Use of CT radiomics to differentiate minimally invasive adenocarcinomas and invasive adenocarcinomas presenting as pure ground-glass nodules larger than 10 mm

A nomogram based on the quantitative and qualitative features of CT imaging for the prediction of the invasiveness of ground glass nodules in lung adenocarcinoma

PURPOSE

Based on the quantitative and qualitative features of CT imaging, a model for predicting the invasiveness of ground-glass nodules (GGNs) was constructed, which could provide a reference value for preoperative planning of GGN patients.

MATERIALS AND METHODS

Altogether, 702 patients with GGNs (including 748 GGNs) were included in this study. The GGNs operated between September 2020 and July 2022 were classified into the training group (n = 555), and those operated between August 2022 and November 2022 were classified into the validation group (n = 193). Clinical data and the quantitative and qualitative features of CT imaging were harvested from these patients. In the training group, the quantitative and qualitative characteristics in CT imaging of GGNs were analyzed by using performing univariate and multivariate logistic regression analyses, followed by constructing a nomogram prediction model. The differentiation, calibration, and clinical practicability in both the training and validation groups were assessed by the nomogram models.

RESULTS

In the training group, multivariate logistic regression analysis disclosed that the maximum diameter (OR = 4.707, 95%CI: 2.06-10.758), consolidation/tumor ratio (CTR) (OR = 1.027, 95%CI: 1.011-1.043), maximum CT value (OR = 1.025, 95%CI: 1.004-1.047), mean CT value (OR = 1.035, 95%CI: 1.008-1.063; P = 0.012), spiculation sign (OR = 2.055, 95%CI: 1.148-3.679), and vascular convergence sign (OR = 2.508, 95%CI: 1.345-4.676) were independent risk parameters for invasive adenocarcinoma. Based on these findings, we established a nomogram model for predicting the invasiveness of GGN, and the AUC was 0.910 (95%CI: 0.885-0.934) and 0.902 (95%CI: 0.859-0.944) in the training group and the validation group, respectively. The internal validation of the Bootstrap method showed an AUC value of 0.905, indicating a good differentiation of the model. Hosmer-Lemeshow goodness of fit test for the training and validation groups indicated that the model had a good fitting effect (P > 0.05). Furthermore, the calibration curve and decision analysis curve of the training and validation groups reflected that the model had a good calibration degree and clinical practicability.

CONCLUSION

Combined with the quantitative and qualitative features of CT imaging, a nomogram prediction model can be created to forecast the invasiveness of GGNs. This model has good prediction efficacy for the invasiveness of GGNs and can provide help for the clinical management and decision-making of GGNs.

Invasiveness identification in pure ground-glass nodules: exploring the generalizability of radiomics based on external validation and stress testing

PURPOSE

This study aimed to apply external validation and stress tests to evaluate the generalizability of radiomics models built using various machine-learning methods for identifying the invasiveness of lung adenocarcinomas manifesting as pure ground-glass nodules (pGGNs).

METHODS

This retrospective study enrolled 495 patients (514 pGGNs) confirmed as lung adenocarcinomas by postoperative pathology from three centers. All nodules were included in the primary cohort (randomly divided into training and test cohorts), two external validation cohorts, and two stress test cohorts. Six machine-learning radiomics models were constructed in the training cohort using the optimal features. Performance of radiomics models and clinical models were compared in primary cohort and external validation cohorts. The stress tests included stratified performance evaluation and shifted performance evaluation and contrastive evaluation under three single-condition modification settings. The predictive performance was validated by area under curve (AUC) of receiver operating characteristic (ROC).

RESULTS

Of the six radiomics models, the best logistic regression (LR) model was able to maintain high differential diagnostic capability (AUC: 0.849 ± 0.049) and good stability (relative standard deviation, 5.719%), but it showed poorer performance (AUC = 0.835) than the clinical model (AUC = 0.862) in the external validation cohort E1. The stress tests suggested LR model had no significant difference in performance between subgroups after stratification and had good consistency in the predictions before and after the three transformations (Kappa = 0.960, 0.840, and 0.933, respectively; p < 0.05, all).

CONCLUSION

The rigorous testing procedure facilitates the selection of high-performance radiomics models with good clinical generalizability.

Radiomics and Artificial Intelligence Can Predict Malignancy of Solitary Pulmonary Nodules in the Elderly

Solitary pulmonary nodules (SPNs) are a diagnostic and therapeutic challenge for thoracic surgeons. Although such lesions are usually benign, the risk of malignancy remains significant, particularly in elderly patients, who represent a large segment of the affected population. Surgical treatment in this subset, which usually presents several comorbidities, requires careful evaluation, especially when pre-operative biopsy is not feasible and comorbidities may jeopardize the outcome. Radiomics and artificial intelligence (AI) are progressively being applied in predicting malignancy in suspicious nodules and assisting the decision-making process. In this study, we analyzed features of the radiomic images of 71 patients with SPN aged more than 75 years (median 79, IQR 76-81) who had undergone upfront pulmonary resection based on CT and PET-CT findings. Three different machine learning algorithms were applied-functional tree, Rep Tree and J48. Histology was malignant in 64.8% of nodules and the best predictive value was achieved by the J48 model (AUC 0.9). The use of AI analysis of radiomic features may be applied to the decision-making process in elderly frail patients with suspicious SPNs to minimize the false positive rate and reduce the incidence of unnecessary surgery.

Feature selection methods and predictive models in CT lung cancer radiomics

Radiomics is a technique that extracts quantitative features from medical images using data-characterization algorithms. Radiomic features can be used to identify tissue characteristics and radiologic phenotyping that is not observable by clinicians. A typical workflow for a radiomics study includes cohort selection, radiomic feature extraction, feature and predictive model selection, and model training and validation. While there has been increasing attention given to radiomic feature extraction, standardization, and reproducibility, currently, there is a lack of rigorous evaluation of feature selection methods and predictive models. Herein, we review the published radiomics investigations in CT lung cancer and provide an overview of the commonly used radiomic feature selection methods and predictive models. We also compare limitations of various methods in clinical applications and present sources of uncertainty associated with those methods. This review is expected to help raise awareness of the impact of radiomic feature and model selection methods on the integrity of radiomics studies.

Pathological components and CT imaging analysis of the area adjacent pleura within the pure ground-glass nodules with pleural deformation in invasive lung adenocarcinoma

Background

Pleural deformation is associated with the invasiveness of lung adenocarcinoma(LAC). Our study focused on the pathological components of the area adjacent pleura in pulmonary pure ground-glass nodules(pGGNs) with pleural deformations(P-pGGNs) confirmed to be invasive LAC without visceral pleural invasion (VPI) pathologically.

Methods

Computed tomography(CT) imaging features of nodules and pathological components of the area adjacent pleura were analyzed and recorded. Statistical analysis was performed for subgroups of P-pGGNs.

Results

The 81 enrolled patients with 81 P-pGGNs were finally involved in the analysis. None of solid/micropapillary group and none of VPI was observed, 54 alveoli/lepidics and 27 acinar/papillarys were observed. In P-pGGN with acinar/papillary components of the area adjacent pleura, invasive adenocarcinoma (IAC) was more common compared to minimally invasive adenocarcinoma (MIA, 74.07% vs. 25.93%; p < 0.001). The distance in alveoli/lepidic group was significantly larger (1.50 mm vs. 0.00 mm; p < 0.001) and the depth was significantly smaller (2.00 mm vs. 6.00 mm; p < 0.001) than that in acinar/papillary group. The CT attenuation value, maximum diameter and maximum vertical diameter was valuable to distinguish acinar/papillary group form alveoli/lepidic group(p < 0.05). The type d pleural deformation was the common pleural deformation in IAC(p = 0.028).

Conclusions

The pathological components of the area adjacent pleura in P-pGGN without VPI confirmed to be invasive LAC could included alveoli/lepidics and acinar/papillarys. Some CT indicators that can identify the pathological invasive components of the area adjacent pleura in P-pGGNs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-022-10043-2.

Prediction of invasive adenocarcinomas manifesting as pure ground-glass nodules based on radiomic signature of low-dose CT in lung cancer screening

OBJECTIVE

To develop a radiomic model based on low-dose CT (LDCT) to distinguish invasive adenocarcinomas (IAs) from adenocarcinoma in situ/minimally invasive adenocarcinomas (AIS/MIAs) manifesting as pure ground-glass nodules (pGGNs) and compare its performance with conventional quantitative and semantic features of LDCT, radiomic model of standard-dose CT, and intraoperative frozen section (FS).

METHODS

A total of 147 consecutive pathologically confirmed pGGNs were divided into primary cohort (43 IAs and 60 AIS/MIAs) and validation cohort (19 IAs and 25 AIS/MIAs). Logistic regression models were built using conventional quantitative and semantic features, selected radiomic features of LDCT and standard-dose CT, and intraoperative FS diagnosis, respectively. The diagnostic performance was assessed by area under curve (AUC) of receiver operating characteristic curve, sensitivity, and specificity.

RESULTS

The AUCs of quantitative-semantic model, radiomic model of LDCT, radiomic model of standard-dose CT, and FS model were 0.879 (95% CI, 0.801-0.935), 0.929 (95% CI, 0.862-0.971), 0.941 (95% CI, 0.876-0.978), and 0.884 (95% CI, 0.805-0.938) in the primary cohort and 0.897 (95% CI, 0.768-0.968), 0.933 (95% CI, 0.815-0.986), 0.901 (95% CI, 0.773-0.970), and 0.828 (95% CI, 0.685-0.925) in the validation cohort. No significant difference of the AUCs was found among these models in both the primary and validation cohorts (all p > 0.05).

CONCLUSION

The LDCT-based quantitative-semantic score and radiomic signature, with good predictive performance, can be pre-operative and non-invasive biomarkers for assessing the invasive risk of pGGNs in lung cancer screening.

ADVANCES IN KNOWLEDGE

The LDCT-based quantitative-semantic score and radiomic signature, with the equivalent performance to the radiomic model of standard-dose CT, can be pre-operative predictors for assessing the invasiveness of pGGNs in lung cancer screening and reducing excess examination and treatment.

Radiomics in Early Lung Cancer Diagnosis: From Diagnosis to Clinical Decision Support and Education

Lung cancer is the most frequent cause of cancer-related death around the world. With the recent introduction of low-dose lung computed tomography for lung cancer screening, there has been an increasing number of smoking- and non-smoking-related lung cancer cases worldwide that are manifesting with subsolid nodules, especially in Asian populations. However, the pros and cons of lung cancer screening also follow the implementation of lung cancer screening programs. Here, we review the literature related to radiomics for early lung cancer diagnosis. There are four main radiomics applications: the classification of lung nodules as being malignant/benign; determining the degree of invasiveness of the lung adenocarcinoma; histopathologic subtyping; and prognostication in lung cancer prediction models. In conclusion, radiomics offers great potential to improve diagnosis and personalized risk stratification in early lung cancer diagnosis through patient–doctor cooperation and shared decision making.

Computed tomography radiomics-based distinction of invasive adenocarcinoma from minimally invasive adenocarcinoma manifesting as pure ground-glass nodules with bubble-like signs

BACKGROUND

To explore an effective model based on radiomics features extracted from nonenhanced computed tomography (CT) images to distinguish invasive adenocarcinoma (IAC) from minimally invasive adenocarcinoma (MIA) presenting as pure ground-glass nodules (pGGNs) with bubble-like (B-pGGNs) signs.

PATIENTS AND METHODS

We retrospectively reviewed 511 nodules (MIA, n = 288; IAC, n = 223) between November 2012 and June 2018 from almost all pGGNs pathologically confirmed MIA or IAC. Eventually, a total of 109 B-pGGNs (MIA, n = 55; IAC, n = 54) from 109 patients fulfilling the criteria were randomly assigned to the training and test cluster at a ratio of 7:3. The gradient boosting decision tree (GBDT) method and logistic regression (LR) analysis were applied to feature selection (radiomics, semantic, and conventional CT features). LR was performed to construct three models (the conventional, radiomics and combined model). The performance of the predictive models was evaluated using the area under the curve (AUC).

RESULTS

The radiomics model had good AUCs of 0.947 in the training cluster and of 0.945 in the test cluster. The combined model produced an AUC of 0.953 in the training cluster and of 0.945 in the test cluster. The combined model yielded no performance improvement (vs. the radiomics model). The rad_score was the only independent predictor of invasiveness.

CONCLUSION

The radiomics model showed excellent predictive performance in discriminating IAC from MIA presenting as B-pGGNs and may provide a necessary reference for extending clinical practice.

A Comparative Study of Radiomics and Deep-Learning Based Methods for Pulmonary Nodule Malignancy Prediction in Low Dose CT Images

OBJECTIVES

Both radiomics and deep learning methods have shown great promise in predicting lesion malignancy in various image-based oncology studies. However, it is still unclear which method to choose for a specific clinical problem given the access to the same amount of training data. In this study, we try to compare the performance of a series of carefully selected conventional radiomics methods, end-to-end deep learning models, and deep-feature based radiomics pipelines for pulmonary nodule malignancy prediction on an open database that consists of 1297 manually delineated lung nodules.

METHODS

Conventional radiomics analysis was conducted by extracting standard handcrafted features from target nodule images. Several end-to-end deep classifier networks, including VGG, ResNet, DenseNet, and EfficientNet were employed to identify lung nodule malignancy as well. In addition to the baseline implementations, we also investigated the importance of feature selection and class balancing, as well as separating the features learned in the nodule target region and the background/context region. By pooling the radiomics and deep features together in a hybrid feature set, we investigated the compatibility of these two sets with respect to malignancy prediction.

RESULTS

The best baseline conventional radiomics model, deep learning model, and deep-feature based radiomics model achieved AUROC values (mean ± standard deviations) of 0.792 ± 0.025, 0.801 ± 0.018, and 0.817 ± 0.032, respectively through 5-fold cross-validation analyses. However, after trying out several optimization techniques, such as feature selection and data balancing, as well as adding context features, the corresponding best radiomics, end-to-end deep learning, and deep-feature based models achieved AUROC values of 0.921 ± 0.010, 0.824 ± 0.021, and 0.936 ± 0.011, respectively. We achieved the best prediction accuracy from the hybrid feature set (AUROC: 0.938 ± 0.010).

CONCLUSION

The end-to-end deep-learning model outperforms conventional radiomics out of the box without much fine-tuning. On the other hand, fine-tuning the models lead to significant improvements in the prediction performance where the conventional and deep-feature based radiomics models achieved comparable results. The hybrid radiomics method seems to be the most promising model for lung nodule malignancy prediction in this comparative study.

Development and Validation of a Radiomics Nomogram for Differentiating Pulmonary Cryptococcosis and Lung Adenocarcinoma in Solitary Pulmonary Solid Nodule

Objective

To establish a CT-based radiomics nomogram model for classifying pulmonary cryptococcosis (PC) and lung adenocarcinoma (LAC) in patients with a solitary pulmonary solid nodule (SPSN) and assess its differentiation ability.

Materials and Methods

A total of 213 patients with PC and 213 cases of LAC (matched based on age and gender) were recruited into this retrospective research with their clinical characteristics and radiological features. High-dimensional radiomics features were acquired from each mask delineated by radiologists manually. We adopted the max-relevance and min-redundancy (mRMR) approach to filter the redundant features and retained the relevant features at first. Then, we used the least absolute shrinkage and operator (LASSO) algorithms as an analysis tool to calculate the coefficients of features and remove the low-weight features. After multivariable logistic regression analysis, a radiomics nomogram model was constructed with clinical characteristics, radiological signs, and radiomics score. We calculated the performance assessment parameters, such as sensitivity, specificity, accuracy, negative predictive value (NPV), and positive predictive value (PPV), in various models. The receiver operating characteristic (ROC) curve analysis and the decision curve analysis (DCA) were drawn to visualize the diagnostic ability and the clinical benefit.

Results

We extracted 1,130 radiomics features from each CT image. The 24 most significant radiomics features in distinguishing PC and LAC were retained, and the radiomics signature was constructed through a three-step feature selection process. Three factors-maximum diameter, lobulation, and pleural retraction-were still statistically significant in multivariate analysis and incorporated into a combined model with radiomics signature to develop the predictive nomogram, which showed excellent classification ability. The area under curve (AUC) yielded 0.91 (sensitivity, 80%; specificity, 83%; accuracy, 82%; NPV, 80%; PPV, 83%) and 0.89 (sensitivity, 81%; specificity, 83%; accuracy, 82%; NPV, 81%; PPV, 82%) in training and test cohorts, respectively. The net reclassification indexes (NRIs) were greater than zero (p < 0.05). The Delong test showed a significant difference (p < 0.0001) between the AUCs from the clinical model and the nomogram.

Conclusions

The radiomics technology can preoperatively differentiate PC and lung adenocarcinoma. The nomogram-integrated CT findings and radiomics feature can provide more clinical benefits in solitary pulmonary solid nodule diagnosis.

Consecutive Serial Non-Contrast CT Scan-Based Deep Learning Model Facilitates the Prediction of Tumor Invasiveness of Ground-Glass Nodules

Introduction

Tumors are continuously evolving biological systems which can be monitored by medical imaging. Previous studies only focus on single timepoint images, whether the performance could be further improved by using serial noncontrast CT imaging obtained during nodule follow-up management remains unclear. In this study, we evaluated DL model for predicting tumor invasiveness of GGNs through analyzing time series CT images.

Methods

A total of 168 pathologically confirmed GGN cases (48 noninvasive lesions and 120 invasive lesions) were retrospectively collected and randomly assigned to the development dataset (n = 123) and independent testing dataset (n = 45). All patients underwent consecutive noncontrast CT examinations, and the baseline CT and 3-month follow-up CT images were collected. The gross region of interest (ROI) patches containing only tumor region and the full ROI patches including both tumor and peritumor regions were cropped from CT images. A baseline model was built on the image features and demographic features. Four DL models were proposed: two single-DL model using gross ROI (model 1) or full ROI patches (model 3) from baseline CT images, and two serial-DL models using gross ROI (model 2) or full ROI patches (model 4) from consecutive CT images (baseline scan and 3-month follow-up scan). In addition, a combined model integrating serial full ROI patches and clinical information was also constructed. The performance of these predictive models was assessed with respect to discrimination and clinical usefulness.

Results

The area under the curve (AUC) of the baseline model, models 1, 2, 3, and 4 were 0.562 [(95% confidence interval (C)], 0.406~0.710), 0.693 (95% CI, 0.538-0.822), 0.787 (95% CI, 0.639-0.895), 0.727 (95% CI, 0.573-0.849), and 0.811 (95% CI, 0.667-0.912) in the independent testing dataset, respectively. The results indicated that the peritumor region had potential to contribute to tumor invasiveness prediction, and the model performance was further improved by integrating imaging scans at multiple timepoints. Furthermore, the combined model showed best discrimination ability, with AUC, sensitivity, specificity, and accuracy achieving 0.831 (95% CI, 0.690-0.926), 86.7%, 73.3%, and 82.2%, respectively.

Conclusion

The DL model integrating full ROIs from serial CT images shows improved predictive performance in differentiating noninvasive from invasive GGNs than the model using only baseline CT images, which could benefit the clinical management of GGNs.

Structured Reporting of Lung Cancer Staging: A Consensus Proposal

Background: Structured reporting (SR) in radiology is becoming necessary and has recently been recognized by major scientific societies. This study aimed to build CT-based structured reports for lung cancer during the staging phase, in order to improve communication between radiologists, members of the multidisciplinary team and patients. Materials and Methods: A panel of expert radiologists, members of the Italian Society of Medical and Interventional Radiology, was established. A modified Delphi exercise was used to build the structural report and to assess the level of agreement for all the report sections. The Cronbach’s alpha (Cα) correlation coefficient was used to assess internal consistency for each section and to perform a quality analysis according to the average inter-item correlation. Results: The final SR version was built by including 16 items in the “Patient Clinical Data” section, 4 items in the “Clinical Evaluation” section, 8 items in the “Exam Technique” section, 22 items in the “Report” section, and 5 items in the “Conclusion” section. Overall, 55 items were included in the final version of the SR. The overall mean of the scores of the experts and the sum of scores for the structured report were 4.5 (range 1–5) and 631 (mean value 67.54, STD 7.53), respectively, in the first round. The items of the structured report with higher accordance in the first round were primary lesion features, lymph nodes, metastasis and conclusions. The overall mean of the scores of the experts and the sum of scores for staging in the structured report were 4.7 (range 4–5) and 807 (mean value 70.11, STD 4.81), respectively, in the second round. The Cronbach’s alpha (Cα) correlation coefficient was 0.89 in the first round and 0.92 in the second round for staging in the structured report. Conclusions: The wide implementation of SR is critical for providing referring physicians and patients with the best quality of service, and for providing researchers with the best quality of data in the context of the big data exploitation of the available clinical data. Implementation is complex, requiring mature technology to successfully address pending user-friendliness, organizational and interoperability challenges.