CT-based deep learning model to differentiate invasive pulmonary adenocarcinomas appearing as subsolid nodules among surgical candidates: comparison of the diagnostic performance with a size-based logistic model and radiologists

Prediction of EGFR Mutations in Lung Adenocarcinoma via CT Images: A Comparative Study of Intratumoral and Peritumoral Radiomics, Deep Learning, and Fusion Models

RATIONALE AND OBJECTIVES

This study aims to analyze the intratumoral and peritumoral characteristics of lung adenocarcinoma patients on the basis of chest CT images via radiomic and deep learning methods and to develop and validate a multimodel fusion strategy for predicting epidermal growth factor receptor (EGFR) mutation statuses.

MATERIALS AND METHODS

Retrospective data from 826 lung adenocarcinoma patients across two hospitals were collected. Data from center1 were used for model training and internal validation, while data from center2 were reserved for external validation. Tumor segmentation was performed using the nnUNet network, and volumes of interest (VOIs) for the tumor and its peritumoral regions (2 mm, 4 mm, 6 mm, 8 mm, 10 mm) were subsequently derived.Radiomics features were extracted from various VOIs using PyRadiomics, and radiomics models were developed using Lasso and multiple machine learning algorithms.Using 2D, 2.5D, and 3D images derived from different VOIs as inputs, multiple deep learning models were trained and their performances compared.The radiomics and deep learning models demonstrating the best predictive performance were selected and integrated with clinical models for model fusion.Multi-model fusion of clinical, radiomics, and deep learning features was achieved using feature-level fusion and various decision-level fusion strategies, including hard voting, soft voting, and stacking ensemble.The predictive performances of various fusion models were evaluated and compared systematically.

RESULTS

Among the available radiomic models, the model based on intratumoral and peritumoral 2-mm regions (VOI_Comb2) achieved the best performance on the internal and external validation sets (AUC=0.843 and 0.803, respectively). Compared with 2D and 2.5D deep learning models, the 3D deep learning model demonstrated superior predictive performance. The 3D deep model based on the VOI_Comb2 region achieved the highest AUC among all the deep learning models on the internal and external validation sets (AUC=0.839 and 0.814, respectively). Among the fusion models, the soft voting strategy achieved the highest AUC on the internal and external validation sets, reaching 0.925 and 0.889, respectively. On the external validation set, the AUC of the soft voting model was significantly greater than that of the hard voting model, early fusion model, or any single modality model.

CONCLUSION

This study demonstrates that combining radiomic and deep learning models based on intratumoral and peritumoral regions is an effective method for capturing comprehensive imaging features in lung adenocarcinoma. The multimodal fusion approach using soft voting leverages the strengths of each modality and provides a robust framework for advanced image feature extraction to support personalized treatment.

Enhancing brain age estimation under uncertainty: A spectral-normalized neural gaussian process approach utilizing 2.5D slicing

Brain age gap, the difference between estimated brain age and chronological age via magnetic resonance imaging, has emerged as a pivotal biomarker in the detection of brain abnormalities. While deep learning is accurate in estimating brain age, the absence of uncertainty estimation may pose risks in clinical use. Moreover, current 3D brain age models are intricate, and using 2D slices hinders comprehensive dimensional data integration. Here, we introduced Spectral-normalized Neural Gaussian Process (SNGP) accompanied by 2.5D slice approach for seamless uncertainty integration in a single network with low computational expenses, and extra dimensional data integration without added model complexity. Subsequently, we compared different deep learning methods for estimating brain age uncertainty via the Pearson correlation coefficient, a metric that helps circumvent systematic underestimation of uncertainty during training. SNGP shows excellent uncertainty estimation and generalization on a dataset of 11 public datasets (N = 6327), with competitive predictive performance (MAE=2.95). Besides, SNGP demonstrates superior generalization performance (MAE=3.47) on an independent validation set (N = 301). Additionally, we conducted five controlled experiments to validate our method. Firstly, uncertainty adjustment in brain age estimation improved the detection of accelerated brain aging in adolescents with ADHD, with a 38% increase in effect size after adjustment. Secondly, the SNGP model exhibited OOD detection capabilities, showing significant differences in uncertainty across Asian and non-Asian datasets. Thirdly, the performance of DenseNet as a backbone for SNGP was slightly better than ResNeXt, attributed to DenseNet's feature reuse capability, with robust generalization on an independent validation set. Fourthly, site effect harmonization led to a decline in model performance, consistent with previous studies. Finally, the 2.5D slice approach significantly outperformed 2D methods, improving model performance without increasing network complexity. In conclusion, we present a cost-effective method for estimating brain age with uncertainty, utilizing 2.5D slicing for enhanced performance, showcasing promise for clinical applications.

Optimizing Deep Learning Models for Luminal and Nonluminal Breast Cancer Classification Using Multidimensional ROI in DCE-MRI-A Multicenter Study

OBJECTIVES

Previous deep learning studies have not explored the synergistic effects of ROI dimensions (2D/2.5D/3D), peritumoral expansion levels (0-8 mm), and segmentation scenarios (ROI only vs. ROI original). Our study aims to evaluate the performance of multidimensional deep transfer learning models in distinguishing molecular subtypes of breast cancer (luminal vs. nonluminal) using DCE-MRI. Under two segmentation scenarios, we systematically compare the effects of ROI dimensions and peritumoral expansion levels to optimize multidimensional deep learning models via transfer learning for distinguishing luminal from nonluminal breast cancers in DCE-MRI-based analysis.

MATERIALS AND METHODS

From October 2020 to October 2023, data from 426 patients with primary invasive breast cancer were retrospectively collected. Patients were divided into three cohorts: (1) training cohort, n = 108, from SYSU Hospital (Zhuhai, China); (2) validation cohort 1, n = 165, from HZ Hospital (Huizhou, China); and (3) validation cohort 2, n = 153, from LY Hospital (Linyi, China). ROIs were delineated, and expansions of 2, 4, 6, and 8 mm beyond the lesion boundary were performed. We assessed the performance of various deep transfer learning models, considering precise segmentation (ROI only and ROI original) and varying peritumoral regions, using ROC curves and decision curve analysis.

RESULTS

The 2.5D1-based deep learning model (ROI original, 4 mm expansion) demonstrated optimal performance, achieving an AUC of 0.808 (95% CI 0.715-0.901) in the training cohort, 0.766 (95% CI 0.682-0.850) in validation cohort 1, and 0.799 (95% CI 0.725-0.874) in validation cohort 2.

CONCLUSION

The study highlights that the 2.5D1-based deep learning model utilizing the three principal slices of the minimum bounding box (ROI original) with a 4 mm peritumoral region is effective in distinguishing between luminal and nonluminal breast cancer tumors, serving as a potential diagnostic tool.

Application of Artificial Intelligence in Thoracic Radiology: A Narrative Review

Thoracic radiology has emerged as a primary field in which artificial intelligence (AI) is extensively researched. Recent advancements highlight the potential to enhance radiologists' performance through AI. AI aids in detecting and classifying abnormalities, and in quantifying both normal and abnormal anatomical structures. Additionally, it facilitates prognostication by leveraging these quantitative values. This review article will discuss the recent achievements of AI in thoracic radiology, focusing primarily on deep learning, and explore the current limitations and future directions of this cutting-edge technique.

MRI transformer deep learning and radiomics for predicting IDH wild type TERT promoter mutant gliomas

This study aims to predict IDH wt with TERTp-mut gliomas using multiparametric MRI sequences through a novel fusion model, while matching model classification metrics with patient risk stratification aids in crafting personalized diagnostic and prognosis evaluations.Preoperative T1CE and T2FLAIR sequences from 1185 glioma patients were analyzed. A MultiChannel_2.5D_DL model and a 2D DL model, both based on the cross-scale attention vision transformer (CrossFormer) neural network, along with a Radiomics model, were developed. These were integrated via ensemble learning into a stacking model. The MultiChannel_2.5D_DL model outperformed the 2D_DL and Radiomics models, with AUCs of 0.806-0.870. The stacking model achieved the highest AUC (0.855-0.904) across validation sets. Patients were stratified into high-risk and low-risk groups based on stacking model scores, with significant survival differences observed via Kaplan-Meier analysis and log-rank tests. The stacking model effectively identifies IDH wt TERTp-mutant gliomas and stratifies patient risk, aiding personalized prognosis.

Preoperative CT-based radiomics nomogram to predict the micropapillary pattern in lung adenocarcinoma of size 2 cm or less

Purpose

To develop and validate a radiomics nomogram model for predicting the micropapillary pattern (MPP) in lung adenocarcinoma (LUAD) tumors of ≤2 cm in size.

Methods

In this study, 300 LUAD patients from our institution were randomly divided into the training cohort (n = 210) and an internal validation cohort (n = 90) at a ratio of 7:3, besides, we selected 65 patients from another hospital as the external validation cohort. The region of interest of the tumor was delineated on the computed tomography (CT) images, and radiomics features were extracted. A nomogram model was established using radiomics features, clinical features and conventional radiographic features. The nomogram model was compared with the radiomics model and the clinical model alone to test its diagnostic validity. Receiver operating characteristic (ROC) curves, areas under the ROC curves and decision curve analysis (DCA) results were plotted to evaluate the model performance and clinical application.

Results

The nomogram model exhibited superior performance, with an AUC of 0.905 (95% confidence interval [CI]: 0.857-0.951) in the training cohort, which decreased to 0.817 (95% CI: 0.698-0.936) in the external validation cohort. The clinical model had AUCs of 0.820 (95% CI: 0.753-0.886) and 0.730 (95% CI: 0.572-0.888) in the training and external validation cohorts, respectively. The radiomics model had AUCs of 0.895 (95% CI: 0.840-0.949) and 0.800 (95% CI: 0.675-0.924) for training and external validation, respectively. DCA confirmed that the nomogram model had the better clinical benefit.

Conclusions

The nomogram model achieved promising prediction efficiency for identifying the presence of the MPP in LUAD tumors ≤2 cm, allowing clinicians to develop more rational and efficacious personalized treatment strategies.

Res-TransNet: A Hybrid deep Learning Network for Predicting Pathological Subtypes of lung Adenocarcinoma in CT Images

This study aims to develop a CT-based hybrid deep learning network to predict pathological subtypes of early-stage lung adenocarcinoma by integrating residual network (ResNet) with Vision Transformer (ViT). A total of 1411 pathologically confirmed ground-glass nodules (GGNs) retrospectively collected from two centers were used as internal and external validation sets for model development. 3D ResNet and ViT were applied to investigate two deep learning frameworks to classify three subtypes of lung adenocarcinoma namely invasive adenocarcinoma (IAC), minimally invasive adenocarcinoma and adenocarcinoma in situ, respectively. To further improve the model performance, four Res-TransNet based models were proposed by integrating ResNet and ViT with different ensemble learning strategies. Two classification tasks involving predicting IAC from Non-IAC (Task1) and classifying three subtypes (Task2) were designed and conducted in this study. For Task 1, the optimal Res-TransNet model yielded area under the receiver operating characteristic curve (AUC) values of 0.986 and 0.933 on internal and external validation sets, which were significantly higher than that of ResNet and ViT models (p < 0.05). For Task 2, the optimal fusion model generated the accuracy and weighted F1 score of 68.3% and 66.1% on the external validation set. The experimental results demonstrate that Res-TransNet can significantly increase the classification performance compared with the two basic models and have the potential to assist radiologists in precision diagnosis.

PET/CT radiomics and deep learning in the diagnosis of benign and malignant pulmonary nodules: progress and challenges

Lung cancer is currently the leading cause of cancer-related deaths, and early diagnosis and screening can significantly reduce its mortality rate. Since some early-stage lung cancers lack obvious clinical symptoms and only present as pulmonary nodules (PNs) in imaging examinations, accurately determining the benign or malignant nature of PNs is crucial for improving patient survival rates. 18F-FDG PET/CT is important in diagnosing PNs, but its specificity needs improvement. Radiomics can provide information beyond traditional visual assessment, overcoming its limitations by extracting high-throughput quantitative features from medical images. Radiomics features based on 18F-FDG PET/CT and deep learning methods have shown great potential in the noninvasive diagnosis of PNs. This paper reviews the latest advancements in these methods and discusses their contributions to improving diagnostic accuracy and the challenges they face.

Application of machine learning for the differentiation of thymomas and thymic cysts using deep transfer learning: A multi-center comparison of diagnostic performance based on different dimensional models

OBJECTIVE

This study aimed to evaluate the feasibility and performance of deep transfer learning (DTL) networks with different types and dimensions in differentiating thymomas from thymic cysts in a retrospective cohort.

MATERIALS AND METHODS

Based on chest-enhanced computed tomography (CT), the region of interest was delineated, and the maximum cross section of the lesion was selected as the input image. Five convolutional neural networks (CNNs) and Vision Transformer (ViT) were used to construct a 2D DTL model. The 2D model constructed by the maximum section (n) and the upper and lower layers (n - 1, n + 1) of the lesion was used for feature extraction, and the features were selected. The remaining features were pre-fused to construct a 2.5D model. The whole lesion image was selected for input and constructing a 3D model.

RESULTS

In the 2D model, the area under curve (AUC) of Resnet50 was 0.950 in the training cohort and 0.907 in the internal validation cohort. In the 2.5D model, the AUCs of Vgg11 in the internal validation cohort and external validation cohort 1 were 0.937 and 0.965, respectively. The AUCs of Inception_v3 in the training cohort and external validation cohort 2 were 0.981 and 0.950, respectively. The AUC values of 3D_Resnet50 in the four cohorts were 0.987, 0.937, 0.938, and 0.905.

CONCLUSIONS

The DTL model based on multiple different dimensions can be used as a highly sensitive and specific tool for the non-invasive differential diagnosis of thymomas and thymic cysts to assist clinicians in decision-making.

Standalone deep learning versus experts for diagnosis lung cancer on chest computed tomography: a systematic review

PURPOSE

To compare the diagnostic performance of standalone deep learning (DL) algorithms and human experts in lung cancer detection on chest computed tomography (CT) scans.

MATERIALS AND METHODS

This study searched for studies on PubMed, Embase, and Web of Science from their inception until November 2023. We focused on adult lung cancer patients and compared the efficacy of DL algorithms and expert radiologists in disease diagnosis on CT scans. Quality assessment was performed using QUADAS-2, QUADAS-C, and CLAIM. Bivariate random-effects and subgroup analyses were performed for tasks (malignancy classification vs invasiveness classification), imaging modalities (CT vs low-dose CT [LDCT] vs high-resolution CT), study region, software used, and publication year.

RESULTS

We included 20 studies on various aspects of lung cancer diagnosis on CT scans. Quantitatively, DL algorithms exhibited superior sensitivity (82%) and specificity (75%) compared to human experts (sensitivity 81%, specificity 69%). However, the difference in specificity was statistically significant, whereas the difference in sensitivity was not statistically significant. The DL algorithms' performance varied across different imaging modalities and tasks, demonstrating the need for tailored optimization of DL algorithms. Notably, DL algorithms matched experts in sensitivity on standard CT, surpassing them in specificity, but showed higher sensitivity with lower specificity on LDCT scans.

CONCLUSION

DL algorithms demonstrated improved accuracy over human readers in malignancy and invasiveness classification on CT scans. However, their performance varies by imaging modality, underlining the importance of continued research to fully assess DL algorithms' diagnostic effectiveness in lung cancer.

CLINICAL RELEVANCE STATEMENT

DL algorithms have the potential to refine lung cancer diagnosis on CT, matching human sensitivity and surpassing in specificity. These findings call for further DL optimization across imaging modalities, aiming to advance clinical diagnostics and patient outcomes.

KEY POINTS

Lung cancer diagnosis by CT is challenging and can be improved with AI integration. DL shows higher accuracy in lung cancer detection on CT than human experts. Enhanced DL accuracy could lead to improved lung cancer diagnosis and outcomes.

CT-based conventional radiomics and quantification of intratumoral heterogeneity for predicting benign and malignant renal lesions

BACKGROUND

With the increasing incidence of renal lesions, pretreatment differentiation between benign and malignant lesions is crucial for optimized management. This study aimed to develop a machine learning model utilizing radiomic features extracted from various regions of interest (ROIs), intratumoral ecological diversity features, and clinical factors to classify renal lesions.

METHODS

CT images (arterial phase) of 1,795 renal lesions with confirmed pathology from three hospital sites were split into development (1184, 66%) and test (611, 34%) cohorts by surgery date. Conventional radiomic features were extracted from eight ROIs of arterial phase images. Intratumoral ecological diversity features were derived from intratumoral subregions. The combined model incorporating these features with clinical factors was developed, and its performance was compared with radiologists' interpretation.

RESULTS

Combining intratumoral and peritumoral radiomic features, along with ecological diversity features yielded the highest AUC of 0.929 among all combinations of features extracted from CT scans. After incorporating clinical factors into the features extracted from CT images, our combined model outperformed the interpretation of radiologists in the whole (AUC = 0.946 vs 0.823, P < 0.001) and small renal lesion (AUC = 0.935 vs 0.745, P < 0.001) test cohorts. Furthermore, the combined model exhibited favorable concordance and provided the highest net benefit across threshold probabilities exceeding 60%. In the whole and small renal lesion test cohorts, the AUCs for subgroups with predicted risk below or above 95% sensitivity and specificity cutoffs were 0.974 and 0.978, respectively.

CONCLUSIONS

The combined model, incorporating intratumoral and peritumoral radiomic features, ecological diversity features, and clinical factors showed good performance for distinguishing benign from malignant renal lesions, surpassing radiologists' diagnoses in both whole and small renal lesions. It has the potential to save patients from unnecessary invasive biopsies/surgeries and to enhance clinical decision-making.

2.5D peritumoural radiomics predicts postoperative recurrence in stage I lung adenocarcinoma

Objective

Radiomics can non-invasively predict the prognosis of a tumour by applying advanced imaging feature algorithms.The aim of this study was to predict the chance of postoperative recurrence by modelling tumour radiomics and peritumour radiomics and clinical features in patients with stage I lung adenocarcinoma (LUAD).

Materials and methods

Retrospective analysis of 190 patients with postoperative pathologically confirmed stage I LUAD from centre 1, who were divided into training cohort and internal validation cohort, with centre 2 added as external validation cohort. To develop a combined radiation-clinical omics model nomogram incorporating clinical features based on images from low-dose lung cancer screening CT plain for predicting postoperative recurrence and to evaluate the performance of the nomogram in the training cohort, internal validation cohort and external validation cohort.

Results

A total of 190 patients were included in the model in centre 1 and randomised into a training cohort of 133 and an internal validation cohort of 57 in a ratio of 7:3, and 39 were included in centre 2 as an external validation cohort. In the training cohort (AUC=0.865, 95% CI 0.824-0.906), internal validation cohort (AUC=0.902, 95% CI 0.851-0.953) and external validation cohort (AUC=0.830,95% CI 0.751-0.908), the combined radiation-clinical omics model had a good predictive ability. The combined model performed significantly better than the conventional single-modality models (clinical model, radiomic model), and the calibration curve and decision curve analysis (DCA) showed high accuracy and clinical utility of the nomogram.

Conclusion

The combined preoperative radiation-clinical omics model provides good predictive value for postoperative recurrence in stage ILUAD and combines the model's superiority in both internal and external validation cohorts, demonstrating its potential to aid in postoperative treatment strategies.

CT-Based Super-Resolution Deep Learning Models with Attention Mechanisms for Predicting Spread Through Air Spaces of Solid or Part-Solid Lung Adenocarcinoma

RATIONALE AND OBJECTIVES

Spread through air space (STAS) is a novel invasive pattern of lung adenocarcinoma (LUAD), and preoperative knowledge of STAS status is helpful in choosing an appropriate surgical approach.

MATERIALS AND METHODS

This retrospective study collected and analyzed 602 patients diagnosed with LUAD from two medical centers: center 1 was randomly partitioned into training (n = 358) and validation cohorts (n = 154) at a 7:3 ratio; and center 2 was the external test cohort (n = 90). Super resolution was performed on all images to acquire high-resolution images, which were used to train the SE-ResNet50 model, before creating an equivalent parameter ResNet50 model. Disparities were compared between the two models using receiver operating characteristic curves, area under the curve, accuracy, precision, sensitivity, and specificity.

RESULTS

In this study, 512 and 90 patients with LUAD were enrolled from centers 1 and 2, respectively. The curve values of the SE-ResNet50 and ResNet50 models were compared for training, validation, and test cohorts, resulting in values of 0.933 vs 0.909, 0.783 vs 0.728, and 0.806 vs 0.695, respectively. In the external test cohort, the accuracy of the SE-ResNet50 model demonstrated a 10% improvement over the ResNet50 model (82.2% vs 72.2%).

CONCLUSION

The SE-ResNet50 model based on computed tomography super-resolution has great potential for predicting STAS status in patients with solid or partially solid LUAD, with superior predictive performance compared to traditional deep learning models.

Deep learning for risk stratification of thymoma pathological subtypes based on preoperative CT images

OBJECTIVES

This study aims to develop an innovative, deep model for thymoma risk stratification using preoperative CT images. Current algorithms predominantly focus on radiomic features or 2D deep features and require manual tumor segmentation by radiologists, limiting their practical applicability.

METHODS

The deep model was trained and tested on a dataset comprising CT images from 147 patients (82 female; mean age, 54 years ± 10) who underwent surgical resection and received subsequent pathological confirmation. The eligible participants were divided into a training cohort (117 patients) and a testing cohort (30 patients) based on the CT scan time. The model consists of two stages: 3D tumor segmentation and risk stratification. The radiomic model and deep model (2D) were constructed for comparative analysis. Model performance was evaluated through dice coefficient, area under the curve (AUC), and accuracy.

RESULTS

In both the training and testing cohorts, the deep model demonstrated better performance in differentiating thymoma risk, boasting AUCs of 0.998 and 0.893 respectively. This was compared to the radiomic model (AUCs of 0.773 and 0.769) and deep model (2D) (AUCs of 0.981 and 0.760). Notably, the deep model was capable of simultaneously identifying lesions, segmenting the region of interest (ROI), and differentiating the risk of thymoma on arterial phase CT images. Its diagnostic prowess outperformed that of the baseline model.

CONCLUSIONS

The deep model has the potential to serve as an innovative decision-making tool, assisting on clinical prognosis evaluation and the discernment of suitable treatments for different thymoma pathological subtypes.

KEY POINTS

• This study incorporated both tumor segmentation and risk stratification. • The deep model, using clinical and 3D deep features, effectively predicted thymoma risk. • The deep model improved AUCs by 16.1pt and 17.5pt compared to radiomic model and deep model (2D) respectively.

Predicting Invasiveness of Lung Adenocarcinoma at Chest CT with Deep Learning Ternary Classification Models

Background Preoperative discrimination of preinvasive, minimally invasive, and invasive adenocarcinoma at CT informs clinical management decisions but may be challenging for classifying pure ground-glass nodules (pGGNs). Deep learning (DL) may improve ternary classification. Purpose To determine whether a strategy that includes an adjudication approach can enhance the performance of DL ternary classification models in predicting the invasiveness of adenocarcinoma at chest CT and maintain performance in classifying pGGNs. Materials and Methods In this retrospective study, six ternary models for classifying preinvasive, minimally invasive, and invasive adenocarcinoma were developed using a multicenter data set of lung nodules. The DL-based models were progressively modified through framework optimization, joint learning, and an adjudication strategy (simulating a multireader approach to resolving discordant nodule classifications), integrating two binary classification models with a ternary classification model to resolve discordant classifications sequentially. The six ternary models were then tested on an external data set of pGGNs imaged between December 2019 and January 2021. Diagnostic performance including accuracy, specificity, and sensitivity was assessed. The χ2 test was used to compare model performance in different subgroups stratified by clinical confounders. Results A total of 4929 nodules from 4483 patients (mean age, 50.1 years ± 9.5 [SD]; 2806 female) were divided into training (n = 3384), validation (n = 579), and internal (n = 966) test sets. A total of 361 pGGNs from 281 patients (mean age, 55.2 years ± 11.1 [SD]; 186 female) formed the external test set. The proposed strategy improved DL model performance in external testing (P < .001). For classifying minimally invasive adenocarcinoma, the accuracy was 85% and 79%, sensitivity was 75% and 63%, and specificity was 89% and 85% for the model with adjudication (model 6) and the model without (model 3), respectively. Model 6 showed a relatively narrow range (maximum minus minimum) across diagnostic indexes (accuracy, 1.7%; sensitivity, 7.3%; specificity, 0.9%) compared with the other models (accuracy, 0.6%-10.8%; sensitivity, 14%-39.1%; specificity, 5.5%-17.9%). Conclusion Combining framework optimization, joint learning, and an adjudication approach improved DL classification of adenocarcinoma invasiveness at chest CT. Published under a CC BY 4.0 license. Supplemental material is available for this article. See also the editorial by Sohn and Fields in this issue.

Development and Validation of Contrast-Enhanced CT-Based Deep Transfer Learning and Combined Clinical-Radiomics Model to Discriminate Thymomas and Thymic Cysts: A Multicenter Study

RATIONALE AND OBJECTIVES

This study aims to evaluate the feasibility and effectiveness of deep transfer learning (DTL) and clinical-radiomics in differentiating thymoma from thymic cysts.

MATERIALS AND METHODS

Clinical and imaging data of 196 patients pathologically diagnosed with thymoma and thymic cysts were retrospectively collected from center 1. (training cohort: n = 137; internal validation cohort: n = 59). An independent external validation cohort comprised 68 thymoma and thymic cyst patients from center 2. Region of interest (ROI) delineation was performed on contrast-enhanced chest computed tomography (CT) images, and eight DTL models including Densenet 169, Mobilenet V2, Resnet 101, Resnet 18, Resnet 34, Resnet 50, Vgg 13, Vgg 16 were constructed. Radiomics features were extracted from the ROI on the CT images of thymoma and thymic cyst patients, and feature selection was performed using intra-observer correlation coefficient (ICC), Spearman correlation analysis, and least absolute shrinkage and selection operator (LASSO) algorithm. Univariate analysis and multivariable logistic regression (LR) were used to select clinical-radiological features. Six machine learning classifiers, including LR, support vector machine (SVM), k-nearest neighbors (KNN), Light Gradient Boosting Machine (LightGBM), Adaptive Boosting (AdaBoost), and Multilayer Perceptron (MLP), were used to construct Radiomics and Clinico-radiologic models. The selected features from the Radiomics and Clinico-radiologic models were fused to build a Combined model. Receiver operating characteristic curve (ROC), calibration curve, and decision curve analysis (DCA) were used to evaluate the discrimination, calibration, and clinical utility of the models, respectively. The Delong test was used to compare the AUC between different models. K-means clustering was used to subdivide the lesions of thymomas or thymic cysts into subregions, and traditional radiomics methods were used to extract features and compare the ability of Radiomics and DTL models to reflect intratumoral heterogeneity using correlation analysis.

RESULTS

The Densenet 169 based on DTL performed the best, with AUC of 0.933 (95% CI: 0.875-0.991) in the internal validation cohort and 0.962 (95% CI: 0.923-1.000) in the external validation cohort. The AdaBoost classifier achieved AUC of 0.965 (95% CI: 0.923-1.000) and 0.959 (95% CI: 0.919-1.000) in the internal and external validation cohorts, respectively, for the Radiomics model. The LightGBM classifier achieved AUC of 0.805 (95% CI: 0.690-0.920) and 0.839 (95% CI: 0.736-0.943) in the Clinico-radiologic model. The AUC of the Combined model in the internal and external validation cohorts was 0.933 (95% CI: 0.866-1.000) and 0.945 (95% CI: 0.897-0.994), respectively. The results of the Delong test showed that the Radiomics model, DTL model, and Combined model outperformed the Clinico-radiologic model in both internal and external validation cohorts (p-values were 0.002, 0.004, and 0.033 in the internal validation cohort, while in the external validation cohort, the p-values were 0.014, 0.006, and 0.015, respectively). But there was no statistical difference in performance among the three models (all p-values <0.05). Correlation analysis showed that radiomics performed better than DTL in quantifying intratumoral heterogeneity differences between thymoma and thymic cysts.

CONCLUSION

The developed DTL model and the Combined model based on radiomics and clinical-radiologic features achieved excellent diagnostic performance in differentiating thymic cysts from thymoma. They can serve as potential tools to assist clinical decision-making, particularly when endoscopic biopsy carries a high risk.

[The contribution of artificial intelligence (AI) subsequent to the processing of thoracic imaging]

The contribution of artificial intelligence (AI) to medical imaging is currently the object of widespread experimentation. The development of deep learning (DL) methods, particularly convolution neural networks (CNNs), has led to performance gains often superior to those achieved by conventional methods such as machine learning. Radiomics is an approach aimed at extracting quantitative data not accessible to the human eye from images expressing a disease. The data subsequently feed machine learning models and produce diagnostic or prognostic probabilities. As for the multiple applications of AI methods in thoracic imaging, they are undergoing evaluation. Chest radiography is a practically ideal field for the development of DL algorithms able to automatically interpret X-rays. Current algorithms can detect up to 14 different abnormalities present either in isolation or in combination. Chest CT is another area offering numerous AI applications. Various algorithms have been specifically formed and validated for the detection and characterization of pulmonary nodules and pulmonary embolism, as well as segmentation and quantitative analysis of the extent of diffuse lung diseases (emphysema, infectious pneumonias, interstitial lung disease). In addition, the analysis of medical images can be associated with clinical, biological, and functional data (multi-omics analysis), the objective being to construct predictive approaches regarding disease prognosis and response to treatment.

Comparing three-dimensional and two-dimensional deep-learning, radiomics, and fusion models for predicting occult lymph node metastasis in laryngeal squamous cell carcinoma based on CT imaging: a multicentre, retrospective, diagnostic study

Background

The occult lymph node metastasis (LNM) of laryngeal squamous cell carcinoma (LSCC) affects the treatment and prognosis of patients. This study aimed to comprehensively compare the performance of the three-dimensional and two-dimensional deep learning models, radiomics model, and the fusion models for predicting occult LNM in LSCC.

Methods

In this retrospective diagnostic study, a total of 553 patients with clinical N0 stage LSCC, who underwent surgical treatment without distant metastasis and multiple primary cancers, were consecutively enrolled from four Chinese medical centres between January 01, 2016 and December 30, 2020. The participant data were manually retrieved from medical records, imaging databases, and pathology reports. The study cohort was divided into a training set (n = 300), an internal test set (n = 89), and two external test sets (n = 120 and 44, respectively). The three-dimensional deep learning (3D DL), two-dimensional deep learning (2D DL), and radiomics model were developed using CT images of the primary tumor. The clinical model was constructed based on clinical and radiological features. Two fusion strategies were utilized to develop the fusion model: the feature-based DLRad_FB model and the decision-based DLRad_DB model. The discriminative ability and correlation of 3D DL, 2D DL and radiomics features were analysed comprehensively. The performances of the predictive models were evaluated based on the pathological diagnosis.

Findings

The 3D DL features had superior discriminative ability and lower internal redundancy compared to 2D DL and radiomics features. The DLRad_DB model achieved the highest AUC (0.89-0.90) among all the study sets, significantly outperforming the clinical model (AUC = 0.73-0.78, P = 0.0001-0.042, Delong test). Compared to the DLRad_DB model, the AUC values for the DLRad_FB, 3D DL, 2D DL, and radiomics models were 0.82-0.84 (P = 0.025-0.46), 0.86-0.89 (P = 0.75-0.97), 0.83-0.86 (P = 0.029-0.66), and 0.79-0.82 (P = 0.0072-0.10), respectively in the study sets. Additionally, the DLRad_DB model exhibited the best sensitivity (82-88%) and specificity (79-85%) in the test sets.

Interpretation

The decision-based fusion model DLRad_DB, which combines 3D DL, 2D DL, radiomics, and clinical data, can be utilized to predict occult LNM in LSCC. This has the potential to minimize unnecessary lymph node dissection and prophylactic radiotherapy in patients with cN0 disease.

Funding

National Natural Science Foundation of China, Natural Science Foundation of Shandong Province.

An ensemble deep learning model for risk stratification of invasive lung adenocarcinoma using thin-slice CT

Lung cancer screening using computed tomography (CT) has increased the detection rate of small pulmonary nodules and early-stage lung adenocarcinoma. It would be clinically meaningful to accurate assessment of the nodule histology by CT scans with advanced deep learning algorithms. However, recent studies mainly focus on predicting benign and malignant nodules, lacking of model for the risk stratification of invasive adenocarcinoma. We propose an ensemble multi-view 3D convolutional neural network (EMV-3D-CNN) model to study the risk stratification of lung adenocarcinoma. We include 1075 lung nodules (≤30 mm and ≥4 mm) with preoperative thin-section CT scans and definite pathology confirmed by surgery. Our model achieves a state-of-art performance of 91.3% and 92.9% AUC for diagnosis of benign/malignant and pre-invasive/invasive nodules, respectively. Importantly, our model outperforms senior doctors in risk stratification of invasive adenocarcinoma with 77.6% accuracy [i.e., Grades 1, 2, 3]). It provides detailed predictive histological information for the surgical management of pulmonary nodules. Finally, for user-friendly access, the proposed model is implemented as a web-based system ( https://seeyourlung.com.cn ).

Imaging of lung cancer

Lung cancer is the leading cause of cancer-related mortality globally. Imaging is essential in the screening, diagnosis, staging, response assessment, and surveillance of patients with lung cancer. Subtypes of lung cancer can have distinguishing imaging appearances. The most frequently used imaging modalities include chest radiography, computed tomography, magnetic resonance imaging, and positron emission tomography. Artificial intelligence algorithms and radiomics are emerging technologies with potential applications in lung cancer imaging.

Neural network-based model for evaluating inert nodules and volume doubling time in T1 lung adenocarcinoma: a nested case-control study

Objective

The purpose of this study is to establish model for assessing inert nodules predicting nodule volume-doubling.

Methods

A total of 201 patients with T1 lung adenocarcinoma were analysed retrospectively pulmonary nodule information was predicted by an AI pulmonary nodule auxiliary diagnosis system. The nodules were classified into two groups: inert nodules (volume-doubling time (VDT)>600 days n=152) noninert nodules (VDT<600 days n=49). Then taking the clinical imaging features obtained at the first examination as predictive variables the inert nodule judgement model <sn</sn>>(INM) volume-doubling time estimation model (VDTM) were constructed based on a deep learning-based neural network. The performance of the INM was evaluated by the area under the curve (AUC) obtained from receiver operating characteristic (ROC) analysis the performance of the VDTM was evaluated by R²(determination coefficient).

Results

The accuracy of the INM in the training and testing cohorts was 81.13% and 77.50%, respectively. The AUC of the INM in the training and testing cohorts was 0.7707 (95% CI 0.6779-0.8636) and 0.7700 (95% CI 0.5988-0.9412), respectively. The INM was effective in identifying inert pulmonary nodules; additionally, the R2 of the VDTM in the training cohort was 0.8008, and that in the testing cohort was 0.6268. The VDTM showed moderate performance in estimating the VDT, which can provide some reference during a patients' first examination and consultation.

Conclusion

The INM and the VDTM based on deep learning can help radiologists and clinicians distinguish among inert nodules and predict the nodule volume-doubling time to accurately treat patients with pulmonary nodules.

Machine Learning and Deep Learning in Cardiothoracic Imaging: A Scoping Review

Machine-learning (ML) and deep-learning (DL) algorithms are part of a group of modeling algorithms that grasp the hidden patterns in data based on a training process, enabling them to extract complex information from the input data. In the past decade, these algorithms have been increasingly used for image processing, specifically in the medical domain. Cardiothoracic imaging is one of the early adopters of ML/DL research, and the COVID-19 pandemic resulted in more research focus on the feasibility and applications of ML/DL in cardiothoracic imaging. In this scoping review, we systematically searched available peer-reviewed medical literature on cardiothoracic imaging and quantitatively extracted key data elements in order to get a big picture of how ML/DL have been used in the rapidly evolving cardiothoracic imaging field. During this report, we provide insights on different applications of ML/DL and some nuances pertaining to this specific field of research. Finally, we provide general suggestions on how researchers can make their research more than just a proof-of-concept and move toward clinical adoption.

Visceral adipose volume is correlated with surgical tissue fibrosis in Crohn's disease of the small bowel

Background

This study explored the diagnostic performance of visceral adiposity to predict the degree of intestinal inflammation and fibrosis.

Methods

The patients with Crohn’s disease (CD) who underwent surgical small bowel resection at the First Affiliated Hospital of Sun Yat-sen University (Guangzhou, China) between January 2007 and December 2017 were enrolled. We evaluated the intestinal imaging features of computed tomography enterography (CTE), including mesenteric inflammatory fat stranding, the target sign, mesenteric hypervascularity, bowel wall thickening, lymphadenopathy, stricture diameter, and maximal upstream diameter. We used A.K. software (Artificial Intelligence Kit, version 1.1) to calculate the visceral fat (VF) and subcutaneous fat (SF) volumes at the third lumbar vertebra level. Pathological tissue information was recorded. Diagnostic models were established based on the multivariate regression analysis results, and their effectiveness was evaluated by area under the curve (AUC) and decision curve analyses.

Results

Overall, 48 patients with CD were included in this study. The abdominal VF/SF volume ratio (odds ratio, 1.20; 95% confidence interval, 1.05–1.38; P = 0.009) and the stenosis diameter/upstream intestinal dilatation diameter (ND) ratio (odds ratio, 0.90; 95% confidence interval, 0.82–0.99; P = 0.034) were independent risk factors for the severe fibrosis of the small intestine. The AUC values of the VF/SF ratio, the ND ratio, and their combination were 0.760, 0.673, and 0.804, respectively. The combination of the VS/SF volume ratio and ND ratio achieved the highest net benefit on the decision curve.

Conclusion

The VF volume on CTE can reflect intestinal fibrosis. The combination of the VF/SF volume ratio and ND ratio of CD patients assessed using CTE can help predict severe fibrosis stenosis of the small intestine.

A radiomics nomogram for invasiveness prediction in lung adenocarcinoma manifesting as part-solid nodules with solid components smaller than 6 mm

Objective

To investigate whether radiomics can help radiologists and thoracic surgeons accurately predict invasive adenocarcinoma (IAC) manifesting as part-solid nodules (PSNs) with solid components <6 mm and provide a basis for rational clinical decision-making.

Materials and Methods

In total, 1,210 patients (mean age ± standard deviation: 54.28 ± 11.38 years, 374 men and 836 women) from our hospital and another hospital with 1,248 PSNs pathologically diagnosed with adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or IAC were enrolled in this study. Among them, 1,050 cases from our hospital were randomly divided into a derivation set (n = 735) and an internal validation set (n = 315), 198 cases from another hospital were used for external validation. Each labeled nodule was segmented, and 105 radiomics features were extracted. Least absolute shrinkage and selection operator (LASSO) was used to calculate Rad-score and build the radiomics model. Multivariable logistic regression was conducted to identify the clinicoradiological predictors and establish the clinical-radiographic model. The combined model and predictive nomogram were developed based on identified clinicoradiological independent predictors and Rad-score using multivariable logistic regression analysis. The predictive performances of the three models were compared via receiver operating characteristic (ROC) curve analysis. Decision curve analysis (DCA) was performed on both the internal and external validation sets to evaluate the clinical utility of the nomogram.

Results

The radiomics model showed superior predictive performance than the clinical-radiographic model in both internal and external validation sets (Az values, 0.884 vs. 0.810, p = 0.001; 0.924 vs. 0.855, p < 0.001, respectively). The combined model showed comparable predictive performance to the radiomics model (Az values, 0.887 vs. 0.884, p = 0.398; 0.917 vs. 0.924, p = 0.271, respectively). The clinical application value of the nomogram developed based on the Rad-score, maximum diameter, and lesion shape was confirmed, and DCA demonstrated that application of the Rad-score would be beneficial for radiologists predicting invasive lesions.

Conclusions

Radiomics has the potential as an independent diagnostic tool to predict the invasiveness of PSNs with solid components <6 mm.

ViSTA: A Novel Network Improving Lung Adenocarcinoma Invasiveness Prediction from Follow-Up CT Series

Simple Summary

Assessing follow-up computed tomography(CT) series is of great importance in clinical practice for lung nodule diagnosis. Deep learning is a thriving data mining method in medical imaging and has obtained surprising results. However, previous studies mostly focused on the analysis of single static time points instead of the entire follow-up series and required regular intervals between CT examinations. In the current study, we propose a new deep learning framework, named ViSTA, that can better evaluate tumor invasiveness using irregularly serial follow-up CT images to avoid aggressive procedures or delay diagnosis in clinical practice. ViSTA provides a new solution for irregularly sampled data. ViSTA delivers superior performance compared with other static or serial deep learning models. The proposed ViSTA framework is capable of improving performance close to the human level in the prediction of invasiveness of lung adenocarcinoma while being transferrable to other tasks analyzing serial medical data.

Abstract

To investigate the value of the deep learning method in predicting the invasiveness of early lung adenocarcinoma based on irregularly sampled follow-up computed tomography (CT) scans. In total, 351 nodules were enrolled in the study. A new deep learning network based on temporal attention, named Visual Simple Temporal Attention (ViSTA), was proposed to process irregularly sampled follow-up CT scans. We conducted substantial experiments to investigate the supplemental value in predicting the invasiveness using serial CTs. A test set composed of 69 lung nodules was reviewed by three radiologists. The performance of the model and radiologists were compared and analyzed. We also performed a visual investigation to explore the inherent growth pattern of the early adenocarcinomas. Among counterpart models, ViSTA showed the best performance (AUC: 86.4% vs. 60.6%, 75.9%, 66.9%, 73.9%, 76.5%, 78.3%). ViSTA also outperformed the model based on Volume Doubling Time (AUC: 60.6%). ViSTA scored higher than two junior radiologists (accuracy of 81.2% vs. 75.4% and 71.0%) and came close to the senior radiologist (85.5%). Our proposed model using irregularly sampled follow-up CT scans achieved promising accuracy in evaluating the invasiveness of the early stage lung adenocarcinoma. Its performance is comparable with senior experts and better than junior experts and traditional deep learning models. With further validation, it can potentially be applied in clinical practice.

A narrative review of deep learning applications in lung cancer research: from screening to prognostication

Background and Objective

Deep learning (DL) algorithms have been developed for various tasks, including lung nodule detection on chest radiographs or lung cancer computed tomography screening, potential candidate selection in lung cancer screening, malignancy prediction for indeterminate pulmonary nodules, lung cancer staging, treatment response prediction, prognostication, and prediction of genetic mutations in lung cancer. Furthermore, these DL algorithms have been applied in various clinical settings in order for them to be generalized in real-world clinical practice. Multiple DL algorithms have been corroborated to be on par with experts or current clinical prediction models for several specific tasks. However, no article has yet comprehensively reviewed DL algorithms dedicated to lung cancer research. This narrative review presents an overview of the literature dealing with DL techniques applied in lung cancer research and briefly summarizes the results according to the DL algorithms’ clinical use cases.

Methods

we performed a narrative review by searching the Embase and OVID-MEDLINE databases for articles published in English from October, 2016 until September, 2021 and reviewing the bibliographies of key references to identify important literature related to DL in lung cancer research. The background, development, results, and clinical implications of each DL algorithm are briefly discussed. Lastly, we end this review article by highlighting future directions in lung cancer research using DL techniques.

Key Content and Findings

DL algorithms have been introduced to show comparable or higher performance than human experts in various clinical settings. Specifically, they have been actively applied to detect lung nodules in chest radiographs or computed tomography (CT) examinations, optimize candidate selection for lung cancer screening (LCS), predict the malignancy of lung nodules, stage lung cancer, and predict treatment response, patients’ prognoses, and genetic mutations in lung cancers.

Conclusions

DL algorithms have corroborated their potential value for various tasks, ranging from lung cancer screening to prognostication of lung cancer patients. Future research is warranted for the clinical application of these algorithms in daily clinical practice and verification of their real-world clinical usefulness.

A deep learning algorithm to improve readers' interpretation and speed of pancreatic cystic lesions on dual-phase enhanced CT

PURPOSE

To develop a deep learning model (DLM) to improve readers' interpretation and speed in the differentiation of pancreatic cystic lesions (PCLs) on dual-phase enhanced CT, and a low contrast media dose, external testing set validated the model.

MATERIALS AND METHODS

Dual-phase enhanced CT images of 363 patients with 368 PCLs obtained from two centers were retrospectively assessed. Based on the examination date, a training and validation set of 266 PCLs, an internal testing set of 52 PCLs were designated from center 1. An external testing set included 50 PCLs from center 2. Clinical and radiological characteristics were compared. The DLM was developed using 3D specially designed densely connected convolutional networks for PCL differentiation. Radiomic features were extracted to build a traditional radiomics model (RM). Performance of the DLM, traditional RM, and three readers was compared.

RESULTS

The accuracy for differential diagnosis was 0.904 with DLM, which was the highest in the internal testing set. Accuracy differences between the DLM and senior radiologist were not significant both in the internal and external testing set (both p > 0.05). With the help of the DLM, the accuracy and specificity of the junior radiologist were significantly improved (all p < 0.05), and all readers' diagnostic time was shortened (all p < 0.05).

CONCLUSION

The DLM achieved senior radiologist-level performance in differentiating benign and malignant PCLs which could improve the junior radiologist's interpretation and speed of PCLs on CT.

Clinical impact of a deep learning system for automated detection of missed pulmonary nodules on routine body computed tomography including the chest region

OBJECTIVES

To evaluate the clinical impact of a deep learning system (DLS) for automated detection of pulmonary nodules on computed tomography (CT) images as a second reader.

METHODS

This single-centre retrospective study screened 21,150 consecutive body CT studies from September 2018 to February 2019. Pulmonary nodules detected by the DLS on axial CT images but not mentioned in initial radiology reports were flagged. Flagged images were scored by four board-certificated radiologists each with at least 5 years of experience. Nodules with scores of 2 (understandable miss) or 3 (should not be missed) were then categorised as unlikely to be clinically significant (2a or 3a) or likely to be clinically significant (2b or 3b) according to the 2017 Fleischner guidelines for pulmonary nodules. The miss rate was defined as the total number of studies receiving scores of 2 or 3 divided by total screened studies.

RESULTS

Among 172 nodules flagged by the DLS, 60 (35%) missed nodules were confirmed by the radiologists. The nodules were further categorised as 2a, 2b, 3a, and 3b in 24, 14, 10, and 12 studies, respectively, with an overall positive predictive value of 35%. Missed pulmonary nodules were identified in 0.3% of all CT images, and one-third of these lesions were considered clinically significant.

CONCLUSIONS

Use of DLS-assisted automated detection as a second reader can identify missed pulmonary nodules, some of which may be clinically significant.

CLINICAL RELEVANCE/APPLICATION

Use of DLS to help radiologists detect pulmonary lesions may improve patient care.

KEY POINTS

• DLS-assisted automated detection as a second reader is feasible in a large consecutive cohort. • Performance of combined radiologists and DLS was better than DLS or radiologists alone. • Pulmonary nodules were missed more frequently in abdomino-pelvis CT than the thoracic CT.

[Clinical Study of Artificial Intelligence-assisted Diagnosis System in Predicting the Invasive Subtypes of Early-stage Lung Adenocarcinoma Appearing as Pulmonary Nodules]

BACKGROUND

Lung cancer is the cancer with the highest mortality at home and abroad at present. The detection of lung nodules is a key step to reducing the mortality of lung cancer. Artificial intelligence-assisted diagnosis system presents as the state of the art in the area of nodule detection, differentiation between benign and malignant and diagnosis of invasive subtypes, however, a validation with clinical data is necessary for further application. Therefore, the aim of this study is to evaluate the effectiveness of artificial intelligence-assisted diagnosis system in predicting the invasive subtypes of early‑stage lung adenocarcinoma appearing as pulmonary nodules.

METHODS

Clinical data of 223 patients with early-stage lung adenocarcinoma appearing as pulmonary nodules admitted to the Lanzhou University Second Hospital from January 1st, 2016 to December 31th, 2021 were retrospectively analyzed, which were divided into invasive adenocarcinoma group (n=170) and non-invasive adenocarcinoma group (n=53), and the non-invasive adenocarcinoma group was subdivided into minimally invasive adenocarcinoma group (n=31) and preinvasive lesions group (n=22). The malignant probability and imaging characteristics of each group were compared to analyze their predictive ability for the invasive subtypes of early-stage lung adenocarcinoma. The concordance between qualitative diagnostic results of artificial intelligence-assisted diagnosis of the invasive subtypes of early-stage lung adenocarcinoma and postoperative pathology was then analyzed.

RESULTS

In different invasive subtypes of early-stage lung adenocarcinoma, the mean CT value of pulmonary nodules (P<0.001), diameter (P<0.001), volume (P<0.001), malignant probability (P<0.001), pleural retraction sign (P<0.001), lobulation (P<0.001), spiculation (P<0.001) were significantly different. At the same time, it was also found that with the increased invasiveness of different invasive subtypes of early-stage lung adenocarcinoma, the proportion of dominant signs of each group gradually increased. On the issue of binary classification, the sensitivity, specificity, and area under the curve (AUC) values of the artificial intelligence-assisted diagnosis system for the qualitative diagnosis of invasive subtypes of early-stage lung adenocarcinoma were 81.76%, 92.45% and 0.871 respectively. On the issue of three classification, the accuracy, recall rate, F1 score, and AUC values of the artificial intelligence-assisted diagnosis system for the qualitative diagnosis of invasive subtypes of early-stage lung adenocarcinoma were 83.86%, 85.03%, 76.46% and 0.879 respectively.

CONCLUSIONS

Artificial intelligence-assisted diagnosis system could predict the invasive subtypes of early‑stage lung adenocarcinoma appearing as pulmonary nodules, and has a certain predictive value. With the optimization of algorithms and the improvement of data, it may provide guidance for individualized treatment of patients.

Identification of Benign and Malignant Lung Nodules in CT Images Based on Ensemble Learning Method

BACKGROUND AND OBJECTIVE

Under the background of urgent need for computer-aided technology to provide physicians with objective decision support, aiming at reducing the false positive rate of nodule CT detection in pulmonary nodules detection and improving the accuracy of lung nodule recognition, this paper puts forward a method based on ensemble learning to distinguish between malignant and benign pulmonary nodules.

METHODS

Firstly, trained on a public data set, a multi-layer feature fusion YOLOv3 network is used to detect lung nodules. Secondly, a CNN was trained to differentiate benign from malignant pulmonary nodules. Then, based on the idea of ensemble learning, the confidence probability of the above two models and the label of the training set are taken as data features to build a Logistic regression model. Finally, two test sets (public data set and private data set) were tested, and the confidence probability output by the two models was fused into the established logistic regression model to determine benign and malignant pulmonary nodules.

RESULTS

The YOLOv3 network was trained to detect chest CT images of the test set. The number of pulmonary nodules detected in the public and private test sets was 356 and 314, respectively. The accuracy, sensitivity and specificity of the two test sets were 80.97%, 81.63%, 78.75% and 79.69%, 86.59%, 72.16%, respectively. With CNN training pulmonary nodules benign and malignant discriminant model analysis of two kinds of test set, the result of accuracy, sensitivity and specificity were 90.12%, 90.66%, 89.47% and 88.57%, 85.62%, 90.87%, respectively. Fused model based on YOLOv3 network and CNN is tested on two test sets, and the result of accuracy, sensitivity and specificity were 93.82%, 94.85%, 92.59% and 92.31%, 92.68%, 91.89%, respectively.

CONCLUSION

The ensemble learning model is more effective than YOLOv3 network and CNN in removing false positives, and the accuracy of the ensemble. Learning model is higher than the other two networks in identifying pulmonary nodules.

Preoperative CT-Based Radiomics Combined With Nodule Type to Predict the Micropapillary Pattern in Lung Adenocarcinoma of Size 2 cm or Less: A Multicenter Study

Purpose

To construct an optimal radiomics model for preoperative prediction micropapillary pattern (MPP) in adenocarcinoma (ADC) of size ≤ 2 cm, nodule type was used for stratification to construct two radiomics models based on high-resolution computed tomography (HRCT) images.

Materials and Methods

We retrospectively analyzed patients with pathologically confirmed ADC of size ≤ 2 cm who presented to three hospitals. Patients presenting to the hospital with the greater number of patients were included in the training set (n = 2386) and those presenting to the other two hospitals were included in the external validation set (n = 119). HRCT images were used for delineation of region of interest of tumor and extraction of radiomics features; dimensionality reduction was performed for the features. Nodule type was used to stratify the data and the random forest method was used to construct two models for preoperative prediction MPP in ADC of size ≤ 2 cm. Model 1 included all nodule types and model 2 included only solid nodules. The receiver operating characteristic curve was used to assess the prediction performance of the two models and independent validation was used to assess its generalizability.

Results

Both models predicted ADC with MPP preoperatively. The area under the curve (AUC) of prediction performance of models 1 and 2 were 0.91 and 0.78, respectively. The prediction performance of model 2 was lower than that of model 1. The AUCs in the external validation set were 0.81 and 0.72, respectively. The DeLong test showed statistically significant differences between the training and validation sets in model 1 (p = 0.0296) with weak generalizability. There was no statistically significant difference between the training and validation sets in model 2 (p = 0.2865) with some generalizability.

Conclusion

Nodule type is an important factor that affects the performance of radiomics predictor model for MPP with ADC of size ≤ 2 cm. The radiomics prediction model constructed based on solid nodules alone, can be used to evaluate MPP and may contribute to proper surgical planning in patients with ADC of size ≤ 2 cm.

Applications of artificial intelligence in the thorax: a narrative review focusing on thoracic radiology

OBJECTIVE

This review will focus on how AI-and, specifically, deep learning-can be applied to complement aspects of the current healthcare system. We describe how AI-based tools can augment existing clinical workflows by discussing the applications of AI to worklist prioritization and patient triage, the performance-boosting effects of AI as a second reader, and the use of AI to facilitate complex quantifications. We also introduce prominent examples of recent AI applications, such as tuberculosis screening in resource-constrained environments, the detection of lung cancer with screening CT, and the diagnosis of COVID-19. We also provide examples of prognostic predictions and new discoveries beyond existing clinical practices.

BACKGROUND

Artificial intelligence (AI) has shown promising performance for thoracic diseases, particularly in the field of thoracic radiology. However, it has not yet been established how AI-based image analysis systems can help physicians in clinical practice.

METHODS

This review included peer-reviewed research articles on AI in the thorax published in English between 2015 and 2021.

CONCLUSIONS

With advances in technology and appropriate preparation of physicians, AI could address various clinical problems that have not been solved due to a lack of clinical resources or technological limitations.

KEYWORDS

Artificial intelligence (AI); deep learning (DL); computer aided diagnosis (CAD); thoracic radiology; pulmonary medicine.

IMAL-Net: Interpretable multi-task attention learning network for invasive lung adenocarcinoma screening in CT images

PURPOSE

Feature maps created from deep convolutional neural networks (DCNNs) have been widely used for visual explanation of DCNN-based classification tasks. However, many clinical applications such as benign-malignant classification of lung nodules normally require quantitative and objective interpretability, rather than just visualization. In this paper, we propose a novel interpretable multi-task attention learning network named IMAL-Net for early invasive adenocarcinoma screening in chest computed tomography images, which takes advantage of segmentation prior to assist interpretable classification.

METHODS

Two sub-ResNets are firstly integrated together via a prior-attention mechanism for simultaneous nodule segmentation and invasiveness classification. Then, numerous radiomic features from the segmentation results are concatenated with high-level semantic features from the classification subnetwork by FC layers to achieve superior performance. Meanwhile, an end-to-end feature selection mechanism (named FSM) is designed to quantify crucial radiomic features greatly affecting the prediction of each sample, and thus it can provide clinically applicable interpretability to the prediction result.

RESULTS

Nodule samples from a total of 1626 patients were collected from two grade-A hospitals for large-scale verification. Five-fold cross validation demonstrated that the proposed IMAL-Net can achieve an AUC score of 93.8% ± 1.1% and a recall score of 93.8% ± 2.8% for identification of invasive lung adenocarcinoma.

CONCLUSIONS

It can be concluded that fusing semantic features and radiomic features can achieve obvious improvements in the invasiveness classification task. Moreover, by learning more fine-grained semantic features and highlighting the most important radiomics features, the proposed attention and FSM mechanisms not only can further improve the performance but also can be used for both visual explanations and objective analysis of the classification results.

Can peritumoral regions increase the efficiency of machine-learning prediction of pathological invasiveness in lung adenocarcinoma manifesting as ground-glass nodules?

BACKGROUND

The peri-tumor microenvironment plays an important role in the occurrence, growth and metastasis of cancer. The aim of this study is to explore the value and application of a CT image-based deep learning model of tumors and peri-tumors in predicting the invasiveness of ground-glass nodules (GGNs).

METHODS

Preoperative thin-section chest CT images were reviewed retrospectively in 622 patients with a total of 687 pulmonary GGNs. GGNs are classified according to clinical management strategies as invasive lesions (IAC) and non-invasive lesions (AAH, AIS and MIA). The two volumes of interest (VOIs) identified on CT were the gross tumor volume (GTV) and the gross volume of tumor incorporating peritumoral region (GPTV). Three dimensional (3D) DenseNet was used to model and predict GGN invasiveness, and five-fold cross validation was performed. We used GTV and GPTV as inputs for the comparison model. Prediction performance was evaluated by sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).

RESULTS

The GTV-based model was able to successfully predict GGN invasiveness, with an AUC of 0.921 (95% CI, 0.896-0.937). Using GPTV, the AUC of the model increased to 0.955 (95% CI, 0.939-0.971).

CONCLUSIONS

The deep learning method performed well in predicting GGN invasiveness. The predictive ability of the GPTV-based model was more effective than that of the GTV-based model.

Deep learning-based differentiation of invasive adenocarcinomas from preinvasive or minimally invasive lesions among pulmonary subsolid nodules

OBJECTIVES

To evaluate a deep learning-based model using model-generated segmentation masks to differentiate invasive pulmonary adenocarcinoma (IPA) from preinvasive lesions or minimally invasive adenocarcinoma (MIA) on CT, making comparisons with radiologist-derived measurements of solid portion size.

METHODS

Four hundred eleven subsolid nodules (SSNs) (120 preinvasive lesions or MIAs and 291 IPAs) in 333 patients who underwent surgery between June 2010 and August 2016 were retrospectively included to develop the model (370 SSNs in 293 patients for training and 41 SSNs in 40 patients for tuning). Ninety SSNs of 2 cm or smaller (45 preinvasive lesions or MIAs and 45 IPAs) resected in 2018 formed a validation set. Six radiologists measured the solid portion of each nodule. Performances of the model and radiologists were assessed using receiver operating characteristics curve analysis.

RESULTS

The deep learning model differentiated IPA from preinvasive lesions or MIA with areas under the curve (AUCs) of 0.914, 0.956, and 0.833 for the training, tuning, and validation sets, respectively. The mean AUC of the radiologists was 0.835 in the validation set, without significant differences between radiologists and the model (p = 0.97). The sensitivity, specificity, and accuracy of the model were 71% (32/45), 87% (39/45), and 79% (71/90), respectively, whereas the corresponding values of the radiologists were 75.2% (203/270), 76.7% (207/270), and 75.9% (410/540) with a 5-mm threshold for the solid portion size.

CONCLUSIONS

The performance of the model for differentiating IPA from preinvasive lesions or MIA was comparable to that of the radiologists' measurements of solid portion size.

KEY POINTS

• A deep learning-based model differentiated IPA from preinvasive lesions or MIA with AUCs of 0.914 and 0.956 for the training and tuning sets, respectively. • In the validation set including subsolid nodules of 2 cm or smaller, the model showed an AUC of 0.833, being on par with the performance of the solid portion size measurements made by the radiologists (AUC, 0.835; p = 0.97). • SSNs with a solid portion measuring > 10 mm on CT showed a high probability of being IPA (positive predictive value, 93.5-100.0%).