentiating change from no-change in follow-up chest radiographs.

A major responsibility of radiologists in routine clinical practice is to read follow-up chest radiographs (CXRs) to identify changes in a patient's condition. Diagnosing meaningful changes in follow-up CXRs is challenging because radiologists must differentiate disease changes from natural or benign variations. Here, we suggest using a multi-task Siamese convolutional vision transformer (MuSiC-ViT) with an anatomy-matching module (AMM) to mimic the radiologist's cognitive process for differentiating baseline change from no-change. MuSiC-ViT uses the convolutional neural networks (CNNs) meet vision transformers model that combines CNN and transformer architecture. It has three major components: a Siamese network architecture, an AMM, and multi-task learning. Because the input is a pair of CXRs, a Siamese network was adopted for the encoder. The AMM is an attention module that focuses on related regions in the CXR pairs. To mimic a radiologist's cognitive process, MuSiC-ViT was trained using multi-task learning, normal/abnormal and change/no-change classification, and anatomy-matching. Among 406 K CXRs studied, 88 K change and 115 K no-change pairs were acquired for the training dataset. The internal validation dataset consisted of 1,620 pairs. To demonstrate the robustness of MuSiC-ViT, we verified the results with two other validation datasets. MuSiC-ViT respectively achieved accuracies and area under the receiver operating characteristic curves of 0.728 and 0.797 on the internal validation dataset, 0.614 and 0.784 on the first external validation dataset, and 0.745 and 0.858 on a second temporally separated validation dataset. All code is available at https://github.com/chokyungjin/MuSiC-ViT.

Prognostic Implications of Synchronous Subsolid Nodules in Patients with Resected Subsolid Lung Adenocarcinoma.

Background For multiple subsolid nodules (SSNs) observed at lung CT, current management focuses on removal of the dominant (≥6 mm) nodule and monitoring of remaining SSNs. Whether the presence of these synchronous SSNs is related to postoperative patient outcomes has not been well established. Purpose To evaluate the prognostic value of single versus multiple synchronous SSNs at preoperative CT in patients with resected subsolid lung adenocarcinoma nodules. Materials and Methods This retrospective study included patients who underwent lobectomy or sublobar resection for lung adenocarcinoma manifesting as an SSN and clinical stage IA from January 2010 to December 2017. The radiologic features of the resected SSN (dominant nodule) and synchronous SSNs were assessed on preoperative CT scans. The effects of synchronous SSNs on time to secondary intervention, time to recurrence (TTR), and overall survival (OS) were evaluated using Cox regression analysis. Results Of the 684 included patients (mean age, 60.9 years ± 9.5 [SD]; 389 female), 515 (75.3%) had a single SSN and 169 (24.7%) had multiple SSNs on preoperative CT scans. During follow-up (median, 71.8 months), 38 secondary interventions were performed, primarily due to growth of synchronous SSNs (21 of 38) or metachronous nodules (14 of 38). As the number of synchronous SSNs greater than or equal to 6 mm in size increased, the time to secondary intervention decreased (P < .001). No association was observed between synchronous SSNs and TTR (P = .53) or OS (P = .65), but these measures were associated with features of the resected nodule, specifically solid portion size for TTR (P = .01) and histologic subtype for TTR and OS (P < .001 for both). Conclusion In patients with subsolid lung adenocarcinoma, the presence of synchronous SSNs on preoperative CT scans was not associated with TTR or OS, but the presence of synchronous SSNs greater than or equal to 6 mm in size was associated with an increased likelihood of secondary intervention. © RSNA, 2023 Supplemental material is available for this article.

Generative Adversarial Network-Based Image Conversion Among Different Computed Tomography Protocols and Vendors: Effects on Accuracy and Variability in Quantifying Regional Disease Patterns of Interstitial Lung Disease.

OBJECTIVE: To assess whether computed tomography (CT) conversion across different scan parameters and manufacturers using a routable generative adversarial network (RouteGAN) can improve the accuracy and variability in quantifying interstitial lung disease (ILD) using a deep learning-based automated software. MATERIALS AND METHODS: This study included patients with ILD who underwent thin-section CT. Unmatched CT images obtained using scanners from four manufacturers (vendors A-D), standard- or low-radiation doses, and sharp or medium kernels were classified into groups 1-7 according to acquisition conditions. CT images in groups 2-7 were converted into the target CT style (Group 1: vendor A, standard dose, and sharp kernel) using a RouteGAN. ILD was quantified on original and converted CT images using a deep learning-based software (Aview, Coreline Soft). The accuracy of quantification was analyzed using the dice similarity coefficient (DSC) and pixel-wise overlap accuracy metrics against manual quantification by a radiologist. Five radiologists evaluated quantification accuracy using a 10-point visual scoring system. RESULTS: Three hundred and fifty CT slices from 150 patients (mean age: 67.6 ± 10.7 years; 56 females) were included. The overlap accuracies for quantifying total abnormalities in groups 2-7 improved after CT conversion (original vs. converted: 0.63 vs. 0.68 for DSC, 0.66 vs. 0.70 for pixel-wise recall, and 0.68 vs. 0.73 for pixel-wise precision; P < 0.002 for all). The DSCs of fibrosis score, honeycombing, and reticulation significantly increased after CT conversion (0.32 vs. 0.64, 0.19 vs. 0.47, and 0.23 vs. 0.54, P < 0.002 for all), whereas those of ground-glass opacity, consolidation, and emphysema did not change significantly or decreased slightly. The radiologists' scores were significantly higher (P < 0.001) and less variable on converted CT. CONCLUSION: CT conversion using a RouteGAN can improve the accuracy and variability of CT images obtained using different scan parameters and manufacturers in deep learning-based quantification of ILD.

Automated CT quantification of interstitial lung abnormality and interstitial lung disease according to the Fleischner Society in patients with resectable lung cancer: prognostic significance.

OBJECTIVE: To assess the prognostic significance of automatically quantified interstitial lung abnormality (ILA) according to the definition by the Fleischner Society in patients with resectable non-small-cell lung cancer (NSCLC). METHODS: Patients who underwent lobectomy or pneumonectomy for NSCLC between January 2015 and December 2019 were retrospectively included. Preoperative CT scans were analyzed using the commercially available deep-learning-based automated quantification software for ILA. According to quantified results and the definition by the Fleischner Society and multidisciplinary discussion, patients were divided into normal, ILA, and interstitial lung disease (ILD) groups. RESULTS: Of the 1524 patients, 87 (5.7%) and 20 (1.3%) patients had ILA and ILD, respectively. Both ILA (HR, 1.81; 95% CI: 1.25-2.61; p = .002) and ILD (HR, 5.26; 95% CI: 2.99-9.24; p < .001) groups had poor recurrence-free survival (RFS). Overall survival (OS) decreased (HR 2.13 [95% CI: 1.27-3.58; p = .004] for the ILA group and 7.20 [95% CI: 3.80-13.62, p < .001] for the ILD group) as the disease severity increased. Both quantified fibrotic and non-fibrotic ILA components were associated with poor RFS (HR, 1.57; 95% CI: 1.12-2.21; p = .009; and HR, 1.11; 95% CI: 1.01-1.23; p = .03) and OS (HR, 1.59; 95% CI: 1.06-2.37; p = .02; and HR, 1.17; 95% CI: 1.03-1.33; and p = .01) in normal and ILA groups. CONCLUSIONS: The automated CT quantification of ILA based on the definition by the Fleischner Society predicts outcomes of patients with resectable lung cancer based on the disease category and quantified fibrotic and non-fibrotic ILA components. CLINICAL RELEVANCE STATEMENT: Quantitative CT assessment of ILA provides prognostic information for lung cancer patients after surgery, which can help in considering active surveillance for recurrence, especially in those with a larger extent of quantified ILA. KEY POINTS: • Of the 1524 patients with resectable lung cancer, 1417 (93.0%) patients were categorized as normal, 87 (5.7%) as interstitial lung abnormality (ILA), and 20 (1.3%) as interstitial lung disease (ILD). • Both ILA and ILD groups were associated with poor recurrence-free survival (hazard ratio [HR], 1.81, p = .002; HR, 5.26, p < .001, respectively) and overall survival (HR, 2.13; p = .004; HR, 7.20; p < .001). • Both quantified fibrotic and non-fibrotic ILA components were associated with recurrence-free survival and overall survival in normal and ILA groups.

High urinary desmosine is associated with long-term mortality in patients with COPD.

COPD patients with high baseline urinary desmosines demonstrated significantly higher mortality than those with lower urinary desmosines. High urinary desmosine is independently associated with an increased risk of long-term mortality in COPD patients. https://bit.ly/4015xZ9.

Interstitial Lung Abnormalities at CT in the Korean National Lung Cancer Screening Program: Prevalence and Deep Learning-based Texture Analysis.

Background Interstitial lung abnormalities (ILAs) are associated with worse clinical outcomes, but ILA with lung cancer screening CT has not been quantitatively assessed. Purpose To determine the prevalence of ILA at CT examinations from the Korean National Lung Cancer Screening Program and define an optimal lung area threshold for ILA detection with CT with use of deep learning-based texture analysis. Materials and Methods This retrospective study included participants who underwent chest CT between April 2017 and December 2020 at two medical centers participating in the Korean National Lung Cancer Screening Program. CT findings were classified by three radiologists into three groups: no ILA, equivocal ILA, and ILA (fibrotic and nonfibrotic). Progression was evaluated between baseline and last follow-up CT scan. The extent of ILA was assessed visually and quantitatively with use of deep learning-based texture analysis. The Youden index was used to determine an optimal cutoff value for detecting ILA with use of texture analysis. Demographics and ILA subcategories were compared between participants with progressive and nonprogressive ILA. Results A total of 3118 participants were included in this study, and ILAs were observed with the CT scans of 120 individuals (4%). The median extent of ILA calculated by the quantitative system was 5.8% for the ILA group, 0.7% for the equivocal ILA group, and 0.1% for the no ILA group (P < .001). A 1.8% area threshold in a lung zone for quantitative detection of ILA showed 100% sensitivity and 99% specificity. Progression was observed in 48% of visually assessed fibrotic ILAs (15 of 31), and quantitative extent of ILA increased by 3.1% in subjects with progression. Conclusion ILAs were detected in 4% of the Korean lung cancer screening population. Deep learning-based texture analysis showed high sensitivity and specificity for detecting ILA with use of a 1.8% lung area cutoff value. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Egashira and Nishino in this issue.

Development of a patient-specific chest computed tomography imaging phantom with realistic lung lesions using silicone casting and three-dimensional printing.

The validation of the accuracy of the quantification software in computed tomography (CT) images is very challenging. Therefore, we proposed a CT imaging phantom that accurately represents patient-specific anatomical structures and randomly integrates various lesions including disease-like patterns and lesions of various shapes and sizes using silicone casting and three-dimensional (3D) printing. Six nodules of various shapes and sizes were randomly added to the patient's modeled lungs to evaluate the accuracy of the quantification software. By using silicone materials, CT intensities suitable for the lesions and lung parenchyma were realized, and their Hounsfield unit (HU) values were evaluated on a CT scan of the phantom. As a result, based on the CT scan of the imaging phantom model, the measured HU values for the normal lung parenchyma, each nodule, fibrosis, and emphysematous lesions were within the target value. The measurement error between the stereolithography model and 3D-printing phantoms was 0.2 ± 0.18 mm. In conclusion, the use of 3D printing and silicone casting allowed the application and evaluation of the proposed CT imaging phantom for the validation of the accuracy of the quantification software in CT images, which could be applied to CT-based quantification and development of imaging biomarkers.

Recurrence Patterns and Patient Outcomes in Resected Lung Adenocarcinoma Differ according to Ground-Glass Opacity at CT.

Background Although lung adenocarcinoma with ground-glass opacity (GGO) is known to have distinct characteristics, limited data exist on whether the recurrence pattern and outcomes in patients with resected lung adenocarcinoma differ according to GGO presence at CT. Purpose To examine recurrence patterns and associations with outcomes in patients with resected lung adenocarcinoma according to GGO at CT. Materials and Methods Patients who underwent CT followed by lobectomy or pneumonectomy for lung adenocarcinoma between July 2010 and December 2017 were retrospectively included. Patients were divided into two groups based on the presence of GGO: GGO adenocarcinoma and solid adenocarcinoma. Recurrence patterns at follow-up CT examinations were investigated and compared between the two groups. The effects of patient grouping on time to recurrence, postrecurrence survival (PRS), and overall survival (OS) were evaluated using Cox regression. Results Of 1019 patients (mean age, 62 years ± 9 [SD]; 520 women), 487 had GGO adenocarcinoma and 532 had solid adenocarcinoma. Recurrences occurred more frequently in patients with solid adenocarcinoma (36.1% [192 of 532 patients]) than in those with GGO adenocarcinoma (16.2% [79 of 487 patients]). Distant metastasis was the most common mode of recurrence in the group with solid adenocarcinoma and all clinical stages. In clinical stage I GGO adenocarcinoma, all regional recurrences appeared as ipsilateral lung metastasis (39.2% [20 of 51]) without regional lymph node metastasis. Brain metastasis was more frequent in patients with clinical stage I solid adenocarcinoma (16.5% [16 of 97 patients]). The presence of GGO was associated with time to recurrence and OS (adjusted hazard ratio [HR], 0.6 [P < .001] for both). Recurrence pattern was an independent risk factor for PRS (adjusted HR, 2.1 for distant metastasis [P < .001] and 3.9 for brain metastasis [P < .001], with local-regional recurrence as the reference). Conclusion Recurrence patterns, time to recurrence, and overall survival differed between patients with and without ground-glass opacity at CT, and recurrence patterns were associated with postrecurrence survival. © RSNA, 2023 Supplemental material is available for this article.

Image Turing test and its applications on synthetic chest radiographs by using the progressive growing generative adversarial network.

The generative adversarial network (GAN) is a promising deep learning method for generating images. We evaluated the generation of highly realistic and high-resolution chest radiographs (CXRs) using progressive growing GAN (PGGAN). We trained two PGGAN models using normal and abnormal CXRs, solely relying on normal CXRs to demonstrate the quality of synthetic CXRs that were 1000 × 1000 pixels in size. Image Turing tests were evaluated by six radiologists in a binary fashion using two independent validation sets to judge the authenticity of each CXR, with a mean accuracy of 67.42% and 69.92% for the first and second trials, respectively. Inter-reader agreements were poor for the first (κ = 0.10) and second (κ = 0.14) Turing tests. Additionally, a convolutional neural network (CNN) was used to classify normal or abnormal CXR using only real images and/or synthetic images mixed datasets. The accuracy of the CNN model trained using a mixed dataset of synthetic and real data was 93.3%, compared to 91.0% for the model built using only the real data. PGGAN was able to generate CXRs that were identical to real CXRs, and this showed promise to overcome imbalances between classes in CNN training.