Assessment of Stability and Discrimination Capacity of Radiomic Features on Apparent Diffusion Coefficient Images

Radiomics features outperform standard radiological measurements in detecting femoroacetabular impingement on three-dimensional magnetic resonance imaging

Femoroacetabular impingement (FAI) is a cause of hip pain and can lead to hip osteoarthritis. Radiological measurements obtained from radiographs or magnetic resonance imaging (MRI) are normally used for FAI diagnosis, but they require time-consuming manual interaction, which limits accuracy and reproducibility. This study compares standard radiologic measurements against radiomics features automatically extracted from MRI for the identification of FAI patients versus healthy subjects. Three-dimensional Dixon MRI of the pelvis were retrospectively collected for 10 patients with confirmed FAI and acquired for 10 healthy subjects. The femur and acetabulum were segmented bilaterally and associated radiomics features were extracted from the four MRI contrasts of the Dixon sequence (water-only, fat-only, in-phase, and out-of-phase). A radiologist collected 21 radiological measurements typically used in FAI. The Gini importance was used to define 9 subsets with the most predictive radiomics features and one subset for the most diagnostically relevant radiological measurements. For each subset, 100 Random Forest machine learning models were trained with different data splits and fivefold cross-validation to classify healthy subjects versus FAI patients. The average performance among the 100 models was computed for each subset and compared against the performance of the radiological measurements. One model trained using the radiomics features datasets yielded 100% accuracy in the detection of FAI, whereas all other radiomics features exceeded 80% accuracy. Radiological measurements yielded 74% accuracy, consistent with previous work. The results of this preliminary work highlight for the first time the potential of radiomics for fully automated FAI diagnosis.

Test-retest repeatability of myocardial radiomic features from quantitative cardiac magnetic resonance T1 and T2 mapping

Radiomics of cardiac magnetic resonance (MR) imaging has proved to be potentially useful in the study of various myocardial diseases. Therefore, assessing the repeatability degree in radiomic features measurement is of fundamental importance. The aim of this study was to assess test-retest repeatability of myocardial radiomic features extracted from quantitative T1 and T2 maps. A representative group of 24 subjects (mean age 54 ± 18 years) referred for clinical cardiac MR imaging were enrolled in the study. For each subject, T1 and T2 mapping through MOLLI and T2-prepared TrueFISP acquisition sequences, respectively, were performed at 1.5 T. Then, 98 radiomic features of different classes (shape, first-order, second-order) were extracted from a region of interest encompassing the whole left ventricle myocardium in a short axis slice. The repeatability was assessed performing different and complementary analyses: intraclass correlation coefficient (ICC) and limits of agreement (LOA) (i.e., the interval within which 95% of the percentage differences between two repeated measures are expected to lie). Radiomic features were characterized by a relatively wide range of repeatability degree in terms of both ICC and LOA. Overall, 44.9% and 38.8% of radiomic features showed ICC values > 0.75 for T1 and T2 maps, respectively, while 25.5% and 23.4% of radiomic features showed LOA between ±10%. A subset of radiomic features for T1 (Mean, Median, 10Percentile, 90Percentile, RootMeanSquared, Imc2, RunLengthNonUniformityNormalized, RunPercentage and ShortRunEmphasis) and T2 (MaximumDiameter, RunLengthNonUniformityNormalized, RunPercentage, ShortRunEmphasis) maps presented both ICC > 0.75 and LOA between ±5%. Overall, radiomic features extracted from T1 maps showed better repeatability performance than those extracted from T2 maps, with shape features characterized by better repeatability than first-order and textural features. Moreover, only a limited subset of 9 and 4 radiomic features for T1 and T2 maps, respectively, showed high repeatability degree in terms of both ICC and LOA. These results confirm the importance of assessing test-retest repeatability degree in radiomic feature estimation and might be useful for a more effective/reliable use of myocardial T1 and T2 mapping radiomics in clinical or research studies.

A Novel Data Augmentation Method for Radiomics Analysis Using Image Perturbations

Radiomics extracts hundreds of features from medical images to quantitively characterize a region of interest (ROI). When applying radiomics, imbalanced or small dataset issues are commonly addressed using under or over-sampling, the latter being applied directly to the extracted features. Aim of this study is to propose a novel balancing and data augmentation technique by applying perturbations (erosion, dilation, contour randomization) to the ROI in cardiac computed tomography images. From the perturbed ROIs, radiomic features are extracted, thus creating additional samples. This approach was tested addressing the clinical problem of distinguishing cardiac amyloidosis (CA) from aortic stenosis (AS) and hypertrophic cardiomyopathy (HCM). Twenty-one CA, thirty-two AS and twenty-one HCM patients were included in the study. From each original and perturbed ROI, 107 radiomic features were extracted. The CA-AS dataset was balanced using the perturbation-based method along with random over-sampling, adaptive synthetic (ADASYN) and the synthetic minority oversampling technique (SMOTE). The same methods were tested to perform data augmentation dealing with CA and HCM. Features were submitted to robustness, redundancy, and relevance analysis testing five feature selection methods (p-value, least absolute shrinkage and selection operator (LASSO), semi-supervised LASSO, principal component analysis (PCA), semi-supervised PCA). Support vector machine performed the classification tasks, and its performance were evaluated by means of a 10-fold cross-validation. The perturbation-based approach provided the best performances in terms of f1 score and balanced accuracy in both CA-AS (f1 score: 80%, AUC: 0.91) and CA-HCM (f1 score: 86%, AUC: 0.92) classifications. These results suggest that ROI perturbations represent a powerful approach to address both data balancing and augmentation issues.

Predicting Osteoporosis and Osteopenia by Fusing Deep Transfer Learning Features and Classical Radiomics Features Based on Single-Source Dual-energy CT Imaging

RATIONALE AND OBJECTIVES

To develop and validate a predictive model for osteoporosis and osteopenia prediction by fusing deep transfer learning (DTL) features and classical radiomics features based on single-source dual-energy computed tomography (CT) virtual monochromatic imaging.

METHODS

A total of 606 lumbar vertebrae with dual-energy CT imaging and quantitative CT (QCT) evaluation were included in the retrospective study and randomly divided into the training (n = 424) and validation (n = 182) cohorts. Radiomics features and DTL features were extracted from 70-keV monochromatic CT images, followed by feature selection and model construction, radiomics and DTL features models were established. Then, we integrated the selected two types of features into a features fusion model. We developed a two-level classifier for the hierarchical pairwise classification of each vertebra. All the vertebrae were first classified into osteoporosis and non-osteoporosis groups, then non-osteoporosis group was classified into osteopenia and normal groups. QCT was used as reference. The predictive performance and clinical usefulness of three models were evaluated and compared.

RESULTS

The area under the curve (AUC) of the features fusion, radiomics and DTL models for the classification between osteoporosis and non-osteoporosis were 0.981, 0.999, 0.997 in the training cohort and 0.979, 0.943, 0.848 in the validation cohort. Furthermore, the AUCs of the previously mentioned models for the differentiation between osteopenia and normal were 0.994, 0.971, 0.996 in the training cohort and 0.990, 0.968, 0.908 in the validation cohort. The overall accuracy of the previously mentioned models for two-level classifications was 0.979, 0.955, 0.908 in the training cohort and 0.918, 0.885, 0.841 in the validation cohort. Decision curve analysis showed that all models had high clinical value.

CONCLUSION

The feature fusion model can be used for osteoporosis and osteopenia prediction with improved predictive ability over a radiomics model or a DTL model alone.

Assessing the stability and discriminative ability of radiomics features in the tumor microenvironment: Leveraging peri-tumoral regions in vestibular schwannoma

PURPOSE

The tumor microenvironment (TME) plays a crucial role in tumor progression and treatment response. Radiomics offers a non-invasive approach to studying the TME by extracting quantitative features from medical images. In this study, we present a novel approach to assess the stability and discriminative ability of radiomics features in the TME of vestibular schwannoma (VS).

METHODS

Magnetic Resonance Imaging (MRI) data from 242 VS patients were analyzed, including contrast-enhanced T1-weighted (ceT1) and high-resolution T2-weighted (hrT2) sequences. Radiomics features were extracted from concentric peri-tumoral regions of varying sizes. The intraclass correlation coefficient (ICC) was used to assess feature stability and discriminative ability, establishing quantile thresholds for ICCmin and ICCmax.

RESULTS

The identified thresholds for ICCmin and ICCmax were 0.45 and 0.72, respectively. Features were classified into four categories: stable and discriminative (S-D), stable and non-discriminative (S-ND), unstable and discriminative (US-D), and unstable and non-discriminative (US-ND). Different feature groups exhibited varying proportions of S-D features across ceT1 and hrT2 sequences. The similarity of S-D features between ceT1 and hrT2 sequences was evaluated using Jaccard's index, with a value of 0.78 for all feature groups which is ranging from 0.68 (intensity features) to 1.00 (Neighbouring Gray Tone Difference Matrix (NGTDM) features).

CONCLUSIONS

This study provides a framework for identifying stable and discriminative radiomics features in the TME, which could serve as potential biomarkers or predictors of patient outcomes, ultimately improving the management of VS patients.

Development and validation of a predictive model for vertebral fracture risk in osteoporosis patients

OBJECTIVE

This study aimed to develop and validate a predictive model for osteoporotic vertebral fractures (OVFs) risk by integrating demographic, bone mineral density (BMD), CT imaging, and deep learning radiomics features from CT images.

METHODS

A total of 169 osteoporosis-diagnosed patients from three hospitals were randomly split into OVFs (n = 77) and Non-OVFs (n = 92) groups for training (n = 135) and test (n = 34). Demographic data, BMD, and CT imaging details were collected. Deep transfer learning (DTL) using ResNet-50 and radiomics features were fused, with the best model chosen via logistic regression. Cox proportional hazards models identified clinical factors. Three models were constructed: clinical, radiomics-DTL, and fusion (clinical-radiomics-DTL). Performance was assessed using AUC, C-index, Kaplan-Meier, and calibration curves. The best model was depicted as a nomogram, and clinical utility was evaluated using decision curve analysis (DCA).

RESULTS

BMD, CT values of paravertebral muscles (PVM), and paravertebral muscles' cross-sectional area (CSA) significantly differed between OVFs and Non-OVFs groups (P < 0.05). No significant differences were found between training and test cohort. Multivariate Cox models identified BMD, CT values of PVM, and CSAPS reduction as independent OVFs risk factors (P < 0.05). The fusion model exhibited the highest predictive performance (C-index: 0.839 in training, 0.795 in test). DCA confirmed the nomogram's utility in OVFs risk prediction.

CONCLUSION

This study presents a robust predictive model for OVFs risk, integrating BMD, CT data, and radiomics-DTL features, offering high sensitivity and specificity. The model's visualizations can inform OVFs prevention and treatment strategies.

Constructing a Deep Learning Radiomics Model Based on X-ray Images and Clinical Data for Predicting and Distinguishing Acute and Chronic Osteoporotic Vertebral Fractures: A Multicenter Study

RATIONALE AND OBJECTIVES

To construct and validate a deep learning radiomics (DLR) model based on X-ray images for predicting and distinguishing acute and chronic osteoporotic vertebral fractures (OVFs).

METHODS

A total of 942 cases (1076 vertebral bodies) with both vertebral X-ray examination and MRI scans were included in this study from three hospitals. They were divided into a training cohort (n = 712), an internal validation cohort (n = 178), an external validation cohort (n = 111), and a prospective validation cohort (n = 75). The ResNet-50 model architecture was used for deep transfer learning (DTL), with pre-training performed on RadImageNet and ImageNet datasets. DTL features and radiomics features were extracted from lateral X-ray images of OVFs patients and fused together. A logistic regression model with the least absolute shrinkage and selection operator was established, with MRI showing bone marrow edema as the gold standard for acute OVFs. The performance of the model was evaluated using receiver operating characteristic curves. Eight machine learning classification models were evaluated for their ability to distinguish between acute and chronic OVFs. The Nomogram was constructed by combining clinical baseline data to achieve visualized classification assessment. The predictive performance of the best RadImageNet model and ImageNet model was compared using the Delong test. The clinical value of the Nomogram was evaluated using decision curve analysis (DCA).

RESULTS

Pre-training resulted in 34 and 39 fused features after feature selection and fusion. The most effective machine learning algorithm in both DLR models was Light Gradient Boosting Machine. Using the Delong test, the area under the curve (AUC) for distinguishing between acute and chronic OVFs in the training cohort was 0.979 and 0.972 for the RadImageNet and ImageNet models, respectively, with no statistically significant difference between them (P = 0.235). In the internal validation cohort, external validation cohort, and prospective validation cohort, the AUCs for the two models were 0.967 vs 0.629, 0.886 vs 0.817, and 0.933 vs 0.661, respectively, with statistically significant differences in all comparisons (P < 0.05). The deep learning radiomics nomogram (DLRN) was constructed by combining the predictive model of RadImageNet with clinical baseline features, resulting in AUCs of 0.981, 0.974, 0.895, and 0.902 in the training cohort, internal validation cohort, external validation cohort, and prospective validation cohort, respectively. Using the Delong test, the AUCs for the fused feature model and the DLRN in the training cohort were 0.979 and 0.981, respectively, with no statistically significant difference between them (P = 0.169). In the internal validation cohort, external validation cohort, and prospective validation cohort, the AUCs for the two models were 0.967 vs 0.974, 0.886 vs 0.895, and 0.933 vs 0.902, respectively, with statistically significant differences in all comparisons (P < 0.05). The Nomogram showed a slight improvement in predictive performance in the internal and external validation cohort, but a slight decrease in the prospective validation cohort (0.933 vs 0.902). DCA showed that the Nomogram provided more benefits to patients compared to the DLR models.

CONCLUSION

Compared to the ImageNet model, the RadImageNet model has higher diagnostic value in distinguishing between acute and chronic OVFs. Furthermore, the diagnostic performance of the model is further improved when combined with clinical baseline features to construct the Nomogram.

Exploring deep learning radiomics for classifying osteoporotic vertebral fractures in X-ray images

Purpose

To develop and validate a deep learning radiomics (DLR) model that uses X-ray images to predict the classification of osteoporotic vertebral fractures (OVFs).

Material and methods

The study encompassed a cohort of 942 patients, involving examinations of 1076 vertebrae through X-ray, CT, and MRI across three distinct hospitals. The OVFs were categorized as class 0, 1, or 2 based on the Assessment System of Thoracolumbar Osteoporotic Fracture. The dataset was divided randomly into four distinct subsets: a training set comprising 712 samples, an internal validation set with 178 samples, an external validation set containing 111 samples, and a prospective validation set consisting of 75 samples. The ResNet-50 architectural model was used to implement deep transfer learning (DTL), undergoing -pre-training separately on the RadImageNet and ImageNet datasets. Features from DTL and radiomics were extracted and integrated using X-ray images. The optimal fusion feature model was identified through least absolute shrinkage and selection operator logistic regression. Evaluation of the predictive capabilities for OVFs classification involved eight machine learning models, assessed through receiver operating characteristic curves employing the "One-vs-Rest" strategy. The Delong test was applied to compare the predictive performance of the superior RadImageNet model against the ImageNet model.

Results

Following pre-training separately on RadImageNet and ImageNet datasets, feature selection and fusion yielded 17 and 12 fusion features, respectively. Logistic regression emerged as the optimal machine learning algorithm for both DLR models. Across the training set, internal validation set, external validation set, and prospective validation set, the macro-average Area Under the Curve (AUC) based on the RadImageNet dataset surpassed those based on the ImageNet dataset, with statistically significant differences observed (P<0.05). Utilizing the binary "One-vs-Rest" strategy, the model based on the RadImageNet dataset demonstrated superior efficacy in predicting Class 0, achieving an AUC of 0.969 and accuracy of 0.863. Predicting Class 1 yielded an AUC of 0.945 and accuracy of 0.875, while for Class 2, the AUC and accuracy were 0.809 and 0.692, respectively.

Conclusion

The DLR model, based on the RadImageNet dataset, outperformed the ImageNet model in predicting the classification of OVFs, with generalizability confirmed in the prospective validation set.

The impact of radiomics in the management of soft tissue sarcoma

INTRODUCTION

Soft tissue sarcomas (STSs) are rare malignancies. Pre-therapeutic tumour grading and assessment are crucial in making treatment decisions. Radiomics is a high-throughput method for analysing imaging data, providing quantitative information beyond expert assessment. This review highlights the role of radiomic texture analysis in STSs evaluation.

MATERIALS AND METHODS

We conducted a systematic review according to the Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A comprehensive search was conducted in PubMed/MEDLINE and Scopus using the search terms: 'radiomics [All Fields] AND ("soft tissue sarcoma" [All Fields] OR "soft tissue sarcomas" [All Fields])'. Only original articles, referring to humans, were included.

RESULTS

A preliminary search conducted on PubMed/MEDLINE and Scopus provided 74 and 93 studies respectively. Based on the previously described criteria, 49 papers were selected, with a publication range from July 2015 to June 2023. The main domains of interest were risk stratification, histological grading prediction, technical feasibility/reproductive aspects, treatment response.

CONCLUSIONS

With an increasing interest over the last years, the use of radiomics appears to have potential for assessing STSs from initial diagnosis to predicting treatment response. However, additional and extensive research is necessary to validate the effectiveness of radiomics parameters and to integrate them into a comprehensive decision support system.

Developing a robust two-step machine learning multiclassification pipeline to predict primary site in head and neck carcinoma from lymph nodes

This study aimed to develop a robust multiclassification pipeline to determine the primary tumor location in patients with head and neck carcinoma of unknown primary using radiomics and machine learning techniques. The dataset included 400 head and neck cancer patients with primary tumor in oropharynx, OPC (n = 162), nasopharynx, NPC (n = 137), oral cavity, OC (n = 63), larynx and hypopharynx, HL (n = 38). Two radiomic-based multiclassification pipelines (P1 and P2) were developed. P1 consisted in a direct identification of the primary sites, whereas P2 was based on a two-step approach: in the first step, the number of classes was reduced by merging the two minority classes which were reclassified in the second step. Diverse correlation thresholds (0.75, 0.80, 0.85), feature selection methods (sequential forwards/backwards selection, sequential floating forward selection, neighborhood component analysis and minimum redundancy maximum relevance), and classification models (neural network, decision tree, naïve Bayes, bagged trees and support vector machine) were assessed. P2 outperformed P1, with the best results obtained with the support vector machine classifier including radiomic and clinical features (accuracies of 75.3 % (HL), 75.4 % (OC), 71.3 % (OPC), 92.9 % (NPC)). These results indicate that the two-step multiclassification pipeline integrating radiomics and clinical information is a promising approach to predict the tumor site of unknown primary.

Comparing effectiveness of image perturbation and test retest imaging in improving radiomic model reliability

Image perturbation is a promising technique to assess radiomic feature repeatability, but whether it can achieve the same effect as test-retest imaging on model reliability is unknown. This study aimed to compare radiomic model reliability based on repeatable features determined by the two methods using four different classifiers. A 191-patient public breast cancer dataset with 71 test-retest scans was used with pre-determined 117 training and 74 testing samples. We collected apparent diffusion coefficient images and manual tumor segmentations for radiomic feature extraction. Random translations, rotations, and contour randomizations were performed on the training images, and intra-class correlation coefficient (ICC) was used to filter high repeatable features. We evaluated model reliability in both internal generalizability and robustness, which were quantified by training and testing AUC and prediction ICC. Higher testing performance was found at higher feature ICC thresholds, but it dropped significantly at ICC = 0.95 for the test-retest model. Similar optimal reliability can be achieved with testing AUC = 0.7-0.8 and prediction ICC > 0.9 at the ICC threshold of 0.9. It is recommended to include feature repeatability analysis using image perturbation in any radiomic study when test-retest is not feasible, but care should be taken when deciding the optimal feature repeatability criteria.

Robustness of pulmonary nodule radiomic features on computed tomography as a function of varying radiation dose levels-a multi-dose in vivo patient study

OBJECTIVE

Analysis of textural features of pulmonary nodules in chest CT, also known as radiomics, has several potential clinical applications, such as diagnosis, prognostication, and treatment response monitoring. For clinical use, it is essential that these features provide robust measurements. Studies with phantoms and simulated lower dose levels have demonstrated that radiomic features can vary with different radiation dose levels. This study presents an in vivo stability analysis of radiomic features for pulmonary nodules against varying radiation dose levels.

METHODS

Nineteen patients with a total of thirty-five pulmonary nodules underwent four chest CT scans at different radiation dose levels (60, 33, 24, and 15 mAs) in a single session. The nodules were manually delineated. To assess the robustness of features, we calculated the intra-class correlation coefficient (ICC). To visualize the effect of milliampere-second variation on groups of features, a linear model was fitted to each feature. We calculated bias and calculated the R2 value as a measure of goodness of fit.

RESULTS

A small minority of 15/100 (15%) radiomic features were considered stable (ICC > 0.9). Bias increased and R2 decreased at lower dose, but shape features seemed to be more robust to milliampere-second variations than other feature classes.

CONCLUSION

A large majority of pulmonary nodule radiomic features were not inherently robust to radiation dose level variations. For a subset of features, it was possible to correct this variability by a simple linear model. However, the correction became increasingly less accurate at lower radiation dose levels.

CLINICAL RELEVANCE STATEMENT

Radiomic features provide a quantitative description of a tumor based on medical imaging such as computed tomography (CT). These features are potentially useful in several clinical tasks such as diagnosis, prognosis prediction, treatment effect monitoring, and treatment effect estimation.

KEY POINTS

• The vast majority of commonly used radiomic features are strongly influenced by variations in radiation dose level. • A small minority of radiomic features, notably the shape feature class, are robust against dose-level variations according to ICC calculations. • A large subset of radiomic features can be corrected by a linear model taking into account only the radiation dose level.

MRI-based radiomic prognostic signature for locally advanced oral cavity squamous cell carcinoma: development, testing and comparison with genomic prognostic signatures

BACKGROUND

. At present, the prognostic prediction in advanced oral cavity squamous cell carcinoma (OCSCC) is based on the tumor-node-metastasis (TNM) staging system, and the most used imaging modality in these patients is magnetic resonance image (MRI). With the aim to improve the prediction, we developed an MRI-based radiomic signature as a prognostic marker for overall survival (OS) in OCSCC patients and compared it with published gene expression signatures for prognosis of OS in head and neck cancer patients, replicated herein on our OCSCC dataset.

METHODS

For each patient, 1072 radiomic features were extracted from T1 and T2-weighted MRI (T1w and T2w). Features selection was performed, and an optimal set of five of them was used to fit a Cox proportional hazard regression model for OS. The radiomic signature was developed on a multi-centric locally advanced OCSCC retrospective dataset (n = 123) and validated on a prospective cohort (n = 108).

RESULTS

The performance of the signature was evaluated in terms of C-index (0.68 (IQR 0.66-0.70)), hazard ratio (HR 2.64 (95% CI 1.62-4.31)), and high/low risk group stratification (log-rank p < 0.001, Kaplan-Meier curves). When tested on a multi-centric prospective cohort (n = 108), the signature had a C-index of 0.62 (IQR 0.58-0.64) and outperformed the clinical and pathologic TNM stage and six out of seven gene expression prognostic signatures. In addition, the significant difference of the radiomic signature between stages III and IVa/b in patients receiving surgery suggests a potential association of MRI features with the pathologic stage.

CONCLUSIONS

Overall, the present study suggests that MRI signatures, containing non-invasive and cost-effective remarkable information, could be exploited as prognostic tools.

A robust radiomic-based machine learning approach to detect cardiac amyloidosis using cardiac computed tomography

Introduction

Cardiac amyloidosis (CA) shares similar clinical and imaging characteristics (e.g., hypertrophic phenotype) with aortic stenosis (AS), but its prognosis is generally worse than severe AS alone. Recent studies suggest that the presence of CA is frequent (1 out of 8 patients) in patients with severe AS. The coexistence of the two diseases complicates the prognosis and therapeutic management of both conditions. Thus, there is an urgent need to standardize and optimize the diagnostic process of CA and AS. The aim of this study is to develop a robust and reliable radiomics-based pipeline to differentiate the two pathologies.

Methods

Thirty patients were included in the study, equally divided between CA and AS. For each patient, a cardiac computed tomography (CCT) was analyzed by extracting 107 radiomics features from the LV wall. Feature robustness was evaluated by means of geometrical transformations to the ROIs and intra-class correlation coefficient (ICC) computation. Various correlation thresholds (0.80, 0.85, 0.90, 0.95, 1), feature selection methods [p-value, least absolute shrinkage and selection operator (LASSO), semi-supervised LASSO, principal component analysis (PCA), semi-supervised PCA, sequential forwards selection] and machine learning classifiers (k-nearest neighbors, support vector machine, decision tree, logistic regression and gradient boosting) were assessed using a leave-one-out cross-validation. Data augmentation was performed using the synthetic minority oversampling technique. Finally, explainability analysis was performed by using the SHapley Additive exPlanations (SHAP) method.

Results

Ninety-two radiomic features were selected as robust and used in the further steps. Best performances of classification were obtained using a correlation threshold of 0.95, PCA (keeping 95% of the variance, corresponding to 9 PCs) and support vector machine classifier reaching an accuracy, sensitivity and specificity of 0.93. Four PCs were found to be mainly dependent on textural features, two on first-order statistics and three on shape and size features.

Conclusion

These preliminary results show that radiomics might be used as non-invasive tool able to differentiate CA from AS using clinical routine available images.

Artificial intelligence-driven radiomics study in cancer: the role of feature engineering and modeling

Modern medicine is reliant on various medical imaging technologies for non-invasively observing patients' anatomy. However, the interpretation of medical images can be highly subjective and dependent on the expertise of clinicians. Moreover, some potentially useful quantitative information in medical images, especially that which is not visible to the naked eye, is often ignored during clinical practice. In contrast, radiomics performs high-throughput feature extraction from medical images, which enables quantitative analysis of medical images and prediction of various clinical endpoints. Studies have reported that radiomics exhibits promising performance in diagnosis and predicting treatment responses and prognosis, demonstrating its potential to be a non-invasive auxiliary tool for personalized medicine. However, radiomics remains in a developmental phase as numerous technical challenges have yet to be solved, especially in feature engineering and statistical modeling. In this review, we introduce the current utility of radiomics by summarizing research on its application in the diagnosis, prognosis, and prediction of treatment responses in patients with cancer. We focus on machine learning approaches, for feature extraction and selection during feature engineering and for imbalanced datasets and multi-modality fusion during statistical modeling. Furthermore, we introduce the stability, reproducibility, and interpretability of features, and the generalizability and interpretability of models. Finally, we offer possible solutions to current challenges in radiomics research.

A radiomics approach to the diagnosis of femoroacetabular impingement

Introduction

Femoroacetabular Impingement (FAI) is a hip pathology characterized by impingement of the femoral head-neck junction against the acetabular rim, due to abnormalities in bone morphology. FAI is normally diagnosed by manual evaluation of morphologic features on magnetic resonance imaging (MRI). In this study, we assess, for the first time, the feasibility of using radiomics to detect FAI by automatically extracting quantitative features from images.

Material and methods

17 patients diagnosed with monolateral FAI underwent pre-surgical MR imaging, including a 3D Dixon sequence of the pelvis. An expert radiologist drew regions of interest on the water-only Dixon images outlining femur and acetabulum in both impingement (IJ) and healthy joints (HJ). 182 radiomic features were extracted for each hip. The dataset numerosity was increased by 60 times with an ad-hoc data augmentation tool. Features were subdivided by type and region in 24 subsets. For each, a univariate ANOVA F-value analysis was applied to find the 5 features most correlated with IJ based on p-value, for a total of 48 subsets. For each subset, a K-nearest neighbor model was trained to differentiate between IJ and HJ using the values of the radiomic features in the subset as input. The training was repeated 100 times, randomly subdividing the data with 75%/25% training/testing.

Results

The texture-based gray level features yielded the highest prediction max accuracy (0.972) with the smallest subset of features. This suggests that the gray image values are more homogeneously distributed in the HJ in comparison to IJ, which could be due to stress-related inflammation resulting from impingement.

Conclusions

We showed that radiomics can automatically distinguish IJ from HJ using water-only Dixon MRI. To our knowledge, this is the first application of radiomics for FAI diagnosis. We reported an accuracy greater than 97%, which is higher than the 90% accuracy for detecting FAI reported for standard diagnostic tests (90%). Our proposed radiomic analysis could be combined with methods for automated joint segmentation to rapidly identify patients with FAI, avoiding time-consuming radiological measurements of bone morphology.

Differentiation of acute and chronic vertebral compression fractures using conventional CT based on deep transfer learning features and hand-crafted radiomics features

BACKGROUND

We evaluated the diagnostic efficacy of deep learning radiomics (DLR) and hand-crafted radiomics (HCR) features in differentiating acute and chronic vertebral compression fractures (VCFs).

METHODS

A total of 365 patients with VCFs were retrospectively analysed based on their computed tomography (CT) scan data. All patients completed MRI examination within 2 weeks. There were 315 acute VCFs and 205 chronic VCFs. Deep transfer learning (DTL) features and HCR features were extracted from CT images of patients with VCFs using DLR and traditional radiomics, respectively, and feature fusion was performed to establish the least absolute shrinkage and selection operator. The MRI display of vertebral bone marrow oedema was used as the gold standard for acute VCF, and the model performance was evaluated using the receiver operating characteristic (ROC).To separately evaluate the effectiveness of DLR, traditional radiomics and feature fusion in the differential diagnosis of acute and chronic VCFs, we constructed a nomogram based on the clinical baseline data to visualize the classification evaluation. The predictive power of each model was compared using the Delong test, and the clinical value of the nomogram was evaluated using decision curve analysis (DCA).

RESULTS

Fifty DTL features were obtained from DLR, 41 HCR features were obtained from traditional radiomics, and 77 features fusion were obtained after feature screening and fusion of the two. The area under the curve (AUC) of the DLR model in the training cohort and test cohort were 0.992 (95% confidence interval (CI), 0.983-0.999) and 0.871 (95% CI, 0.805-0.938), respectively. While the AUCs of the conventional radiomics model in the training cohort and test cohort were 0.973 (95% CI, 0.955-0.990) and 0.854 (95% CI, 0.773-0.934), respectively. The AUCs of the features fusion model in the training cohort and test cohort were 0.997 (95% CI, 0.994-0.999) and 0.915 (95% CI, 0.855-0.974), respectively. The AUCs of nomogram constructed by the features fusion in combination with clinical baseline data were 0.998 (95% CI, 0.996-0.999) and 0.946 (95% CI, 0.906-0.987) in the training cohort and test cohort, respectively. The Delong test showed that the differences between the features fusion model and the nomogram in the training cohort and the test cohort were not statistically significant (P values were 0.794 and 0.668, respectively), and the differences in the other prediction models in the training cohort and the test cohort were statistically significant (P < 0.05). DCA showed that the nomogram had high clinical value.

CONCLUSION

The features fusion model can be used for the differential diagnosis of acute and chronic VCFs, and its differential diagnosis ability is improved when compared with that when either radiomics is used alone. At the same time, the nomogram has a high predictive value for acute and chronic VCFs and can be a potential decision-making tool to assist clinicians, especially when a patient is unable to undergo spinal MRI examination.

Prognostic radiomic signature for head and neck cancer: development and validation on a multi-centric MRI dataset

BACKGROUND AND PURPOSE

Prognosis in locally advanced head and neck cancer (HNC) is currently based on TNM staging system and tumor subsite. However, quantitative imaging features (i.e., radiomic features) from magnetic resonance imaging (MRI) may provide additional prognostic info. The aim of this work is to develop and validate an MRI-based prognostic radiomic signature for locally advanced HNC.

MATERIALS AND METHODS

Radiomic features were extracted from T1- and T2-weighted MRI (T1w and T2w) using the segmentation of the primary tumor as mask. In total 1072 features (536 per image type) were extracted for each tumor. A retrospective multi-centric dataset (n=285) was used for features selection and model training. The selected features were used to fit a Cox proportional hazard regression model for overall survival (OS) that outputs the radiomic signature. The signature was then validated on a prospective multi-centric dataset (n=234). Prognostic performance for OS and disease-free survival (DFS) was evaluated using C-index. Additional prognostic value of the radiomic signature was explored.

RESULTS

The radiomic signature had C-index=0.64 for OS and C-index=0.60 for DFS in the validation set. The addition of the radiomic signature to other clinical features (TNM staging and tumor subsite) increased prognostic ability for both OS (HPV- C-index 0.63 to 0.65; HPV+ C-index 0.75 to 0.80) and DFS (HPV- C-index 0.58 to 0.61; HPV+ C-index 0.64 to 0.65).

CONCLUSION

An MRI-based prognostic radiomic signature was developed and prospectively validated. Such signature can successfully integrate clinical factors in both HPV+ and HPV- tumors. Grant support: European Union Horizon 2020 Framework Programme, Grant/Award, Number: 689715.

Radiomic Features Associated with Lymphoma Development in the Parotid Glands of Patients with Primary Sjögren's Syndrome

Non-Hodgkin Lymphoma (NHL) represents a severe complication and the main cause of morbidity in patients with primary Sjögren's syndrome (pSS). This study aimed to assess the role of textural analysis (TA) in revealing lymphoma-associated imaging parameters in the parotid gland (PG) parenchyma of patients with pSS. This retrospective study included a total of 36 patients (54.93 ± 13.34 years old; 91.6% females) diagnosed with pSS according to the American College of Rheumatology and the European League Against Rheumatism criteria (24 subjects with pSS and no lymphomatous proliferation; 12 subjects with pSS and NHL development in the PG, confirmed by the histopathological analysis). All subjects underwent MR scanning between January 2018 and October 2022. The coronal STIR PROPELLER sequence was employed to segment PG and perform TA using the MaZda5 software. A total of 65 PGs underwent segmentation and texture feature extraction (48 PGs were included in the pSS control group, and 17 PGs were included in the pSS NHL group). Following parameter reduction techniques, univariate analysis, multivariate regression, and receiver operating characteristics (ROC) analysis, the following TA parameters proved to be independently associated with NHL development in pSS: CH4S6_Sum_Variance and CV4S6_Inverse_Difference_Moment, with an area under ROC of 0.800 and 0.875, respectively. The radiomic model (resulting by combining the two previously independent TA features), presented 94.12% sensitivity and 85.42% specificity in differentiating between the two studied groups, reaching the highest area under ROC of 0.931 for the chosen cutoff value of 1.556. This study suggests the potential role of radiomics in revealing new imaging biomarkers that might serve as useful predictors for lymphoma development in patients with pSS. Further research on multicentric cohorts is warranted to confirm the obtained results and the added benefit of TA in risk stratification for patients with pSS.

Repeatability and reproducibility of MRI-radiomic features: A phantom experiment on a 1.5 T scanner

PURPOSE

Aim of this study is to assess the repeatability of radiomic features on magnetic resonance images (MRI) and their stability to variations in time of repetition (TR), time of echo (TE), slice thickness (ST), and pixel spacing (PS) using vegetable phantoms.

METHODS

The organic phantom was realized using two cucumbers placed inside a cylindrical container, and the analysis was performed using T1-weighted (T1w), T2-weighted (T2w), and diffusion-weighted images. One dataset was used to test the repeatability of the radiomic features, whereas other four datasets were used to test the sensitivity of the different MRI sequences to image acquisition parameters (TR, TE, ST, and PS). Four regions of interest (ROIs) were segmented: two for the central part of each cucumber and two for the external parts. Radiomic features were extracted from each ROI using Pyradiomics. To assess the effect of preprocessing on the reduction of variability, features were extracted both before and after the preprocessing. The coefficient of variation (CV) and intra-class correlation coefficient (ICC) were used to evaluate variability.

RESULTS

The use of intensity standardization increased the stability for the first-order statistics features. Shape and size features were always stable for all the analyses. Textural features were particularly sensitive to changes in ST and PS, although some increase in stability could be obtained by voxel size resampling. When images underwent image preprocessing, the number of stable features (ICC > 0.75 and mean absolute CV < 0.3) was 33 for apparent diffusion coefficient (ADC), 52 for T1w, and 73 for T2w.

CONCLUSIONS

The most critical source of variability is related to changes in voxel size (either caused by changes in ST or PS). Preprocessing increases features stability to both test-retest and variation of the image acquisition parameters for all the types of analyzed MRI (T1w, T2w, and ADC), except for ST.

3D vs. 2D MRI radiomics in skeletal Ewing sarcoma: Feature reproducibility and preliminary machine learning analysis on neoadjuvant chemotherapy response prediction

OBJECTIVE

The extent of response to neoadjuvant chemotherapy predicts survival in Ewing sarcoma. This study focuses on MRI radiomics of skeletal Ewing sarcoma and aims to investigate feature reproducibility and machine learning prediction of response to neoadjuvant chemotherapy.

MATERIALS AND METHODS

This retrospective study included thirty patients with biopsy-proven skeletal Ewing sarcoma, who were treated with neoadjuvant chemotherapy before surgery at two tertiary sarcoma centres. 7 patients were poor responders and 23 were good responders based on pathological assessment of the surgical specimen. On pre-treatment T1-weighted and T2-weighted MRI, 2D and 3D tumour segmentations were manually performed. Features were extracted from original and wavelet-transformed images. Feature reproducibility was assessed through small geometrical transformations of the regions of interest mimicking multiple manual delineations, and intraclass correlation coefficient >0.75 defined feature reproducibility. Feature selection also consisted of collinearity and significance analysis. After class balancing in the training cohort, three machine learning classifiers were trained and tested on unseen data using hold-out cross-validation.

RESULTS

1303 (77%) 3D and 620 (65%) 2D radiomic features were reproducible. 4 3D and 4 2D features passed feature selection. Logistic regression built upon 3D features achieved the best performance with 85% accuracy (AUC=0.9) in predicting response to neoadjuvant chemotherapy.

CONCLUSION

Compared to 2D approach, 3D MRI radiomics of Ewing sarcoma had superior reproducibility and higher accuracy in predicting response to neoadjuvant chemotherapy, particularly when using logistic regression classifier.

Investigation of the Inter- and Intrascanner Reproducibility and Repeatability of Radiomics Features in T1-Weighted Brain MRI

BACKGROUND

Radiomics is the high throughput analysis of medical images using computer algorithms, which specifically assess textural features. It has increasingly been proposed as a tool for the development of imaging biomarkers. However, an important acknowledged limitation of radiomics is the lack of reproducibility of features produced.

PURPOSE

To assess reproducibility and repeatability of radiomics variables in brain MRI through a multivisit, multicenter study.

STUDY TYPE

Retrospective.

POPULATION

Fourteen individuals visiting three institutions twice, 10 males with the mean age of 36.3 years and age range 25-51.

FIELD STRENGTH

3D T1W inversion recovery on three 1.5-T General Electric scanners.

ASSESSMENT

Radiomics analysis by a consultant radiologist performed on the T1W images of the whole brain on all visits. All possible radiomics features were generated.

STATISTICAL TEST

Concordance correlation coefficient (CCC) and dynamic range (DR) for all variables were calculated to assess the test-retest repeatability. Intraclass correlation coefficients (ICCs) were calculated to investigate the reproducibility of features across centers.

RESULTS

Of 1596 features generated, 57 from center 1, 15 from center 2, and 22 from center 3 had a CCC > 0.9 and DR > 0.9. Eight variables had CCC > 0.9 and DR > 0.9 in all centers. Forty-one variables had an ICC of >0.9. No variables had CCC > 0.9, DR > 0.9, and ICC > 0.9.

DATA CONCLUSION

Repeatability and reproducibility of variables is a significant limitation of radiomics analysis in 3DT1W brain MRI. Careful selection of radiomic features is required.

LEVEL OF EVIDENCE

4 TECHNICAL EFFICACY STAGE: 2.

Robustness of radiomics to variations in segmentation methods in multimodal brain MRI

Radiomics in neuroimaging uses fully automatic segmentation to delineate the anatomical areas for which radiomic features are computed. However, differences among these segmentation methods affect radiomic features to an unknown extent. A scan-rescan dataset (n = 46) of T1-weighted and diffusion tensor images was used. Subjects were split into a sleep-deprivation and a control group. Scans were segmented using four segmentation methods from which radiomic features were computed. First, we measured segmentation agreement using the Dice-coefficient. Second, robustness and reproducibility of radiomic features were measured using the intraclass correlation coefficient (ICC). Last, difference in predictive power was assessed using the Friedman-test on performance in a radiomics-based sleep deprivation classification application. Segmentation agreement was generally high (interquartile range = 0.77–0.90) and median feature robustness to segmentation method variation was higher (ICC > 0.7) than scan-rescan reproducibility (ICC 0.3–0.8). However, classification performance differed significantly among segmentation methods (p < 0.001) ranging from 77 to 84%. Accuracy was higher for more recent deep learning-based segmentation methods. Despite high agreement among segmentation methods, subtle differences significantly affected radiomic features and their predictive power. Consequently, the effect of differences in segmentation methods should be taken into account when designing and evaluating radiomics-based research methods.

Radiomic Features From Diffusion-Weighted MRI of Retroperitoneal Soft-Tissue Sarcomas Are Repeatable and Exhibit Change After Radiotherapy

Background

Size-based assessments are inaccurate indicators of tumor response in soft-tissue sarcoma (STS), motivating the requirement for new response imaging biomarkers for this rare and heterogeneous disease. In this study, we assess the test-retest repeatability of radiomic features from MR diffusion-weighted imaging (DWI) and derived maps of apparent diffusion coefficient (ADC) in retroperitoneal STS and compare baseline repeatability with changes in radiomic features following radiotherapy (RT).

Materials and Methods

Thirty patients with retroperitoneal STS received an MR examination prior to treatment, of whom 23/30 were investigated in our repeatability analysis having received repeat baseline examinations and 14/30 patients were investigated in our post-treatment analysis having received an MR examination after completing pre-operative RT. One hundred and seven radiomic features were extracted from the full manually delineated tumor region using PyRadiomics. Test-retest repeatability was assessed using an intraclass correlation coefficient (baseline ICC), and post-radiotherapy variance analysis (post-RT-IMS) was used to compare the change in radiomic feature value to baseline repeatability.

Results

For the ADC maps and DWI images, 101 and 102 features demonstrated good baseline repeatability (baseline ICC > 0.85), respectively. Forty-three and 2 features demonstrated both good baseline repeatability and a high post-RT-IMS (>0.85), respectively. Pearson correlation between the baseline ICC and post-RT-IMS was weak (0.432 and 0.133, respectively).

Conclusions

The ADC-based radiomic analysis shows better test-retest repeatability compared with features derived from DWI images in STS, and some of these features are sensitive to post-treatment change. However, good repeatability at baseline does not imply sensitivity to post-treatment change.

The stability of oncologic MRI radiomic features and the potential role of deep learning: a review

The use of MRI radiomic models for the diagnosis, prognosis and treatment response prediction of tumors has been increasingly reported in literature. However, its widespread adoption in clinics is hampered by issues related to features stability. In the MRI radiomic workflow, the main factors that affect radiomic features computation can be found in the image acquisition and reconstruction phase, in the image pre-processing steps, and in the segmentation of the region of interest on which radiomic indices are extracted. Deep Neural Networks (DNNs), having shown their potentiality in the medical image processing and analysis field, can be seen as an attractive strategy to partially overcome the issues related to radiomic stability and mitigate their impact. In fact, DNN approaches can be prospectively integrated in the MRI radiomic workflow to improve image quality, obtain accurate and reproducible segmentations and generate standardized images. In this review, DNN methods that can be included in the image processing steps of the radiomic workflow are described and discussed, in the light of a detailed analysis of the literature in the context of MRI radiomic reliability.

Relevance of apparent diffusion coefficient features for a radiomics-based prediction of response to induction chemotherapy in sinonasal cancer

In this paper, several radiomics-based predictive models of response to induction chemotherapy (IC) in sinonasal cancers (SNCs) are built and tested. Models were built as a combination of radiomic features extracted from three types of MRI images: T1-weighted images, T2-weighted images and apparent diffusion coefficient (ADC) maps. Fifty patients (aged 54 ± 12 years, 41 men) were included in this study. Patients were classified according to their response to IC (25 responders and 25 nonresponders). Not all types of images were acquired for all of the patients: 49 had T1-weighted images, 50 had T2-weighted images and 34 had ADC maps. Only in a subset of 33 patients were all three types of image acquired. Eighty-nine radiomic features were extracted from the MRI images. Dimensionality reduction was performed by using principal component analysis (PCA) and by selecting only the three main components. Different algorithms (trees ensemble, K-nearest neighbors, support vector machine, naïve Bayes) were used to classify the patients as either responders or nonresponders. Several radiomic models (either monomodality or multimodality obtained by a combination of T1-weighted, T2-weighted and ADC images) were developed and the performance was assessed through 100 iterations of train and test split. The area under the curve (AUC) of the models ranged from 0.56 to 0.78. Trees ensemble, support vector machine and naïve Bayes performed similarly, but in all cases ADC-based models performed better. Trees ensemble gave the highest AUC (0.78 for the T1-weighted+T2-weighted+ADC model) and was used for further analyses. For trees ensemble, the models based on ADC features performed better than those models that did not use those features (P < 0.02 for one-tail Hanley test, AUC range 0.68-0.78 vs 0.56-0.69) except the T1-weighted+ADC model (AUC 0.71 vs 0.69, nonsignificant differences). The results suggest the relevance of ADC-based radiomics for prediction of response to IC in SNCs.

Diffusion-weighted MRI radiomics of spine bone tumors: feature stability and machine learning-based classification performance

PURPOSE

To evaluate stability and machine learning-based classification performance of radiomic features of spine bone tumors using diffusion- and T2-weighted magnetic resonance imaging (MRI).

MATERIAL AND METHODS

This retrospective study included 101 patients with histology-proven spine bone tumor (22 benign; 38 primary malignant; 41 metastatic). All tumor volumes were manually segmented on morphologic T2-weighted sequences. The same region of interest (ROI) was used to perform radiomic analysis on ADC map. A total of 1702 radiomic features was considered. Feature stability was assessed through small geometrical transformations of the ROIs mimicking multiple manual delineations. Intraclass correlation coefficient (ICC) quantified feature stability. Feature selection consisted of stability-based (ICC > 0.75) and significance-based selections (ranking features by decreasing Mann-Whitney p-value). Class balancing was performed to oversample the minority (i.e., benign) class. Selected features were used to train and test a support vector machine (SVM) to discriminate benign from malignant spine tumors using tenfold cross-validation.

RESULTS

A total of 76.4% radiomic features were stable. The quality metrics for the SVM were evaluated as a function of the number of selected features. The radiomic model with the best performance and the lowest number of features for classifying tumor types included 8 features. The metrics were 78% sensitivity, 68% specificity, 76% accuracy and AUC 0.78.

CONCLUSION

SVM classifiers based on radiomic features extracted from T2- and diffusion-weighted imaging with ADC map are promising for classification of spine bone tumors. Radiomic features of spine bone tumors show good reproducibility rates.

MRI-Derived Radiomics Features of Hepatic Fat Predict Metabolic States in Individuals without Cardiovascular Disease

RATIONALE AND OBJECTIVES

To investigate radiomics features of hepatic fat as potential biomarkers of type 2 diabetes mellitus (T2DM) and metabolic syndrome (MetS) in individuals without overt cardiovascular disease, and benchmarking against hepatic proton density fat fraction (PDFF) and the body mass index (BMI).

MATERIALS AND METHODS

This study collected liver radiomics features of 310 individuals that were part of a case-controlled imaging substudy embedded in a prospective cohort. Individuals had known T2DM (n = 39; 12.6 %) and MetS (n = 107; 34.5 %) status, and were divided into stratified training (n = 232; 75 %) and validation (n = 78; 25 %) sets. Six hundred eighty-four MRI radiomics features were extracted for each liver volume of interest (VOI) on T₁-weighted dual-echo Dixon relative fat water content (rfwc) maps. Test-retest and inter-rater variance was simulated by additionally extracting radiomics features using noise augmented rfwc maps and deformed volume of interests. One hundred and seventy-one features with test-retest reliability (ICC(1,1)) and inter-rater agreement (ICC(3,k)) of ≥0.85 on the training set were considered stable. To construct predictive random forest (RF) models, stable features were filtered using univariate RF analysis followed by sequential forward aggregation. The predictive performance was evaluated on the independent validation set with area under the curve of the receiver operating characteristic (AUROC) and balanced accuracy (Accuracy_B).

RESULTS

On the validation set, the radiomics RF models predicted T2DM with AUROC of 0.835 and Accuracy_B of 0.822 and MetS with AUROC of 0.838 and Accuracy_B of 0.787, outperforming the RF models trained on the benchmark parameters PDFF and BMI.

CONCLUSION

Hepatic radiomics features may serve as potential imaging biomarkers for T2DM and MetS.

Radiomics feature reliability assessed by intraclass correlation coefficient: a systematic review

Radiomics research is rapidly growing in recent years, but more concerns on radiomics reliability are also raised. This review attempts to update and overview the current status of radiomics reliability research in the ever expanding medical literature from the perspective of a single reliability metric of intraclass correlation coefficient (ICC). To conduct this systematic review, Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed. After literature search and selection, a total of 481 radiomics studies using CT, PET, or MRI, covering a wide range of subject and disease types, were included for review. In these highly heterogeneous studies, feature reliability to image segmentation was much more investigated than reliability to other factors, such as image acquisition, reconstruction, post-processing, and feature quantification. The reported ICCs also suggested high radiomics feature reliability to image segmentation. Image acquisition was found to introduce much more feature variability than image segmentation, in particular for MRI, based on the reported ICC values. Image post-processing and feature quantification yielded different levels of radiomics reliability and might be used to mitigate image acquisition-induced variability. Some common flaws and pitfalls in ICC use were identified, and suggestions on better ICC use were given. Due to the extremely high study heterogeneities and possible risks of bias, the degree of radiomics feature reliability that has been achieved could not yet be safely synthesized or derived in this review. More future researches on radiomics reliability are warranted.

Radiomics feature stability of open-source software evaluated on apparent diffusion coefficient maps in head and neck cancer

Radiomics is a promising technique for discovering image based biomarkers of therapy response in cancer. Reproducibility of radiomics features is a known issue that is addressed by the image biomarker standardisation initiative (IBSI), but it remains challenging to interpret previously published radiomics signatures. This study investigates the reproducibility of radiomics features calculated with two widely used radiomics software packages (IBEX, MaZda) in comparison to an IBSI compliant software package (PyRadiomics). Intensity histogram, shape and textural features were extracted from 334 diffusion weighted magnetic resonance images of 59 head and neck cancer (HNC) patients from the PREDICT-HN observational radiotherapy study. Based on name and linear correlation, PyRadiomics shares 83 features with IBEX and 49 features with MaZda, a sub-set of well correlated features are considered reproducible (IBEX: 15 features, MaZda: 18 features). We explore the impact of including non-reproducible radiomics features in a HNC radiotherapy response model. It is possible to classify equivalent patient groups using radiomic features from either software, but only when restricting the model to reliable features using a correlation threshold method. This is relevant for clinical biomarker validation trials as it provides a framework to assess the reproducibility of reported radiomic signatures from existing trials.

Effects of Interobserver Variability on 2D and 3D CT- and MRI-Based Texture Feature Reproducibility of Cartilaginous Bone Tumors

This study aims to investigate the influence of interobserver manual segmentation variability on the reproducibility of 2D and 3D unenhanced computed tomography (CT)- and magnetic resonance imaging (MRI)-based texture analysis. Thirty patients with cartilaginous bone tumors (10 enchondromas, 10 atypical cartilaginous tumors, 10 chondrosarcomas) were retrospectively included. Three radiologists independently performed manual contour-focused segmentation on unenhanced CT and T1-weighted and T2-weighted MRI by drawing both a 2D region of interest (ROI) on the slice showing the largest tumor area and a 3D ROI including the whole tumor volume. Additionally, a marginal erosion was applied to both 2D and 3D segmentations to evaluate the influence of segmentation margins. A total of 783 and 1132 features were extracted from original and filtered 2D and 3D images, respectively. Intraclass correlation coefficient ≥ 0.75 defined feature stability. In 2D vs. 3D contour-focused segmentation, the rates of stable features were 74.71% vs. 86.57% (p < 0.001), 77.14% vs. 80.04% (p = 0.142), and 95.66% vs. 94.97% (p = 0.554) for CT and T1-weighted and T2-weighted images, respectively. Margin shrinkage did not improve 2D (p = 0.343) and performed worse than 3D (p < 0.001) contour-focused segmentation in terms of feature stability. In 2D vs. 3D contour-focused segmentation, matching stable features derived from CT and MRI were 65.8% vs. 68.7% (p = 0.191), and those derived from T1-weighted and T2-weighted images were 76.0% vs. 78.2% (p = 0.285). 2D and 3D radiomic features of cartilaginous bone tumors extracted from unenhanced CT and MRI are reproducible, although some degree of interobserver segmentation variability highlights the need for reliability analysis in future studies.

Radiomics in prostate cancer imaging for a personalized treatment approach - current aspects of methodology and a systematic review on validated studies

Prostate cancer (PCa) is one of the most frequently diagnosed malignancies of men in the world. Due to a variety of treatment options in different risk groups, proper diagnostic and risk stratification is pivotal in treatment of PCa. The development of precise medical imaging procedures simultaneously to improvements in big data analysis has led to the establishment of radiomics - a computer-based method of extracting and analyzing image features quantitatively. This approach bears the potential to assess and improve PCa detection, tissue characterization and clinical outcome prediction. This article gives an overview on the current aspects of methodology and systematically reviews available literature on radiomics in PCa patients, showing its potential for personalized therapy approaches. The qualitative synthesis includes all imaging modalities and focuses on validated studies, putting forward future directions.

A CT-Based Radiomic Signature Can Be Prognostic for 10-Months Overall Survival in Metastatic Tumors Treated with Nivolumab: An Exploratory Study

Baseline clinical prognostic factors for recurrent and/or metastatic (RM) head and neck squamous cell carcinoma (HNSCC) treated with immunotherapy are lacking. CT-based radiomics may provide additional prognostic information. A total of 85 patients with RM-HNSCC were enrolled for this study. For each tumor, radiomic features were extracted from the segmentation of the largest tumor mass. A pipeline including different feature selection steps was used to train a radiomic signature prognostic for 10-month overall survival (OS). Features were selected based on their stability to geometrical transformation of the segmentation (intraclass correlation coefficient, ICC > 0.75) and their predictive power (area under the curve, AUC > 0.7). The predictive model was developed using the least absolute shrinkage and selection operator (LASSO) in combination with the support vector machine. The model was developed based on the first 68 enrolled patients and tested on the last 17 patients. Classification performance of the radiomic risk was evaluated accuracy and the AUC. The same metrics were computed for some baseline predictors used in clinical practice (volume of largest lesion, total tumor volume, number of tumor lesions, number of affected organs, performance status). The AUC in the test set was 0.67, while accuracy was 0.82. The performance of the radiomic score was higher than the one obtainable with the clinical variables (largest lesion volume: accuracy 0.59, AUC = 0.55; number of tumoral lesions: accuracy 0.71, AUC 0.36; number of affected organs: accuracy 0.47; AUC 0.42; total tumor volume: accuracy 0.59, AUC 0.53; performance status: accuracy 0.41, AUC = 0.47). Radiomics may provide additional baseline prognostic value compared to the variables used in clinical practice.

Uncertainty measurement of radiomics features against inherent quantum noise in computed tomography imaging

OBJECTIVES

Quantum noise is a random process in X-ray-based imaging systems. We addressed and measured the uncertainty of radiomics features against this quantum noise in computed tomography (CT) images.

METHODS

A clinical multi-detector CT scanner, two homogeneous phantom sets, and four heterogeneous samples were used. A solid tumor tissue removed from a male BALB/c mouse was included. We the placed phantom sets on the CT scanning table and repeated 20 acquisitions with identical imaging settings. Regions of interest were delineated for feature extraction. Statistical quantities-average, standard deviation, and percentage uncertainty-were calculated from these 20 repeated scans. Percentage uncertainty was used to measure and quantify feature stability against quantum noise. Twelve radiomics features were measured. Random noise was added to study the robustness of machine learning classifiers against feature uncertainty.

RESULTS

We found the ranges of percentage uncertainties from homogeneous soft tissue phantoms, homogeneous bone phantoms, and solid tumor tissue to be 0.01-2138%, 0.02-15%, and 0.18-16%, respectively. Overall, it was found that the CT features ShortRunHighGrayLevelEmpha (SRHGE) (0.01-0.18%), ShortRunLowGrayLevelEmpha (SRLGE) (0.01-0.41%), LowGrayLevelRunEmpha (LGRE) (0.01-0.39%), and LongRunLowGrayLevelEmpha (LRLGE) (0.02-0.66%) were the most stable features against the inherent quantum noise. The most unstable features were cluster shade (1-2138%) and max probability (1-16%). The impact of random noise to the prediction accuracy by different machine learning classifiers was found to be between 0 and 12%.

CONCLUSIONS

Twelve features were used for uncertainty measurements. The upper and lower bounds of percentage uncertainties were determined. The quantum noise effect on machine learning classifiers is model dependent.

KEY POINTS

• Quantum noise is a random process and is intrinsic to X-ray-based imaging systems. This inherent quantum noise creates unpredictable fluctuations in the gray-level intensities of image pixels. Extra cautions and further validations are strongly recommended when unstable radiomics features are selected by a predictive model for disease classification or treatment outcome prognosis. • We addressed and used the statistical quantity of percentage uncertainty to measure the uncertainty of radiomics features against the inherent quantum noise in computed tomography (CT) images. • A clinical multi-detector CT scanner, two homogeneous phantom sets, and four heterogeneous samples were used in the stability measurement. A solid tumor tissue removed from a male BALB/c mouse was included in the heterogeneous sample.

Standardization of imaging methods for machine learning in neuro-oncology

Radiomics is a novel technique in which quantitative phenotypes or features are extracted from medical images. Machine learning enables analysis of large quantities of medical imaging data generated by radiomic feature extraction. A growing number of studies based on these methods have developed tools for neuro-oncology applications. Despite the initial promises, many of these imaging tools remain far from clinical implementation. One major limitation hindering the use of these models is their lack of reproducibility when applied across different institutions and clinical settings. In this article, we discuss the importance of standardization of methodology and reporting in our effort to improve reproducibility. Ongoing efforts of standardization for neuro-oncological imaging are reviewed. Challenges related to standardization and potential disadvantages in over-standardization are also described. Ultimately, greater multi-institutional collaborative effort is needed to provide and implement standards for data acquisition and analysis methods to facilitate research results to be interoperable and reliable for integration into different practice environments.

Repeatability of Cardiac Magnetic Resonance Radiomics: A Multi-Centre Multi-Vendor Test-Retest Study

Aims: To evaluate the repeatability of cardiac magnetic resonance (CMR) radiomics features on test-retest scanning using a multi-centre multi-vendor dataset with a varied case-mix.

Methods and Results: The sample included 54 test-retest studies from the VOLUMES resource (thevolumesresource.com). Images were segmented according to a pre-defined protocol to select three regions of interest (ROI) in end-diastole and end-systole: right ventricle, left ventricle (LV), and LV myocardium. We extracted radiomics shape features from all three ROIs and, additionally, first-order and texture features from the LV myocardium. Overall, 280 features were derived per study. For each feature, we calculated intra-class correlation coefficient (ICC), within-subject coefficient of variation, and mean relative difference. We ranked robustness of features according to mean ICC stratified by feature category, ROI, and cardiac phase, demonstrating a wide range of repeatability. There were features with good and excellent repeatability (ICC ≥ 0.75) within all feature categories and ROIs. A high proportion of first-order and texture features had excellent repeatability (ICC ≥ 0.90), however, these categories also contained features with the poorest repeatability (ICC < 0.50).

Conclusion: CMR radiomic features have a wide range of repeatability. This paper is intended as a reference for future researchers to guide selection of the most robust features for clinical CMR radiomics models. Further work in larger and richer datasets is needed to further define the technical performance and clinical utility of CMR radiomics.

Baseline MRI-Radiomics Can Predict Overall Survival in Non-Endemic EBV-Related Nasopharyngeal Carcinoma Patients

Advanced stage nasopharyngeal cancer (NPC) shows highly variable treatment outcomes, suggesting the need for independent prognostic factors. This study aims at developing a magnetic resonance imaging (MRI)-based radiomic signature as a prognostic marker for different clinical endpoints in NPC patients from non-endemic areas. A total 136 patients with advanced NPC and available MRI imaging (T1-weighted and T2-weighted) were selected. For each patient, 2144 radiomic features were extracted from the main tumor and largest lymph node. A multivariate Cox regression model was trained on a subset of features to obtain a radiomic signature for overall survival (OS), which was also applied for the prognosis of other clinical endpoints. Validation was performed using 10-fold cross-validation. The added prognostic value of the radiomic features to clinical features and volume was also evaluated. The radiomics-based signature had good prognostic power for OS and loco-regional recurrence-free survival (LRFS), with C-index of 0.68 and 0.72, respectively. In all the cases, the addition of radiomics to clinical features improved the prognostic performance. Radiomic features can provide independent prognostic information in NPC patients from non-endemic areas.

Methodology and technology for the development of a prognostic MRI-based radiomic model for the outcome of head and neck cancer patients

The purpose of this study was to establish a methodology and technology for the development of an MRI-based radiomic signature for prognosis of overall survival (OS) in nasopharyngeal cancer from non-endemic areas. The signature was trained using 1072 features extracted from the main tumor in T1-weighted and T2-weighted images of 142 patients. A model with 2 radiomic features was obtained (RAD model). Tumor volume and a signature obtained by training the model on permuted survival data (RADperm model) were used as a reference. A 10-fold cross-validation was used to validate the signature. Harrel's C-index was used as performance metric. A statistical comparison of the RAD, RADperm and volume was performed using Wilcoxon signed rank tests. The C-index for the RAD model was higher compared to the one of the RADperm model (0.69±0.08 vs 0.47±0.05), which ensures absence of overfitting. Also, the signature obtained with the RAD model had an improved C-index compared to tumor volume alone (0.69±0.08 vs 0.65±0.06), suggesting that the radiomic signature provides additional prognostic information.

Radiomics in diffusion data: a test-retest, inter- and intra-reader DWI phantom study

AIM

The aim of this study was to evaluate the robustness of radiomics features of a MRI (magnetic resonance imaging) phantom in quantitative diffusion-weighted imaging (DWI) and depending on the image resolution.

MATERIALS AND METHODS

Scanning of an in-house developed DWI phantom was performed at a 1.5 T MRI scanner (Magnetom AERA, Siemens, Erlangen, Germany) using an echo planar imaging (EPI) DWI sequence (b=0,500,1,000 s/mm²) with low (3×3 mm²) and high (2×2 mm²) image resolutions. Scans were repeated after phantom repositioning to evaluate retest reliability. Radiomics features were extracted after semi-automatic segmentation and standardised pre-processing. Intra-/interobserver reproducibility and test-retest robustness were assessed using intraclass correlation coefficients (ICC). Differences were tested with non-parametric Wilcoxon's signed-rank and Friedman's test (p < 0.05) with Dunn's post-hoc analysis.

RESULTS

Test-retest ICC was overall high with >0.90 for 39/46 radiomics features in all sequences/resolutions. Decreased test-retest ICCs were pronounced for conventional Min-value (overall ICC=0.817), and grey-level zone length matrix (GLZLM) features Short-Zone Emphasis (SZE) and Short-Zone Low Grey-level Emphasis (SZLGE) (for both overall ICC=0.927). Test-retest reproducibility was significantly different between b=500, 1,000 and apparent diffusion coefficient (ADC) (mean 0.975±0.050, 0.974±0.051 and 0.966±0.063), which remained significant after post-hoc analysis between b=1,000 and ADC (p = 0.022). ICCs were not significantly different between resolutions of 2×2 and 3×3 mm² regarding b=500 (mean: 0.977±0.052 and 0.974±0.049, p = 0.612), b=1,000 (mean: 0.973±0.059 and 0.974±0.054, p = 0.516), and ADC (mean: 0.972±0.049 and 0.955±0.101, p = 0.851). Inter- and intra-observer reliability was consistently high for all sequences (overall mean 0.992±0.021 and 0.990±0.028).

CONCLUSION

Under ex-vivo conditions, DWI provided robust radiomics features with those from ADC being slightly less robust than from raw DWI (b=500, 1,000 s/mm²). No significant difference was detected for different resolutions. Although, ex-vivo reliability of DWI radiomics features was high, no implications can be made regarding in-vivo analyses.

Reliability of CT radiomic features reflecting tumour heterogeneity according to image quality and image processing parameters

The reliability of radiomics features (RFs) is crucial for quantifying tumour heterogeneity. We assessed the influence of imaging, segmentation, and processing conditions (quantization range, bin number, signal-to-noise ratio [SNR], and unintended outliers) on RF measurement. Low SNR and unintended outliers increased the standard deviation and mean values of histograms to calculate the first-order RFs. Variations in imaging processing conditions significantly altered the shape of the probability distribution (centre of distribution, extent of dispersion, and segmentation of probability clusters) in second-order RF matrices (i.e. grey-level co-occurrence and grey-level run length), thereby eventually causing fluctuations in RF estimation. Inconsistent imaging and processing conditions decreased the number of reliably measured RFs in terms of individual RF values (intraclass correlation coefficient ≥0.75) and inter-lesion RF ratios (coefficient of variation <15%). No RF could be reliably estimated under inconsistent SNR and inclusion of outlier conditions. By contrast, with high SNR and no outliers, all first-order RFs, 11 (42%) grey-level co-occurrence RFs and five (42%) grey-level run length RFs showed acceptable reliability. Our study suggests that optimization of SNR, exclusion of outliers, and application of relevant quantization range and bin number should be performed to ensure the robustness of radiomics studies assessing tumor heterogeneity.

Assessment of the effect of intensity standardization on the reliability of T1-weighted MRI radiomic features: experiment on a virtual phantom

The effect of time of repetition (TR) and time of echo (TE) on radiomic features was evaluated using a virtual phantom. Forty-two T1-weighted MRI images of the same virtual phantom were simulated with TR and TE in a range used in clinical practice. Fifty-eight radiomic features were considered for this analysis. Features were extracted from 3 different regions of interest (ROIs) from the original images and from images that underwent intensity standardization (linear intensity standardization, Z-score standardization and histogram matching). Intraclass correlation coefficient (ICC) was used to assess the reliability of the radiomic features and a threshold of 0.75 was used to discriminate features with good or bad reliability. The coefficient of determination R² was used to quantify correlation between features and image acquisition parameters. The majority of radiomic features (76%) had good reliability (ICC>0.75) and 66% of the features were uncorrelated with TR and TE (R²<; 0.5). Intensity standardization (in particular histogram matching) significantly reduced the correlation. Intensity standardization also increased the reliability of FOS features, but histogram matching significantly reduced the reliability of GLCM features.

Robustness of radiomic features in magnetic resonance imaging: review and a phantom study

Radiomic analysis has exponentially increased the amount of quantitative data that can be extracted from a single image. These imaging biomarkers can aid in the generation of prediction models aimed to further personalized medicine. However, the generalizability of the model is dependent on the robustness of these features. The purpose of this study is to review the current literature regarding robustness of radiomic features on magnetic resonance imaging. Additionally, a phantom study is performed to systematically evaluate the behavior of radiomic features under various conditions (signal to noise ratio, region of interest delineation, voxel size change and normalization methods) using intraclass correlation coefficients. The features extracted in this phantom study include first order, shape, gray level cooccurrence matrix and gray level run length matrix. Many features are found to be non-robust to changing parameters. Feature robustness assessment prior to feature selection, especially in the case of combining multi-institutional data, may be warranted. Further investigation is needed in this area of research.

Repeatability of Multiparametric Prostate MRI Radiomics Features

In this study we assessed the repeatability of radiomics features on small prostate tumors using test-retest Multiparametric Magnetic Resonance Imaging (mpMRI). The premise of radiomics is that quantitative image-based features can serve as biomarkers for detecting and characterizing disease. For such biomarkers to be useful, repeatability is a basic requirement, meaning its value must remain stable between two scans, if the conditions remain stable. We investigated repeatability of radiomics features under various preprocessing and extraction configurations including various image normalization schemes, different image pre-filtering, and different bin widths for image discretization. Although we found many radiomics features and preprocessing combinations with high repeatability (Intraclass Correlation Coefficient > 0.85), our results indicate that overall the repeatability is highly sensitive to the processing parameters. Neither image normalization, using a variety of approaches, nor the use of pre-filtering options resulted in consistent improvements in repeatability. We urge caution when interpreting radiomics features and advise paying close attention to the processing configuration details of reported results. Furthermore, we advocate reporting all processing details in radiomics studies and strongly recommend the use of open source implementations.

NCTN Assessment on Current Applications of Radiomics in Oncology

Radiomics is a fast-growing research area based on converting standard-of-care imaging into quantitative minable data and building subsequent predictive models to personalize treatment. Radiomics has been proposed as a study objective in clinical trial concepts and a potential biomarker for stratifying patients across interventional treatment arms. In recognizing the growing importance of radiomics in oncology, a group of medical physicists and clinicians from NRG Oncology reviewed the current status of the field and identified critical issues, providing a general assessment and early recommendations for incorporation in oncology studies.

Stability of radiomic features of apparent diffusion coefficient (ADC) maps for locally advanced rectal cancer in response to image pre-processing

Quantitative imaging features (radiomics) extracted from apparent diffusion coefficient (ADC) maps of rectal cancer patients can provide additional information to support treatment decision. Most available radiomic computational packages allow extraction of hundreds to thousands of features. However, two major factors can influence the reproducibility of radiomic features: interobserver variability, and imaging filtering applied prior to features extraction. In this exploratory study we seek to determine to what extent various commonly-used features are reproducible with regards to the mentioned factors using ADC maps from two different clinics (56 patients). Features derived from intensity distribution histograms are less sensitive to manual tumour delineation differences, noise in ADC images, pixel size resampling and intensity discretization. Shape features appear to be strongly affected by delineation quality. On the whole, textural features appear to be poorly or moderately reproducible with respect to the image pre-processing perturbations we reproduced.

Delta-radiomics signature predicts treatment outcomes after preoperative chemoradiotherapy and surgery in rectal cancer

Background

To develop and compare delta-radiomics signatures from 2- (2D) and 3-dimensional (3D) features that predict treatment outcomes following preoperative chemoradiotherapy (CCRT) and surgery for locally advanced rectal cancer.

Methods

In total, 101 patients (training cohort, n = 67; validation cohort, n = 34) with locally advanced rectal adenocarcinoma between 2008 and 2015 were included. We extracted 55 features from T2-weighted magnetic resonance imaging (MRI) scans. Delta-radiomics feature was defined as the difference in radiomics feature before and after CCRT. Signatures were developed to predict local recurrence (LR), distant metastasis (DM), and disease-free survival (DFS) from 2D and 3D features. The least absolute shrinkage and selection operator regression was used to select features and build signatures. The delta-radiomics signatures and clinical factors were integrated into Cox regression analysis to determine if the signatures were independent prognostic factors.

Results

The radiomics signatures for LR, DM, and DFS were developed and validated using both 2D and 3D features. Outcomes were significantly different in the low- and high-risk patients dichotomized by optimal cutoff in both the training and validation cohorts. In multivariate analysis, the signatures were independent prognostic factors even when considering the clinical parameters. There were no significant differences in C-index from 2D vs. 3D signatures.

Conclusions

This is the first study to develop delta-radiomics signatures for rectal cancer. The signatures successfully predicted the outcomes and were independent prognostic factors. External validation is warranted to ensure their performance.

Electronic supplementary material

The online version of this article (10.1186/s13014-019-1246-8) contains supplementary material, which is available to authorized users.

Potentials of radiomics for cancer diagnosis and treatment in comparison with computer-aided diagnosis

Computer-aided diagnosis (CAD) is a field that is essentially based on pattern recognition that improves the accuracy of a diagnosis made by a physician who takes into account the computer's "opinion" derived from the quantitative analysis of radiological images. Radiomics is a field based on data science that massively and comprehensively analyzes a large number of medical images to extract a large number of phenotypic features reflecting disease traits, and explores the associations between the features and patients' prognoses for precision medicine. According to the definitions for both, you may think that radiomics is not a paraphrase of CAD, but you may also think that these definitions are "image manipulation". However, there are common and different features between the two fields. This review paper elaborates on these common and different features and introduces the potential of radiomics for cancer diagnosis and treatment by comparing it with CAD.