Non-invasive classification of non-small cell lung cancer: a comparison between random forest models utilising radiomic and semantic features

Machine learning for differentiating lung squamous cell cancer from adenocarcinoma using Clinical-Metabolic characteristics and 18F-FDG PET/CT radiomics

Noninvasive differentiation between the squamous cell carcinoma (SCC) and adenocarcinoma (ADC) subtypes of non-small cell lung cancer (NSCLC) could benefit patients who are unsuitable for invasive diagnostic procedures. Therefore, this study evaluates the predictive performance of a PET/CT-based radiomics model. It aims to distinguish between the histological subtypes of lung adenocarcinoma and squamous cell carcinoma, employing four different machine learning techniques. A total of 255 Non-Small Cell Lung Cancer (NSCLC) patients were retrospectively analyzed and randomly divided into the training (n = 177) and validation (n = 78) sets, respectively. Radiomics features were extracted, and the Least Absolute Shrinkage and Selection Operator (LASSO) method was employed for feature selection. Subsequently, models were constructed using four distinct machine learning techniques, with the top-performing algorithm determined by evaluating metrics such as accuracy, sensitivity, specificity, and the area under the curve (AUC). The efficacy of the various models was appraised and compared using the DeLong test. A nomogram was developed based on the model with the best predictive efficiency and clinical utility, and it was validated using calibration curves. Results indicated that the logistic regression classifier had better predictive power in the validation cohort of the radiomic model. The combined model (AUC 0.870) exhibited superior predictive power compared to the clinical model (AUC 0.848) and the radiomics model (AUC 0.774). In this study, we discovered that the combined model, refined by the logistic regression classifier, exhibited the most effective performance in classifying the histological subtypes of NSCLC.

A radiomics strategy based on CT intra-tumoral and peritumoral regions for preoperative prediction of neoadjuvant chemoradiotherapy for esophageal cancer

OBJECTIVE

Develop a method for selecting esophageal cancer patients achieving pathological complete response with pre-neoadjuvant therapy chest-enhanced CT scans.

METHODS

Two hundred and one patients from center 1 were enrolled, split into training and testing sets (7:3 ratio), with an external validation set of 30 patients from center 2. Radiomics features from intra-tumoral and peritumoral images were extracted and dimensionally reduced using Student's t-test and least absolute shrinkage and selection operator. Four machine learning classifiers were employed to build models, with the best-performing models selected based on accuracy and stability. ROC curves were utilized to determine the top prediction model, and its generalizability was evaluated on the external validation set.

RESULTS

Among 16 models, the integrated-XGBoost and integrated-random forest models performed the best, with average ROC AUCs of 0.906 and 0.918, respectively, and RSDs of 6.26 and 6.89 in the training set. In the testing set, AUCs were 0.845 and 0.871, showing no significant difference in ROC curves. External validation set AUCs for integrated-XGBoost and integrated-random forest models were 0.650 and 0.749.

CONCLUSION

Incorporating peritumoral radiomics features into the analysis enhances predictive performance for esophageal cancer patients undergoing neoadjuvant chemoradiotherapy, paving the way for improved treatment outcomes.

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.

Evaluating Histological Subtypes Classification of Primary Lung Cancers on Unenhanced Computed Tomography Based on Random Forest Model

Lung cancer is the leading cause of cancer-related death in many countries, and an accurate histopathological diagnosis is of great importance in subsequent treatment. The aim of this study was to establish the random forest (RF) model based on radiomic features to automatically classify and predict lung adenocarcinoma (ADC), lung squamous cell carcinoma (SCC), and small cell lung cancer (SCLC) on unenhanced computed tomography (CT) images. Eight hundred and fifty-two patients (mean age: 61.4, range: 29-87, male/female: 536/316) with preoperative unenhanced CT and postoperative histopathologically confirmed primary lung cancers, including 525 patients with ADC, 161 patients with SCC, and 166 patients with SCLC, were included in this retrospective study. Radiomic features were extracted, selected, and then used to establish the RF classification model to analyse and classify primary lung cancers into three subtypes, including ADC, SCC, and SCLC according to histopathological results. The training (446 ADC, 137 SCC, and 141 SCLC) and testing cohorts (79 ADC, 24 SCC, and 25 SCLC) accounted for 85% and 15% of the whole datasets, respectively. The prediction performance of the RF classification model was evaluated by F1 scores and the receiver operating characteristic (ROC) curve. On the testing cohort, the areas under the ROC curve (AUC) of the RF model in classifying ADC, SCC, and SCLC were 0.74, 0.77, and 0.88, respectively. The F1 scores achieved 0.80, 0.40, and 0.73 in ADC, SCC, and SCLC, respectively, and the weighted average F1 score was 0.71. In addition, for the RF classification model, the precisions were 0.72, 0.64, and 0.70; the recalls were 0.86, 0.29, and 0.76; and the specificities were 0.55, 0.96, and 0.92 in ADC, SCC, and SCLC. The primary lung cancers were feasibly and effectively classified into ADC, SCC, and SCLC based on the combination of RF classification model and radiomic features, which has the potential for noninvasive predicting histological subtypes of primary lung cancers.

Potential impact of texture analysis in contrast enhanced CT in non-small cell lung cancer as a marker of survival: A retrospective feasibility study

The objective of this feasibility study was to assess computed tomography (CT) texture analysis (CTTA) of pulmonary lesions as a predictor of overall survival in patients with suspected lung cancer on contrast-enhanced computed tomography (CECT). In a retrospective pilot study, 94 patients (52 men and 42 women; mean age, 67.2 ± 10.8 yrs) from 1 center with non-small cell lung cancer (NSCLC) underwent CTTA on the primary lesion by 2 individual readers. Both simple and multivariate Cox regression analyses correlating textural parameters with overall survival were performed. Statistically significant parameters were selected, and optimal cutoff values were determined. Kaplan-Meier plots based on these results were produced. Simple Cox regression analysis showed that normalized uniformity had a hazard ratio (HR) of 16.059 (3.861-66.788, P < .001), and skewness had an HR of 1.914 (1.330-2.754, P < .001). The optimal cutoff values for both parameters were 0.8602 and 0.1554, respectively. Normalized uniformity, clinical stage, and skewness were found to be prognostic factors for overall survival in multivariate analysis. Tumor heterogeneity, assessed by normalized uniformity and skewness on CECT may be a prognostic factor for overall survival.

Construction and Validation of Early Warning Model of Lung Cancer Based on Machine Learning: A Retrospective Study

Background: This study is a retrospective study. The purpose of this study is to construct and validate an early warning model of lung cancer through machine learning. Methods: The CDKN2A gene expression profile and clinical information were downloaded from The Cancer Genome Atlas (TCGA) database and divided into a tumor group and a normal group (n = 57). The top 5 somatic mutation-related genes were extracted from 567 somatic mutation data downloaded from TCGA database using random forest algorithm. Cox proportional hazard model and nomogram were constructed combining CDKN2A, 5 somatic mutation-related genes, gender, and smoking index. Patients were divided into high-risk and low-risk groups according to risk score. The predictability of the model in the prognosis of lung cancer was estimated by Kaplan-Meier survival analysis and receiver operating characteristics curve. Results: We constructed a prognostic model consisting of 5 somatic mutation-related genes (sphingosine 1-phosphate receptor 1 [S1PR1], dedicator of cytokinesis 7 [DOCK7], DEAD-box helicase 4 [DDX4], laminin subunit beta 3 [LAMB3], and importin 5 [IPO5]), cyclin-dependent kinase inhibitor 2A (CDKN2A), gender, and smoking indicators. The high-risk group had a lower overall survival rate compared to the low-risk group (hazard ratio = 2.14, P = 0 .0323). The area under the curve predicted for 3-year, 5-year, and 10-year survival rates are 0.609, 0.673, and 0.698, respectively. The accuracy, sensitivity, and specificity of the model for predicting the 10-year survival rate of lung cancer are 76.19%, 56.71%, and 86.23%. Conclusion: The lung cancer early warning model and nomogram may provide an essential reference for patients with lung cancer management in the clinic.

18F-FDG PET/CT radiomics nomogram for predicting occult lymph node metastasis of non-small cell lung cancer

Purpose

To investigate the ability of a PET/CT-based radiomics nomogram to predict occult lymph node metastasis in patients with clinical stage N0 non-small cell lung cancer (NSCLC).

Materials and methods

This retrospective study included 228 patients with surgically confirmed NSCLC (training set, 159 patients; testing set, 69 patients). ITKsnap3.8.0 was used for image(CT and PET images) segmentation, AK version 3.2.0 was used for radiomics feature extraction, and Python3.7.0 was used for radiomics feature screening. A radiomics model for predicting occult lymph node metastasis was established using a logistic regression algorithm. A nomogram was constructed by combining radiomics scores with selected clinical predictors. Receiver operating characteristic (ROC) curves were used to verify the performance of the radiomics model and nomogram in the training and testing sets.

Results

The radiomics nomogram comprising six selected features achieved good prediction efficiency, including radiomics characteristics and tumor location information (central or peripheral), which demonstrated good calibration and discrimination ability in the training (area under the ROC curve [AUC] = 0.884, 95% confidence interval [CI]: 0.826-0.941) and testing (AUC = 0.881, 95% CI: 0.8031-0.959) sets. Clinical decision curves demonstrated that the nomogram was clinically useful.

Conclusion

The PET/CT-based radiomics nomogram is a noninvasive tool for predicting occult lymph node metastasis in NSCLC.

Value of CT Radiomics and Clinical Features in Predicting Bone Metastases in Patients with NSCLC

Objective

To explore the CT radiomic features and clinical imaging features of the primary tumor in patients with nonsmall cell lung cancer (NSCLC) before treatment and their predictive value for the occurrence of bone metastases.

Methods

From June 2017 to June 2021, 195 patients with NSCLC who were pathologically diagnosed without any treatment in the Cancer Hospital Affiliated to Hainan Medical College were retrospectively analyzed, and they were divided into a bone metastasis group and a nonbone metastasis group. The relationship between clinical imaging features and bone metastasis in patients was analyzed by the t-test, rank sum test, and χ 2 test. At the same time, ITK software was used to extract the radiomic characteristics of the primary tumor of the patients, and the patients were randomly divided into a training group and a validation group in a ratio of 7 : 3. The training model was validated in the validation group, and the performance of the model for predicting bone metastases in NSCLC patients was verified by the ROC curve, and a multivariate logistic regression prediction model was established based on the omics parameters extracted from the best prediction model combined with clinical image features.

Results

Seven features were screened from the primary tumor by LASSO to establish a model for predicting metastasis. The area under the curve was 0.82 and 0.73 in the training and validation sets. The best omics signature and univariate analysis suggested clinical imaging factors (P < 0.05) associated with bone metastases were included in multivariate binary logistic analysis to obtain clinical characteristics of the primary tumor such as gender (OR = 0.141, 95% CI: 0.022-0.919, P = 0.04), increased Cyfra21-1 (OR = 0.12, 95% CI: 0.018-0.782, P = 0.027), Fe content in blood (OR = 0.774, 95% CI: 0.626-0.958, P = 0.018), CT signs such as lesion homogeneity (OR = 0.052, 95% CI: 0.006-0.419, P = 0.006), pleural indentation sign (OR = 0.007, 95% CI: 0.001-0.696, P = 0.034), and omics characteristics glszm_Small Area High Gray Level Emphasis (OR = 0.016, 95% CI: 0.001-0.286, P = 0.005) were independent risk factors for bone metastasis in patients.

Conclusion

The prediction model established based on radiomics and clinical imaging features has high predictive performance for the occurrence of bone metastasis in NSCLC patients.

Reconstruction-assisted Feature Encoding Network for Histologic Subtype Classification of Non-small Cell Lung Cancer

Accurate histological subtype classification between adenocarcinoma (ADC) and squamous cell carcinoma (SCC) using computed tomography (CT) images is of great importance to assist clinicians in determining treatment and therapy plans for non-small cell lung cancer (NSCLC) patients. Although current deep learning approaches have achieved promising progress in this field, they are often difficult to capture efficient tumor representations due to inadequate training data, and in consequence show limited performance. In this study, we propose a novel and effective reconstruction-assisted feature encoding network (RAFENet) for histological subtype classification by leveraging an auxiliary image reconstruction task to enable extra guidance and regularization for enhanced tumor feature representations. Different from existing reconstruction-assisted methods that directly use generalizable features obtained from shared encoder for primary task, a dedicated task-aware encoding module is utilized in RAFENet to perform refinement of generalizable features. Specifically, a cascade of cross-level non-local blocks are introduced to progressively refine generalizable features at different levels with the aid of lower-level task-specific information, which can successfully learn multi-level task-specific features tailored to histological subtype classification. Moreover, in addition to widely adopted pixel-wise reconstruction loss, we introduce a powerful semantic consistency loss function to explicitly supervise the training of RAFENet, which combines both feature consistency loss and prediction consistency loss to ensure semantic invariance during image reconstruction. Extensive experimental results show that RAFENet effectively addresses the difficult issues that cannot be resolved by existing reconstruction-based methods and consistently outperforms other state-of-the-art methods on both public and in-house NSCLC datasets.

Diagnosis of Idiopathic Pulmonary Fibrosis in High-Resolution Computed Tomography Scans Using a Combination of Handcrafted Radiomics and Deep Learning

Purpose

To develop handcrafted radiomics (HCR) and deep learning (DL) based automated diagnostic tools that can differentiate between idiopathic pulmonary fibrosis (IPF) and non-IPF interstitial lung diseases (ILDs) in patients using high-resolution computed tomography (HRCT) scans.

Material and Methods

In this retrospective study, 474 HRCT scans were included (mean age, 64.10 years ± 9.57 [SD]). Five-fold cross-validation was performed on 365 HRCT scans. Furthermore, an external dataset comprising 109 patients was used as a test set. An HCR model, a DL model, and an ensemble of HCR and DL model were developed. A virtual in-silico trial was conducted with two radiologists and one pulmonologist on the same external test set for performance comparison. The performance was compared using DeLong method and McNemar test. Shapley Additive exPlanations (SHAP) plots and Grad-CAM heatmaps were used for the post-hoc interpretability of HCR and DL models, respectively.

Results

In five-fold cross-validation, the HCR model, DL model, and the ensemble of HCR and DL models achieved accuracies of 76.2 ± 6.8, 77.9 ± 4.6, and 85.2 ± 2.7%, respectively. For the diagnosis of IPF and non-IPF ILDs on the external test set, the HCR, DL, and the ensemble of HCR and DL models achieved accuracies of 76.1, 77.9, and 85.3%, respectively. The ensemble model outperformed the diagnostic performance of clinicians who achieved a mean accuracy of 66.3 ± 6.7% (p < 0.05) during the in-silico trial. The area under the receiver operating characteristic curve (AUC) for the ensemble model on the test set was 0.917 which was significantly higher than the HCR model (0.817, p = 0.02) and the DL model (0.823, p = 0.005). The agreement between HCR and DL models was 61.4%, and the accuracy and specificity for the predictions when both the models agree were 93 and 97%, respectively. SHAP analysis showed the texture features as the most important features for IPF diagnosis and Grad-CAM showed that the model focused on the clinically relevant part of the image.

Conclusion

Deep learning and HCR models can complement each other and serve as useful clinical aids for the diagnosis of IPF and non-IPF ILDs.

Towards a safe and efficient clinical implementation of machine learning in radiation oncology by exploring model interpretability, explainability and data-model dependency

The interest in machine learning (ML) has grown tremendously in recent years, partly due to the performance leap that occurred with new techniques of deep learning, convolutional neural networks for images, increased computational power, and wider availability of large datasets. Most fields of medicine follow that popular trend and, notably, radiation oncology is one of those that are at the forefront, with already a long tradition in using digital images and fully computerized workflows. ML models are driven by data, and in contrast with many statistical or physical models, they can be very large and complex, with countless generic parameters. This inevitably raises two questions, namely, the tight dependence between the models and the datasets that feed them, and the interpretability of the models, which scales with its complexity. Any problems in the data used to train the model will be later reflected in their performance. This, together with the low interpretability of ML models, makes their implementation into the clinical workflow particularly difficult. Building tools for risk assessment and quality assurance of ML models must involve then two main points: interpretability and data-model dependency. After a joint introduction of both radiation oncology and ML, this paper reviews the main risks and current solutions when applying the latter to workflows in the former. Risks associated with data and models, as well as their interaction, are detailed. Next, the core concepts of interpretability, explainability, and data-model dependency are formally defined and illustrated with examples. Afterwards, a broad discussion goes through key applications of ML in workflows of radiation oncology as well as vendors' perspectives for the clinical implementation of ML.

Radiomics: a quantitative imaging biomarker in precision oncology

Cancer treatment is heading towards precision medicine driven by genetic and biochemical markers. Various genetic and biochemical markers are utilized to render personalized treatment in cancer. In the last decade, noninvasive imaging biomarkers have also been developed to assist personalized decision support systems in oncology. The imaging biomarkers i.e., radiomics is being researched to develop specific digital phenotype of tumor in cancer. Radiomics is a process to extract high throughput data from medical images by using advanced mathematical and statistical algorithms. The radiomics process involves various steps i.e., image generation, segmentation of region of interest (e.g. a tumor), image preprocessing, radiomic feature extraction, feature analysis and selection and finally prediction model development. Radiomics process explores the heterogeneity, irregularity and size parameters of the tumor to calculate thousands of advanced features. Our study investigates the role of radiomics in precision oncology. Radiomics research has witnessed a rapid growth in the last decade with several studies published that show the potential of radiomics in diagnosis and treatment outcome prediction in oncology. Several radiomics based prediction models have been developed and reported in the literature to predict various prediction endpoints i.e., overall survival, progression-free survival and recurrence in various cancer i.e., brain tumor, head and neck cancer, lung cancer and several other cancer types. Radiomics based digital phenotypes have shown promising results in diagnosis and treatment outcome prediction in oncology. In the coming years, radiomics is going to play a significant role in precision oncology.

Artificial intelligence in thoracic surgery: a narrative review

Objective

The aim of this article is to review the current applications of artificial intelligence in thoracic surgery, from diagnosis and pulmonary disease management, to preoperative risk-assessment, surgical planning, and outcomes prediction.

Background

Artificial intelligence implementation in healthcare settings is rapidly growing, though its widespread use in clinical practice is still limited. The employment of machine learning algorithms in thoracic surgery is wide-ranging, including all steps of the clinical pathway.

Methods

We performed a narrative review of the literature on Scopus, PubMed and Cochrane databases, including all the relevant studies published in the last ten years, until March 2021.

Conclusion

Machine learning methods are promising encouraging results throughout the key issues of thoracic surgery, both clinical, organizational, and educational. Artificial intelligence-based technologies showed remarkable efficacy to improve the perioperative evaluation of the patient, to assist the decision-making process, to enhance the surgical performance, and to optimize the operating room scheduling. Still, some concern remains about data supply, protection, and transparency, thus further studies and specific consensus guidelines are needed to validate these technologies for daily common practice.

Keywords

Artificial intelligence (AI); thoracic surgery; machine learning; lung resection; perioperative medicine

Identification of Significantly Expressed Gene Mutations for Automated Classification of Benign and Malignant Prostate Cancer

Among males, prostate cancer (Pca) is the cancer type with the highest prevalence and the second leading cause of cancer deaths. The current screening methods for prostate cancer lack effectiveness such as prostate-specific antigen (PSA) and digital rectal exam (DRE). Machine learning models have been used to predict Pca progression, Gleason score, and laterality. In this research paper, we have employed novel Machine learning techniques such as Bayesian approach, Support vector machines (SVM), Decision Trees, Logistic Regression, K-Nearest Neighbors, Random Forest and AdaBoost for detecting malignant prostate cancers from benign ones. Moreover, different feature extracting strategies are proposed to improve the detection performance and identify potential genomic biomarkers. The results show the Lasso feature set yielded high performance from the models with SVM achieving exemplary classification accuracy of 97%. The Lasso and SVM combination reported many significant biomarker genes and gene mutations including but not restricted to CA2320112, CA2328529, and CA2436168.

Histological Subtypes Classification of Lung Cancers on CT Images Using 3D Deep Learning and Radiomics

RATIONALE AND OBJECTIVES

Histological subtypes of lung cancers are critical for clinical treatment decision. In this study, we attempt to use 3D deep learning and radiomics methods to automatically distinguish lung adenocarcinomas (ADC), squamous cell carcinomas (SCC), and small cell lung cancers (SCLC) respectively on Computed Tomography images, and then compare their performance.

MATERIALS AND METHODS

920 patients (mean age 61.2, range, 17-87; 340 Female and 580 Male) with lung cancer, including 554 patients with ADC, 175 patients with lung SCC and 191 patients with SCLC, were included in this retrospective study from January 2013 to August 2018. Histopathologic analysis was available for every patient. The classification models based on 3D deep learning (named the ProNet) and radiomics (named com_radNet) were designed to classify lung cancers into the three types mentioned above according to histopathologic results. The training, validation and testing cohorts counted 0.70, 0.15, and 0.15 of the whole datasets respectively.

RESULTS

The ProNet model used to classify the three types of lung cancers achieved the F1-scores of 90.0%, 72.4%, 83.7% in ADC, SCC, and SCLC respectively, and the weighted average F1-score of 73.2%. For com_radNet, the F1-scores achieved 83.1%, 75.4%, 85.1% in ADC, SCC, and SCLC, and the weighted average F1-score was 72.2%. The area under the receiver operating characteristic curve of the ProNet model and com_radNet were 0.840 and 0.789, and the accuracy were 71.6% and 74.7% respectively.

CONCLUSION

The ProNet and com_radNet models we developed can achieve high performance in distinguishing ADC, SCC, and SCLC and may be promising approaches for non-invasive predicting histological subtypes of lung cancers.

Non-small cell lung carcinoma histopathological subtype phenotyping using high-dimensional multinomial multiclass CT radiomics signature

OBJECTIVE

The aim of this study was to identify the most important features and assess their discriminative power in the classification of the subtypes of NSCLC.

METHODS

This study involved 354 pathologically proven NSCLC patients including 134 squamous cell carcinoma (SCC), 110 large cell carcinoma (LCC), 62 not other specified (NOS), and 48 adenocarcinoma (ADC). In total, 1433 radiomics features were extracted from 3D volumes of interest drawn on the malignant lesion identified on CT images. Wrapper algorithm and multivariate adaptive regression splines were implemented to identify the most relevant/discriminative features. A multivariable multinomial logistic regression was employed with 1000 bootstrapping samples based on the selected features to classify four main subtypes of NSCLC.

RESULTS

The results revealed that the texture features, specifically gray level size zone matrix features (GLSZM), were the significant indicators of NSCLC subtypes. The optimized classifier achieved an average precision, recall, F1-score, and accuracy of 0.710, 0.703, 0.706, and 0.865, respectively, based on the selected features by the wrapper algorithm.

CONCLUSIONS

Our CT radiomics approach demonstrated impressive potential for the classification of the four main histological subtypes of NSCLC, It is anticipated that CT radiomics could be useful in treatment planning and precision medicine.

Study Progress of Noninvasive Imaging and Radiomics for Decoding the Phenotypes and Recurrence Risk of Bladder Cancer

Urinary bladder cancer (BCa) is a highly prevalent disease among aged males. Precise diagnosis of tumor phenotypes and recurrence risk is of vital importance in the clinical management of BCa. Although imaging modalities such as CT and multiparametric MRI have played an essential role in the noninvasive diagnosis and prognosis of BCa, radiomics has also shown great potential in the precise diagnosis of BCa and preoperative prediction of the recurrence risk. Radiomics-empowered image interpretation can amplify the differences in tumor heterogeneity between different phenotypes, i.e., high-grade vs. low-grade, early-stage vs. advanced-stage, and nonmuscle-invasive vs. muscle-invasive. With a multimodal radiomics strategy, the recurrence risk of BCa can be preoperatively predicted, providing critical information for the clinical decision making. We thus reviewed the rapid progress in the field of medical imaging empowered by the radiomics for decoding the phenotype and recurrence risk of BCa during the past 20 years, summarizing the entire pipeline of the radiomics strategy for the definition of BCa phenotype and recurrence risk including region of interest definition, radiomics feature extraction, tumor phenotype prediction and recurrence risk stratification. We particularly focus on current pitfalls, challenges and opportunities to promote massive clinical applications of radiomics pipeline in the near future.

Introduction to Artificial Intelligence and Machine Learning for Pathology

CONTEXT.—

Recent developments in machine learning have stimulated intense interest in software that may augment or replace human experts. Machine learning may impact pathology practice by offering new capabilities in analysis, interpretation, and outcomes prediction using images and other data. The principles of operation and management of machine learning systems are unfamiliar to pathologists, who anticipate a need for additional education to be effective as expert users and managers of the new tools.

OBJECTIVE.—

To provide a background on machine learning for practicing pathologists, including an overview of algorithms, model development, and performance evaluation; to examine the current status of machine learning in pathology and consider possible roles and requirements for pathologists in local deployment and management of machine learning systems; and to highlight existing challenges and gaps in deployment methodology and regulation.

DATA SOURCES.—

Sources include the biomedical and engineering literature, white papers from professional organizations, government reports, electronic resources, and authors' experience in machine learning. References were chosen when possible for accessibility to practicing pathologists without specialized training in mathematics, statistics, or software development.

CONCLUSIONS.—

Machine learning offers an array of techniques that in recent published results show substantial promise. Data suggest that human experts working with machine learning tools outperform humans or machines separately, but the optimal form for this combination in pathology has not been established. Significant questions related to the generalizability of machine learning systems, local site verification, and performance monitoring remain to be resolved before a consensus on best practices and a regulatory environment can be established.

Stage-Specific PET Radiomic Prediction Model for the Histological Subtype Classification of Non-Small-Cell Lung Cancer

Purpose

To investigate the impact of staging on differences in glucose metabolic heterogeneity between lung adenocarcinoma (ADC) and squamous cell carcinoma (SCC) by ¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET) textural analysis and to develop a stage-specific PET radiomic prediction model to distinguish lung ADC from SCC.

Patients and Methods

Patients who were histologically diagnosed with lung ADC or SCC and underwent pretreatment ¹⁸F-FDG PET/CT scans were retrospectively identified. Radiomic features were extracted from a semiautomatically outlined tumor region in the Chang-Gung Image Texture Analysis (CGITA) software package. The differences in radiomic parameters between lung ADC and SCC were compared stage-by-stage in 253 consecutive NSCLC patients with stages I to III disease. The least absolute shrinkage and selection operator (LASSO) algorithm was used for feature selection. A radiomic signature for each stage was subsequently constructed and evaluated. Then, an individual nomogram incorporating the radiomic signature and clinical risk factors was established and evaluated. The performance of the constructed models was assessed by receiver operating characteristic (ROC) curve analysis, and the nomogram was further validated by calibration curve analysis.

Results

The performance of the radiomic signature for distinguishing lung ADC and SCC in both the training and validation cohorts was good, with AUCs of 0.883, 0.854, and 0.895 in the training cohort and 0.932, 0.944, and 0.886 in the validation cohort for stages I, II, and III NSCLC, respectively. The radiomic-clinical nomogram integrating radiomic features with independent clinical predictors exhibited more favorable discriminative performance, with AUCs of 0.982, 0.963, and 0.979 in the training cohort and 0.989, 0.984, and 0.978 in the validation cohort for stages I, II, and III, respectively.

Conclusion

Differences in PET radiomic features between lung ADC and SCC varied in different stages. Stage-specific PET radiomic prediction models provided more favorable performance for discriminating the histological subtype of NSCLC.

Identifying Cross-Scale Associations between Radiomic and Pathomic Signatures of Non-Small Cell Lung Cancer Subtypes: Preliminary Results

(1) Background: Despite the complementarity between radiology and histopathology, both from a diagnostic and a prognostic perspective, quantitative analyses of these modalities are usually performed in disconnected silos. This work presents initial results for differentiating two major non-small cell lung cancer (NSCLC) subtypes by exploring cross-scale associations between Computed Tomography (CT) images and corresponding digitized pathology images. (2) Methods: The analysis comprised three phases, (i) a multi-resolution cell density quantification to identify discriminant pathomic patterns for differentiating adenocarcinoma (ADC) and squamous cell carcinoma (SCC), (ii) radiomic characterization of CT images by using Haralick descriptors to quantify tumor textural heterogeneity as represented by gray-level co-occurrences to discriminate the two pathological subtypes, and (iii) quantitative correlation analysis between the multi-modal features to identify potential associations between them. This analysis was carried out using two publicly available digitized pathology databases (117 cases from TCGA and 54 cases from CPTAC) and a public radiological collection of CT images (101 cases from NSCLC-R). (3) Results: The top-ranked cell density pathomic features from the histopathology analysis were correlation, contrast, homogeneity, sum of entropy and difference of variance; which yielded a cross-validated AUC of 0.72 ± 0.02 on the training set (CPTAC) and hold-out validation AUC of 0.77 on the testing set (TCGA). Top-ranked co-occurrence radiomic features within NSCLC-R were contrast, correlation and sum of entropy which yielded a cross-validated AUC of 0.72 ± 0.01. Preliminary but significant cross-scale associations were identified between cell density statistics and CT intensity values using matched specimens available in the TCGA cohort, which were used to significantly improve the overall discriminatory performance of radiomic features in differentiating NSCLC subtypes (AUC = 0.78 ± 0.01). (4) Conclusions: Initial results suggest that cross-scale associations may exist between digital pathology and CT imaging which can be used to identify relevant radiomic and histopathology features to accurately distinguish lung adenocarcinomas from squamous cell carcinomas.

Prediction of KRAS, NRAS and BRAF status in colorectal cancer patients with liver metastasis using a deep artificial neural network based on radiomics and semantic features

There is a critical need for development of improved methods capable of accurately predicting the RAS (KRAS and NRAS) and BRAF gene mutation status in patients with advanced colorectal cancer (CRC). The purpose of this study was to investigate whether radiomics and/or semantic features could improve the detection accuracy of RAS/BRAF gene mutation status in patients with colorectal liver metastasis (CRLM). In this retrospective study, 159 patients who had been diagnosed with CRLM in two hospitals were enrolled. All patients received lung and abdominal contrast-enhanced CT (CECT) scans prior to radiation therapy and chemotherapy. Semantic features were independently assessed by two radiologists. Radiomics features were extracted from the portal venous phase (PVP) of the CT scan for each patient. Seven machine learning algorithms were used to establish three scores based on the semantic, radiomics and the combination of both features. Two semantic and 851 radiomics features were used to predict the mutation status of RAS and BRAF using an artificial neural network method (ANN). This approach performed best out of the seven tested algorithms. We constructed three scores which were based on radiomics, semantic features and the combined scores. The combined score could distinguish between wild-type and mutant patients with an AUC of 0.95 in the primary cohort and 0.79 in the validation cohort. This study proved that the application of radiomics together with semantic features can improve non-invasive assessment of the gene mutation status of RAS (KRAS and NRAS) and BRAF in CRLM.

Role of SOX Protein Groups F and H in Lung Cancer Progression

Simple Summary

The expression of SOX proteins has been demonstrated in many tissues at various stages of embryogenesis, where they play the role of transcription factors. The SOX18 protein (along with SOX7 and SOX17) belongs to the SOXF group and is mainly involved in the development of the cardiovascular system, where its expression was found in the endothelium. SOX18 expression was also demonstrated in neoplastic lines of gastric, pancreatic and colon adenocarcinomas. The prognostic role of SOX30 expression has only been studied in lung adenocarcinomas, where a low expression of this factor in the stromal tumor was associated with a worse prognosis for patients. Because of the complexity of non-small-cell lung cancer (NSCLC) development, the role of the SOX proteins in this malignancy is still not fully understood. Many recently published papers show that SOX family protein members play a crucial role in the progression of NSCLC.

Abstract

The SOX family proteins are proved to play a crucial role in the development of the lymphatic ducts and the cardiovascular system. Moreover, an increased expression level of the SOX18 protein has been found in many malignances, such as melanoma, stomach, pancreatic breast and lung cancers. Another SOX family protein, the SOX30 transcription factor, is responsible for the development of male germ cells. Additionally, recent studies have shown its proapoptotic character in non-small cell lung cancer cells. Our preliminary studies showed a disparity in the amount of mRNA of the SOX18 gene relative to the amount of protein. This is why our attention has been focused on microRNA (miRNA) molecules, which could regulate the SOX18 gene transcript level. Recent data point to the fact that, in practically all types of cancer, hundreds of genes exhibit an abnormal methylation, covering around 5–10% of the thousands of CpG islands present in the promoter sequences, which in normal cells should not be methylated from the moment the embryo finishes its development. It has been demonstrated that in non-small-cell lung cancer (NSCLC) cases there is a large heterogeneity of the methylation process. The role of the SOX18 and SOX30 expression in non-small-cell lung cancers (NSCLCs) is not yet fully understood. However, if we take into account previous reports, these proteins may be important factors in the development and progression of these malignancies.

A Non-invasive Method to Diagnose Lung Adenocarcinoma

Purpose: To find out the CT radiomics features of differentiating lung adenocarcinoma from another lung cancer histological type. Methods: This was a historical cohort study, three independent lung cancer cohorts included. One cohort was used to evaluate the stability of radiomics features, one cohort was used to feature selection, and the last was used to construct and evaluate classification models. The research is divided into four steps: region of interest segmentation, feature extraction, feature selection, and model building and validation. The feature selection methods included the intraclass correlation coefficient, ReliefF coefficient, and Partition-Membership filter. The performance metrics of the classification model included accuracy (Acc), precision (Pre), area under curve (AUC), and kappa statistics. Results: The 10 features (First order shape features: Sphericity and Compacity, Gray-Level Run Length Matrix: Short-Run Emphasis, Low Gray-level Run Emphasis, and High Gray-level Run Emphasis, Gray Level Co-occurrence Matrix: Homogeneity, Energy, Contrast, Correlation, and Dissimilarity) showed the most stable and classification capability. The 6 classifiers, Logistic regression classifier (LR), Sequence Minimum Optimization algorithm, Random Forest, KStar, Naive Bayes and Random Committee, have great performance both on the train and the test sets, and especially LR has the best performance on the test set (Acc = 98.72, Pre = 0.988, AUC = 1, and kappa = 0.974). Conclusion: Lung adenocarcinoma can be identified based on CT radiomics features. We can diagnose lung adenocarcinoma with CT non-invasively.

The diagnostic accuracy of artificial intelligence in thoracic diseases: A protocol for systematic review and meta-analysis

INTRODUCTION

Thoracic diseases include a variety of common human primary malignant tumors, among which lung cancer and esophageal cancer are among the top 10 in cancer incidence and mortality. Early diagnosis is an important part of cancer treatment, so artificial intelligence (AI) systems have been developed for the accurate and automated detection and diagnosis of thoracic tumors. However, the complicated AI structure and image processing made the diagnosis result of AI-based system unstable. The purpose of this study is to systematically review published evidence to explore the accuracy of AI systems in diagnosing thoracic cancers.

METHODS AND ANALYSIS

We will conduct a systematic review and meta-analysis of the diagnostic accuracy of AI systems for the prediction of thoracic diseases. The primary objective is to assess the diagnostic accuracy of thoracic cancers, including assessing potential biases and calculating combined estimates of sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). The secondary objective is to evaluate the factors associated with different models, classifiers, and radiomics information. We will search databases such as PubMed/MEDLINE, Embase (via OVID), and the Cochrane Library. Two reviewers will independently screen titles and abstracts, perform full article reviews and extract study data. We will report study characteristics and assess methodological quality using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool. RevMan 5.3 and Meta-disc 1.4 software will be used for data synthesis. If pooling is appropriate, we will produce summary receiver operating characteristic (SROC) curves, summary operating points (pooled sensitivity and specificity), and 95% confidence intervals around the summary operating points. Methodological subgroup and sensitivity analyses will be performed to explore heterogeneity.

PROSPERO REGISTRATION NUMBER

CRD42019135247.

Elaboration of a multimodal MRI-based radiomics signature for the preoperative prediction of the histological subtype in patients with non-small-cell lung cancer

BACKGROUND

Non-invasive discrimination between lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) subtypes of non-small-cell lung cancer (NSCLC) could be very beneficial to the patients unfit for the invasive diagnostic procedures. The aim of this study was to investigate the feasibility of utilizing the multimodal magnetic resonance imaging (MRI) radiomics and clinical features in classifying NSCLC. This retrospective study involved 148 eligible patients with postoperative pathologically confirmed NSCLC. The study was conducted in three steps: (1) feature extraction was performed using the online freely available package with the multimodal MRI data; (2) feature selection was performed using the Student's t test and support vector machine (SVM)-based recursive feature elimination method with the training cohort (n = 100), and the performance of these selected features was evaluated using both the training and the validation cohorts (n = 48) with a non-linear SVM classifier; (3) a Radscore model was then generated using logistic regression algorithm; (4) Integrating the Radscore with the semantic clinical features, a radiomics-clinical nomogram was developed, and its overall performance was evaluated with both cohorts.

RESULTS

Thirteen optimal features achieved favorable discrimination performance with both cohorts, with area under the curve (AUC) of 0.819 and 0.824, respectively. The radiomics-clinical nomogram integrating the Radscore with the independent clinical predictors exhibited more favorable discriminative power, with AUC improved to 0.901 and 0.872 in both cohorts, respectively. The Hosmer-Lemeshow test and decision curve analysis results furtherly showed good predictive precision and clinical usefulness of the nomogram.

CONCLUSION

Non-invasive histological subtype stratification of NSCLC can be done favorably using multimodal MRI radiomics features. Integrating the radiomics features with the clinical features could further improve the performance of the histological subtype stratification in patients with NSCLC.