Radiomics and deep learning methods for the prediction of 2-year overall survival in LUNG1 dataset

Predicting the risk of type 2 diabetes mellitus (T2DM) emergence in 5 years using mammography images: a comparison study between radiomics and deep learning algorithm

Purpose

The prevalence of type 2 diabetes mellitus (T2DM) has been steadily increasing over the years. We aim to predict the occurrence of T2DM using mammography images within 5 years using two different methods and compare their performance.

Approach

We examined 312 samples, including 110 positive cases (developed T2DM after 5 years) and 202 negative cases (did not develop T2DM) using two different methods. In the first method, a radiomics-based approach, we utilized radiomics features and machine learning (ML) algorithms. The entire breast region was chosen as the region of interest for extracting radiomics features. Then, a binary breast image was created from which we extracted 668 features and analyzed them using various ML algorithms. In the second method, a complex convolutional neural network (CNN) with a modified ResNet architecture and various kernel sizes was applied to raw mammography images for the prediction task. A nested, stratified five-fold cross-validation was done for both parts A and B to compute accuracy, sensitivity, specificity, and area under the receiver operating curve (AUROC). Hyperparameter tuning was also done to enhance the model's performance and reliability.

Results

The radiomics approach's light gradient boosting model gave 68.9% accuracy, 30.7% sensitivity, 89.5% specificity, and 0.63 AUROC. The CNN method achieved an AUROC of 0.58 over 20 epochs.

Conclusion

Radiomics outperformed CNN by 0.05 in terms of AUROC. This may be due to the more straightforward interpretability and clinical relevance of predefined radiomics features compared with the complex, abstract features learned by CNNs.

Predicting early stage lung cancer recurrence and survival from combined tumor motion amplitude and radiomics on free-breathing 4D-CT

BACKGROUND

Cancer control outcomes of lung cancer are hypothesized to be affected by several confounding factors, including tumor heterogeneity and patient history, which have been hypothesized to mitigate the dose delivery effectiveness when treated with radiation therapy. Providing an accurate predictive model to identify patients at risk would enable tailored follow-up strategies during treatment.

PURPOSE

Our goal is to demonstrate the added prognostic value of including tumor displacement amplitude in a predictive model that combines clinical features and computed tomography (CT) radiomics for 2-year recurrence and survival in non-small-cell lung cancer (NSCLC) patients treated with curative-intent stereotactic body radiation therapy.

METHODS

A cohort of 381 patients treated for primary lung cancer with radiotherapy was collected, each including a planning CT with a dosimetry plan, 4D-CT, and clinical information. From this cohort, 101 patients (26.5%) experienced cancer progression (locoregional/distant metastasis) or death within 2 years of the end of treatment. Imaging data was analyzed for radiomics features from the tumor segmented image, as well as tumor motion amplitude measured on 4D-CT. A random forest (RF) model was developed to predict the overall outcomes, which was compared to three other approaches - logistic regression, support vector machine, and convolutional neural networks.

RESULTS

A 6-fold cross-validation study yielded an area under the receiver operating characteristic curve of 72% for progression-free survival when combining clinical data with radiomics features and tumor motion using a RF model (72% sensitivity and 81% specificity). The combined model showed significant improvement compared to standard clinical data. Model performances for loco-regional recurrence and overall survival sub-outcomes were established at 73% and 70%, respectively. No comparative methods reached statistical significance in any data configuration.

CONCLUSIONS

Combined tumor respiratory motion and radiomics features from planning CT showed promising predictive value for 2-year tumor control and survival, indicating the potential need for improving motion management strategies in future studies using machine learning-based prognosis models.

Integrating radiomic and 3D autoencoder-based features for Non-Small Cell Lung Cancer survival analysis

BACKGROUND AND OBJECTIVES

The aim of this study is to develop a radiomic and deep learning-based signature for survival analysis of patients with Non-Small Cell Lung Cancer.

METHODS

Four-hundred twenty-two patients from "Lung1" dataset were included in the study. A 3D convolutional autoencoder (AE) was built and features from the latent space extracted for further analysis. Radiomic features were derived from the 3D volume of the tumor region using PyRadiomics. Both radiomic and AE-based features underwent feature selection, by removing: i) highly correlated and ii) constant features. The selected variables were then used to derive both mono-domain (radiomics, AE and clinic) and multi-domain signatures fitting a Cox Proportional Hazard model with LASSO penalization and evaluated considering the concordance (C)-index as performance metric.

RESULTS

Both mono-domain and multi-domain signatures could significantly differentiate high risk from low risk patients. Among the mono-domain signatures, the highest hazard ratio (HR) in the test set was obtained using radiomics (HR = 1.5428) followed by the AE-based signature (HR = 1.5012) and the clinical signature (HR = 1.4770). The best overall performance was achieved by combining all three signatures, resulting in the highest HR (HR = 1.7383), while the combination of AE-based and clinical signatures yielded the highest C-index (C-index = 0.6309).

CONCLUSIONS

These preliminary results show that combining information carried by AE, radiomic and clinical domain shows potential for improving the prediction of overall survival in NSCLC patients.

A dual-radiomics model for overall survival prediction in early-stage NSCLC patient using pre-treatment CT images

Introduction

Radiation therapy (RT) is one of the primary treatment options for early-stage non-small cell lung cancer (ES-NSCLC). Therefore, accurately predicting the overall survival (OS) rate following radiotherapy is crucial for implementing personalized treatment strategies. This work aims to develop a dual-radiomics (DR) model to (1) predict 3-year OS in ES-NSCLC patients receiving RT using pre-treatment CT images, and (2) provide explanations between feature importanceand model prediction performance.

Methods

The publicly available TCIA Lung1 dataset with 132 ES-NSCLC patients received RT were studied: 89/43 patients in the under/over 3-year OS group. For each patient, two types of radiomic features were examined: 56 handcrafted radiomic features (HRFs) extracted within gross tumor volume, and 512 image deep features (IDFs) extracted using a pre-trained U-Net encoder. They were combined as inputs to an explainable boosting machine (EBM) model for OS prediction. The EBM's mean absolute scores for HRFs and IDFs were used as feature importance explanations. To evaluate identified feature importance, the DR model was compared with EBM using either (1) key or (2) non-key feature type only. Comparison studies with other models, including supporting vector machine (SVM) and random forest (RF), were also included. The performance was evaluated by the area under the receiver operating characteristic curve (AUCROC), accuracy, sensitivity, and specificity with a 100-fold Monte Carlo cross-validation.

Results

The DR model showed highestperformance in predicting 3-year OS (AUCROC=0.81 ± 0.04), and EBM scores suggested that IDFs showed significantly greater importance (normalized mean score=0.0019) than HRFs (score=0.0008). The comparison studies showed that EBM with key feature type (IDFs-only demonstrated comparable AUCROC results (0.81 ± 0.04), while EBM with non-key feature type (HRFs-only) showed limited AUCROC (0.64 ± 0.10). The results suggested that feature importance score identified by EBM is highly correlated with OS prediction performance. Both SVM and RF models were unable to explain key feature type while showing limited overall AUCROC=0.66 ± 0.07 and 0.77 ± 0.06, respectively. Accuracy, sensitivity, and specificity showed a similar trend.

Discussion

In conclusion, a DR model was successfully developed to predict ES-NSCLC OS based on pre-treatment CT images. The results suggested that the feature importance from DR model is highly correlated to the model prediction power.

Developing a Radiomics Atlas Dataset of normal Abdominal and Pelvic computed Tomography (RADAPT)

Atlases of normal genomics, transcriptomics, proteomics, and metabolomics have been published in an attempt to understand the biological phenotype in health and disease and to set the basis of comprehensive comparative omics studies. No such atlas exists for radiomics data. The purpose of this study was to systematically create a radiomics dataset of normal abdominal and pelvic radiomics that can be used for model development and validation. Young adults without any previously known disease, aged > 17 and ≤ 36 years old, were retrospectively included. All patients had undergone CT scanning for emergency indications. In case abnormal findings were identified, the relevant anatomical structures were excluded. Deep learning was used to automatically segment the majority of visible anatomical structures with the TotalSegmentator model as applied in 3DSlicer. Radiomics features including first order, texture, wavelet, and Laplacian of Gaussian transformed features were extracted with PyRadiomics. A Github repository was created to host the resulting dataset. Radiomics data were extracted from a total of 531 patients with a mean age of 26.8 ± 5.19 years, including 250 female and 281 male patients. A maximum of 53 anatomical structures were segmented and used for subsequent radiomics data extraction. Radiomics features were derived from a total of 526 non-contrast and 400 contrast-enhanced (portal venous) series. The dataset is publicly available for model development and validation purposes.

Enhancing Lung Cancer Survival Prediction: 3D CNN Analysis of CT Images Using Novel GTV1-SliceNum Feature and PEN-BCE Loss Function

Lung cancer is a prevalent malignancy associated with a high mortality rate, with a 5-year relative survival rate of 23%. Traditional survival analysis methods, reliant on clinician judgment, may lack accuracy due to their subjective nature. Consequently, there is growing interest in leveraging AI-based systems for survival analysis using clinical data and medical imaging. The purpose of this study is to improve survival classification for lung cancer patients by utilizing a 3D-CNN architecture (ResNet-34) applied to CT images from the NSCLC-Radiomics dataset. Through comprehensive ablation studies, we evaluate the effectiveness of different features and methodologies in classification performance. Key contributions include the introduction of a novel feature (GTV1-SliceNum), the proposal of a novel loss function (PEN-BCE) accounting for false negatives and false positives, and the showcasing of their efficacy in classification. Experimental work demonstrates results surpassing those of the existing literature, achieving a classification accuracy of 0.7434 and an ROC-AUC of 0.7768. The conclusions of this research indicate that the AI-driven approach significantly improves survival prediction for lung cancer patients, highlighting its potential for enhancing personalized treatment strategies and prognostic modeling.

Radiomics-Based Analysis of Intestinal Ultrasound Images for Inflammatory Bowel Disease: A Feasibility Study

Background

The increasing adoption of intestinal ultrasound (IUS) for monitoring inflammatory bowel diseases (IBD) by IBD providers has uncovered new challenges regarding standardized image interpretation and limitations as a research tool. Artificial intelligence approaches can help address these challenges. We aim to determine the feasibility of radiomic analysis of IUS images and to determine if a radiomics-based classification model can accurately differentiate between normal and abnormal IUS images. We will also compare the radiomic-based model's performance to a convolutional neural network (CNN)-based classification model to understand which method is more effective for extracting meaningful information from IUS images.

Methods

Retrospectively analyzing IUS images obtained during routine outpatient visits, we developed and tested radiomic-based and CNN-based models to distinguish between normal and abnormal images, with abnormal images defined as bowel wall thickness > 3 mm or bowel hyperemia with modified Limberg score ≥ 1 (both are surrogate markers for inflammation). Model performances were measured by area under the receiver operator curve (AUC).

Results

For this feasibility study, 125 images (33% abnormal) were analyzed. A radiomic-based model using XG boost yielded the best classifier model with average test AUC 0.98%, 93.8% sensitivity, 93.8% specificity, and 93.7% accuracy. The CNN-based classification model yielded an average testing AUC of 0.75.

Conclusions

Radiomic analysis of IUS images is feasible, and a radiomic-based classification model could accurately differentiate abnormal from normal images. Our findings establish methods to facilitate future radiomic-based IUS studies that can help standardize image interpretation and expand IUS research capabilities.

Foundation model for cancer imaging biomarkers

Foundation models in deep learning are characterized by a single large-scale model trained on vast amounts of data serving as the foundation for various downstream tasks. Foundation models are generally trained using self-supervised learning and excel in reducing the demand for training samples in downstream applications. This is especially important in medicine, where large labelled datasets are often scarce. Here, we developed a foundation model for cancer imaging biomarker discovery by training a convolutional encoder through self-supervised learning using a comprehensive dataset of 11,467 radiographic lesions. The foundation model was evaluated in distinct and clinically relevant applications of cancer imaging-based biomarkers. We found that it facilitated better and more efficient learning of imaging biomarkers and yielded task-specific models that significantly outperformed conventional supervised and other state-of-the-art pretrained implementations on downstream tasks, especially when training dataset sizes were very limited. Furthermore, the foundation model was more stable to input variations and showed strong associations with underlying biology. Our results demonstrate the tremendous potential of foundation models in discovering new imaging biomarkers that may extend to other clinical use cases and can accelerate the widespread translation of imaging biomarkers into clinical settings.

Sacroiliitis diagnosis based on interpretable features and multi-task learning

Objective.Sacroiliitis is an early pathological manifestation of ankylosing spondylitis (AS), and a positive sacroiliitis test on imaging may help clinical practitioners diagnose AS early. Deep learning based automatic diagnosis algorithms can deliver grading findings for sacroiliitis, however, it requires a large amount of data with precise labels to train the model and lacks grading features visualization. In this paper, we aimed to propose a radiomics and deep learning based deep feature visualization positive diagnosis algorithm for sacroiliitis on CT scans. Visualization of grading features can enhance clinical interpretability with visual grading features, which assist doctors in diagnosis and treatment more effectively.Approach.The region of interest (ROI) is identified by segmenting the sacroiliac joint (SIJ) 3D CT images using a combination of the U-net model and certain statistical approaches. Then, in addition to extracting spatial and frequency domain features from ROI according to the radiographic manifestations of sacroiliitis, the radiomics features have also been integrated into the proposed encoder module to obtain a powerful encoder and extract features effectively. Finally, a multi-task learning technique and five-class labels are utilized to help with performing positive tests to reduce discrepancies in the evaluation of several radiologists.Main results.On our private dataset, proposed methods have obtained an accuracy rate of 87.3%, which is 9.8% higher than the baseline and consistent with assessments made by qualified medical professionals.Significance.The results of the ablation experiment and interpreting analysis demonstrated that the proposed methods are applied in automatic CT scan sacroiliitis diagnosis due to their excellently interpretable and portable advantages.

AI-Driven Models for Diagnosing and Predicting Outcomes in Lung Cancer: A Systematic Review and Meta-Analysis

(1) Background: Lung cancer's high mortality due to late diagnosis highlights a need for early detection strategies. Artificial intelligence (AI) in healthcare, particularly for lung cancer, offers promise by analyzing medical data for early identification and personalized treatment. This systematic review evaluates AI's performance in early lung cancer detection, analyzing its techniques, strengths, limitations, and comparative edge over traditional methods. (2) Methods: This systematic review and meta-analysis followed the PRISMA guidelines rigorously, outlining a comprehensive protocol and employing tailored search strategies across diverse databases. Two reviewers independently screened studies based on predefined criteria, ensuring the selection of high-quality data relevant to AI's role in lung cancer detection. The extraction of key study details and performance metrics, followed by quality assessment, facilitated a robust analysis using R software (Version 4.3.0). The process, depicted via a PRISMA flow diagram, allowed for the meticulous evaluation and synthesis of the findings in this review. (3) Results: From 1024 records, 39 studies met the inclusion criteria, showcasing diverse AI model applications for lung cancer detection, emphasizing varying strengths among the studies. These findings underscore AI's potential for early lung cancer diagnosis but highlight the need for standardization amidst study variations. The results demonstrate promising pooled sensitivity and specificity of 0.87, signifying AI's accuracy in identifying true positives and negatives, despite the observed heterogeneity attributed to diverse study parameters. (4) Conclusions: AI demonstrates promise in early lung cancer detection, showing high accuracy levels in this systematic review. However, study variations underline the need for standardized protocols to fully leverage AI's potential in revolutionizing early diagnosis, ultimately benefiting patients and healthcare professionals. As the field progresses, validated AI models from large-scale perspective studies will greatly benefit clinical practice and patient care in the future.

The effect of data resampling methods in radiomics

Radiomic datasets can be class-imbalanced, for instance, when the prevalence of diseases varies notably, meaning that the number of positive samples is much smaller than that of negative samples. In these cases, the majority class may dominate the model's training and thus negatively affect the model's predictive performance, leading to bias. Therefore, resampling methods are often utilized to class-balance the data. However, several resampling methods exist, and neither their relative predictive performance nor their impact on feature selection has been systematically analyzed. In this study, we aimed to measure the impact of nine resampling methods on radiomic models utilizing a set of fifteen publicly available datasets regarding their predictive performance. Furthermore, we evaluated the agreement and similarity of the set of selected features. Our results show that applying resampling methods did not improve the predictive performance on average. On specific datasets, slight improvements in predictive performance (+ 0.015 in AUC) could be seen. A considerable disagreement on the set of selected features was seen (only 28.7% of features agreed), which strongly impedes feature interpretability. However, selected features are similar when considering their correlation (82.9% of features correlated on average).

Multi-institutional prognostic modeling of survival outcomes in NSCLC patients treated with first-line immunotherapy using radiomics

BACKGROUND

Immune checkpoint inhibitors (ICIs) have emerged as one of the most promising first-line therapeutics in the management of non-small cell lung cancer (NSCLC). However, only a subset of these patients responds to ICIs, highlighting the clinical need to develop better predictive and prognostic biomarkers. This study will leverage pre-treatment imaging profiles to develop survival risk models for NSCLC patients treated with first-line immunotherapy.

METHODS

Advanced NSCLC patients (n = 149) were retrospectively identified from two institutions who were treated with first-line ICIs. Radiomics features extracted from pretreatment imaging scans were used to build the predictive models for progression-free survival (PFS) and overall survival (OS). A compendium of five feature selection methods and seven machine learning approaches were utilized to build the survival risk models. The concordance index (C-index) was used to evaluate model performance.

RESULTS

From our results, we found several combinations of machine learning algorithms and feature selection methods to achieve similar performance. K-nearest neighbourhood (KNN) with ReliefF (RL) feature selection was the best-performing model to predict PFS (C-index = 0.61 and 0.604 in discovery and validation cohorts), while XGBoost with Mutual Information (MI) feature selection was the best-performing model for OS (C-index = 0.7 and 0.655 in discovery and validation cohorts).

CONCLUSION

The results of this study highlight the importance of implementing an appropriate feature selection method coupled with a machine learning strategy to develop robust survival models. With further validation of these models on external cohorts when available, this can have the potential to improve clinical decisions by systematically analyzing routine medical images.

Foundation Models for Quantitative Biomarker Discovery in Cancer Imaging

Foundation models represent a recent paradigm shift in deep learning, where a single large-scale model trained on vast amounts of data can serve as the foundation for various downstream tasks. Foundation models are generally trained using self-supervised learning and excel in reducing the demand for training samples in downstream applications. This is especially important in medicine, where large labeled datasets are often scarce. Here, we developed a foundation model for imaging biomarker discovery by training a convolutional encoder through self-supervised learning using a comprehensive dataset of 11,467 radiographic lesions. The foundation model was evaluated in distinct and clinically relevant applications of imaging-based biomarkers. We found that they facilitated better and more efficient learning of imaging biomarkers and yielded task-specific models that significantly outperformed their conventional supervised counterparts on downstream tasks. The performance gain was most prominent when training dataset sizes were very limited. Furthermore, foundation models were more stable to input and inter-reader variations and showed stronger associations with underlying biology. Our results demonstrate the tremendous potential of foundation models in discovering novel imaging biomarkers that may extend to other clinical use cases and can accelerate the widespread translation of imaging biomarkers into clinical settings.

A Radiomics-Clinical Model Predicts Overall Survival of Non-Small Cell Lung Cancer Patients Treated with Immunotherapy: A Multicenter Study

BACKGROUND

Immune checkpoint inhibitors (ICIs) are a great breakthrough in cancer treatments and provide improved long-term survival in a subset of non-small cell lung cancer (NSCLC) patients. However, prognostic and predictive biomarkers of immunotherapy still remain an unmet clinical need. In this work, we aim to leverage imaging data and clinical variables to develop survival risk models among advanced NSCLC patients treated with immunotherapy.

METHODS

This retrospective study includes a total of 385 patients from two institutions who were treated with ICIs. Radiomics features extracted from pretreatment CT scans were used to build predictive models. The objectives were to predict overall survival (OS) along with building a classifier for short- and long-term survival groups. We employed the XGBoost learning method to build radiomics and integrated clinical-radiomics predictive models. Feature selection and model building were developed and validated on a multicenter cohort.

RESULTS

We developed parsimonious models that were associated with OS and a classifier for short- and long-term survivor groups. The concordance indices (C-index) of the radiomics model were 0.61 and 0.57 to predict OS in the discovery and validation cohorts, respectively. While the area under the curve (AUC) values of the radiomic models for short- and long-term groups were found to be 0.65 and 0.58 in the discovery and validation cohorts. The accuracy of the combined radiomics-clinical model resulted in 0.63 and 0.62 to predict OS and in 0.77 and 0.62 to classify the survival groups in the discovery and validation cohorts, respectively.

CONCLUSIONS

We developed and validated novel radiomics and integrated radiomics-clinical survival models among NSCLC patients treated with ICIs. This model has important translational implications, which can be used to identify a subset of patients who are not likely to benefit from immunotherapy. The developed imaging biomarkers may allow early prediction of low-group survivors, though additional validation of these radiomics models is warranted.

Radiomics approaches to predict PD-L1 and PFS in advanced non-small cell lung patients treated with immunotherapy: a multi-institutional study

With the increasing use of immune checkpoint inhibitors (ICIs), there is an urgent need to identify biomarkers to stratify responders and non-responders using programmed death-ligand (PD-L1) expression, and to predict patient-specific outcomes such as progression free survival (PFS). The current study is aimed to determine the feasibility of building imaging-based predictive biomarkers for PD-L1 and PFS through systematically evaluating a combination of several machine learning algorithms with different feature selection methods. A retrospective, multicenter study of 385 advanced NSCLC patients amenable to ICIs was undertaken in two academic centers. Radiomic features extracted from pretreatment CT scans were used to build predictive models for PD-L1 and PFS (short-term vs. long-term survivors). We first employed the LASSO methodology followed by five feature selection methods and seven machine learning approaches to build the predictors. From our analyses, we found several combinations of feature selection methods and machine learning algorithms to achieve a similar performance. Logistic regression with ReliefF feature selection (AUC = 0.64, 0.59 in discovery and validation cohorts) and SVM with Anova F-test feature selection (AUC = 0.64, 0.63 in discovery and validation datasets) were the best-performing models to predict PD-L1 and PFS. This study elucidates the application of suitable feature selection approaches and machine learning algorithms to predict clinical endpoints using radiomics features. Through this study, we identified a subset of algorithms that should be considered in future investigations for building robust and clinically relevant predictive models.

Additional Value of PET and CT Image-Based Features in the Detection of Occult Lymph Node Metastases in Lung Cancer: A Systematic Review of the Literature

Lung cancer represents the second most common malignancy worldwide and lymph node (LN) involvement serves as a crucial prognostic factor for tailoring treatment approaches. Invasive methods, such as mediastinoscopy and endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA), are employed for preoperative LN staging. Among the preoperative non-invasive diagnostic methods, computed tomography (CT) and, recently, positron emission tomography (PET)/CT with fluorine-18-fludeoxyglucose ([¹⁸F]FDG) are routinely recommended by several guidelines; however, they can both miss pathologically proven LN metastases, with an incidence up to 26% for patients staged with [¹⁸F]FDG PET/CT. These undetected metastases, known as occult LN metastases (OLMs), are usually cases of micro-metastasis or small LN metastasis (shortest radius below 10 mm). Hence, it is crucial to find novel approaches to increase their discovery rate. Radiomics is an emerging field that seeks to uncover and quantify the concealed information present in biomedical images by utilising machine or deep learning approaches. The extracted features can be integrated into predictive models, as numerous reports have emphasised their usefulness in the staging of lung cancer. However, there is a paucity of studies examining the detection of OLMs using quantitative features derived from images. Hence, the objective of this review was to investigate the potential application of PET- and/or CT-derived quantitative radiomic features for the identification of OLMs.

A CT-based transfer learning approach to predict NSCLC recurrence: The added-value of peritumoral region

Non-small cell lung cancer (NSCLC) represents 85% of all new lung cancer diagnoses and presents a high recurrence rate after surgery. Thus, an accurate prediction of recurrence risk in NSCLC patients at diagnosis could be essential to designate risk patients to more aggressive medical treatments. In this manuscript, we apply a transfer learning approach to predict recurrence in NSCLC patients, exploiting only data acquired during its screening phase. Particularly, we used a public radiogenomic dataset of NSCLC patients having a primary tumor CT image and clinical information. Starting from the CT slice containing the tumor with maximum area, we considered three different dilatation sizes to identify three Regions of Interest (ROIs): CROP (without dilation), CROP 10 and CROP 20. Then, from each ROI, we extracted radiomic features by means of different pre-trained CNNs. The latter have been combined with clinical information; thus, we trained a Support Vector Machine classifier to predict the NSCLC recurrence. The classification performances of the devised models were finally evaluated on both the hold-out training and hold-out test sets, in which the original sample has been previously divided. The experimental results showed that the model obtained analyzing CROP 20 images, which are the ROIs containing more peritumoral area, achieved the best performances on both the hold-out training set, with an AUC of 0.73, an Accuracy of 0.61, a Sensitivity of 0.63, and a Specificity of 0.60, and on the hold-out test set, with an AUC value of 0.83, an Accuracy value of 0.79, a Sensitivity value of 0.80, and a Specificity value of 0.78. The proposed model represents a promising procedure for early predicting recurrence risk in NSCLC patients.