The effect of harmonization on the variability of PET radiomic features extracted using various segmentation methods

PURPOSE

This study aimed to examine the robustness of positron emission tomography (PET) radiomic features extracted via different segmentation methods before and after ComBat harmonization in patients with non-small cell lung cancer (NSCLC).

METHODS

We included 120 patients (positive recurrence = 46 and negative recurrence = 74) referred for PET scanning as a routine part of their care. All patients had a biopsy-proven NSCLC. Nine segmentation methods were applied to each image, including manual delineation, K-means (KM), watershed, fuzzy-C-mean, region-growing, local active contour (LAC), and iterative thresholding (IT) with 40, 45, and 50% thresholds. Diverse image discretizations, both without a filter and with different wavelet decompositions, were applied to PET images. Overall, 6741 radiomic features were extracted from each image (749 radiomic features from each segmented area). Non-parametric empirical Bayes (NPEB) ComBat harmonization was used to harmonize the features. Linear Support Vector Classifier (LinearSVC) with L1 regularization For feature selection and Support Vector Machine classifier (SVM) with fivefold nested cross-validation was performed using StratifiedKFold with 'n_splits' set to 5 to predict recurrence in NSCLC patients and assess the impact of ComBat harmonization on the outcome.

RESULTS

From 749 extracted radiomic features, 206 (27%) and 389 (51%) features showed excellent reliability (ICC ≥ 0.90) against segmentation method variation before and after NPEB ComBat harmonization, respectively. Among all, 39 features demonstrated poor reliability, which declined to 10 after ComBat harmonization. The 64 fixed bin widths (without any filter) and wavelets (LLL)-based radiomic features set achieved the best performance in terms of robustness against diverse segmentation techniques before and after ComBat harmonization. The first-order and GLRLM and also first-order and NGTDM feature families showed the largest number of robust features before and after ComBat harmonization, respectively. In terms of predicting recurrence in NSCLC, our findings indicate that using ComBat harmonization can significantly enhance machine learning outcomes, particularly improving the accuracy of watershed segmentation, which initially had fewer reliable features than manual contouring. Following the application of ComBat harmonization, the majority of cases saw substantial increase in sensitivity and specificity.

CONCLUSION

Radiomic features are vulnerable to different segmentation methods. ComBat harmonization might be considered a solution to overcome the poor reliability of radiomic features.

Differential privacy preserved federated learning for prognostic modeling in COVID-19 patients using large multi-institutional chest CT dataset

BACKGROUND

Notwithstanding the encouraging results of previous studies reporting on the efficiency of deep learning (DL) in COVID-19 prognostication, clinical adoption of the developed methodology still needs to be improved. To overcome this limitation, we set out to predict the prognosis of a large multi-institutional cohort of patients with COVID-19 using a DL-based model.

PURPOSE

This study aimed to evaluate the performance of deep privacy-preserving federated learning (DPFL) in predicting COVID-19 outcomes using chest CT images.

METHODS

After applying inclusion and exclusion criteria, 3055 patients from 19 centers, including 1599 alive and 1456 deceased, were enrolled in this study. Data from all centers were split (randomly with stratification respective to each center and class) into a training/validation set (70%/10%) and a hold-out test set (20%). For the DL model, feature extraction was performed on 2D slices, and averaging was performed at the final layer to construct a 3D model for each scan. The DensNet model was used for feature extraction. The model was developed using centralized and FL approaches. For FL, we employed DPFL approaches. Membership inference attack was also evaluated in the FL strategy. For model evaluation, different metrics were reported in the hold-out test sets. In addition, models trained in two scenarios, centralized and FL, were compared using the DeLong test for statistical differences.

RESULTS

The centralized model achieved an accuracy of 0.76, while the DPFL model had an accuracy of 0.75. Both the centralized and DPFL models achieved a specificity of 0.77. The centralized model achieved a sensitivity of 0.74, while the DPFL model had a sensitivity of 0.73. A mean AUC of 0.82 and 0.81 with 95% confidence intervals of (95% CI: 0.79-0.85) and (95% CI: 0.77-0.84) were achieved by the centralized model and the DPFL model, respectively. The DeLong test did not prove statistically significant differences between the two models (p-value = 0.98). The AUC values for the inference attacks fluctuate between 0.49 and 0.51, with an average of 0.50 ± 0.003 and 95% CI for the mean AUC of 0.500 to 0.501.

CONCLUSION

The performance of the proposed model was comparable to centralized models while operating on large and heterogeneous multi-institutional datasets. In addition, the model was resistant to inference attacks, ensuring the privacy of shared data during the training process.

Development and validation of survival prognostic models for head and neck cancer patients using machine learning and dosiomics and CT radiomics features: a multicentric study

BACKGROUND

This study aimed to investigate the value of clinical, radiomic features extracted from gross tumor volumes (GTVs) delineated on CT images, dose distributions (Dosiomics), and fusion of CT and dose distributions to predict outcomes in head and neck cancer (HNC) patients.

METHODS

A cohort of 240 HNC patients from five different centers was obtained from The Cancer Imaging Archive. Seven strategies, including four non-fusion (Clinical, CT, Dose, DualCT-Dose), and three fusion algorithms (latent low-rank representation referred (LLRR),Wavelet, weighted least square (WLS)) were applied. The fusion algorithms were used to fuse the pre-treatment CT images and 3-dimensional dose maps. Overall, 215 radiomics and Dosiomics features were extracted from the GTVs, alongside with seven clinical features incorporated. Five feature selection (FS) methods in combination with six machine learning (ML) models were implemented. The performance of the models was quantified using the concordance index (CI) in one-center-leave-out 5-fold cross-validation for overall survival (OS) prediction considering the time-to-event.

RESULTS

The mean CI and Kaplan-Meier curves were used for further comparisons. The CoxBoost ML model using the Minimal Depth (MD) FS method and the glmnet model using the Variable hunting (VH) FS method showed the best performance with CI = 0.73 ± 0.15 for features extracted from LLRR fused images. In addition, both glmnet-Cindex and Coxph-Cindex classifiers achieved a CI of 0.72 ± 0.14 by employing the dose images (+ incorporated clinical features) only.

CONCLUSION

Our results demonstrated that clinical features, Dosiomics and fusion of dose and CT images by specific ML-FS models could predict the overall survival of HNC patients with acceptable accuracy. Besides, the performance of ML methods among the three different strategies was almost comparable.

Prevalence of electrocardiographic abnormalities among Iranian children and adolescents and associations with blood pressure and obesity: findings from the SHED LIGHT study

BACKGROUND

There are few studies for detecting rhythm abnormalities among healthy children and adolescents. The aim of the study was to investigate the prevalence of abnormal electrocardiographic findings in the young Iranian population and its association with blood pressure and obesity.

METHODS

A total of 15084 children and adolescents were examined in a randomly selected population of Tehran city, Iran, between October 2017 and December 2018. Anthropometric values and blood pressure measurements were also assessed. A standard 12-lead electrocardiogram was recorded by a unique recorder, and those were examined by electrophysiologists.

RESULTS

All students mean age was 12.3 ± 3.1 years (6-18 years), and 52% were boys. A total of 2900 students (192.2/1000 persons; 95% confidence interval 186-198.6) had electrocardiographic abnormalities. The rate of electrocardiographic abnormalities was higher in boys than girls (p < 0.001). Electrocardiographic abnormalities were significantly higher in thin than obese students (p < 0.001), and there was a trend towards hypertensive individuals to have more electrocardiographic abnormalities compared to normotensive individuals (p = 0.063). Based on the multivariable analysis, individuals with electrocardiographic abnormalities were less likely to be girls (odds ratio 0.745, 95% confidence interval 0.682-0.814) and had a lower body mass index (odds ratio 0.961, 95% confidence interval 0.944-0.979).

CONCLUSIONS

In this large-scale study, there was a high prevalence of electrocardiographic abnormalities among young population. In addition, electrocardiographic findings were significantly influenced by increasing age, sex, obesity, and blood pressure levels. This community-based study revealed the implications of electrocardiographic screening to improve the care delivery by early detection.

Information fusion for fully automated segmentation of head and neck tumors from PET and CT images

BACKGROUND

PET/CT images combining anatomic and metabolic data provide complementary information that can improve clinical task performance. PET image segmentation algorithms exploiting the multi-modal information available are still lacking.

PURPOSE

Our study aimed to assess the performance of PET and CT image fusion for gross tumor volume (GTV) segmentations of head and neck cancers (HNCs) utilizing conventional, deep learning (DL), and output-level voting-based fusions.

METHODS

The current study is based on a total of 328 histologically confirmed HNCs from six different centers. The images were automatically cropped to a 200 × 200 head and neck region box, and CT and PET images were normalized for further processing. Eighteen conventional image-level fusions were implemented. In addition, a modified U2-Net architecture as DL fusion model baseline was used. Three different input, layer, and decision-level information fusions were used. Simultaneous truth and performance level estimation (STAPLE) and majority voting to merge different segmentation outputs (from PET and image-level and network-level fusions), that is, output-level information fusion (voting-based fusions) were employed. Different networks were trained in a 2D manner with a batch size of 64. Twenty percent of the dataset with stratification concerning the centers (20% in each center) were used for final result reporting. Different standard segmentation metrics and conventional PET metrics, such as SUV, were calculated.

RESULTS

In single modalities, PET had a reasonable performance with a Dice score of 0.77 ± 0.09, while CT did not perform acceptably and reached a Dice score of only 0.38 ± 0.22. Conventional fusion algorithms obtained a Dice score range of [0.76-0.81] with guided-filter-based context enhancement (GFCE) at the low-end, and anisotropic diffusion and Karhunen-Loeve transform fusion (ADF), multi-resolution singular value decomposition (MSVD), and multi-level image decomposition based on latent low-rank representation (MDLatLRR) at the high-end. All DL fusion models achieved Dice scores of 0.80. Output-level voting-based models outperformed all other models, achieving superior results with a Dice score of 0.84 for Majority_ImgFus, Majority_All, and Majority_Fast. A mean error of almost zero was achieved for all fusions using SUVpeak , SUVmean and SUVmedian .

CONCLUSION

PET/CT information fusion adds significant value to segmentation tasks, considerably outperforming PET-only and CT-only methods. In addition, both conventional image-level and DL fusions achieve competitive results. Meanwhile, output-level voting-based fusion using majority voting of several algorithms results in statistically significant improvements in the segmentation of HNC.

Differential privacy preserved federated transfer learning for multi-institutional 68Ga-PET image artefact detection and disentanglement

PURPOSE

Image artefacts continue to pose challenges in clinical molecular imaging, resulting in misdiagnoses, additional radiation doses to patients and financial costs. Mismatch and halo artefacts occur frequently in gallium-68 (68Ga)-labelled compounds whole-body PET/CT imaging. Correcting for these artefacts is not straightforward and requires algorithmic developments, given that conventional techniques have failed to address them adequately. In the current study, we employed differential privacy-preserving federated transfer learning (FTL) to manage clinical data sharing and tackle privacy issues for building centre-specific models that detect and correct artefacts present in PET images.

METHODS

Altogether, 1413 patients with 68Ga prostate-specific membrane antigen (PSMA)/DOTA-TATE (TOC) PET/CT scans from 3 countries, including 8 different centres, were enrolled in this study. CT-based attenuation and scatter correction (CT-ASC) was used in all centres for quantitative PET reconstruction. Prior to model training, an experienced nuclear medicine physician reviewed all images to ensure the use of high-quality, artefact-free PET images (421 patients' images). A deep neural network (modified U2Net) was trained on 80% of the artefact-free PET images to utilize centre-based (CeBa), centralized (CeZe) and the proposed differential privacy FTL frameworks. Quantitative analysis was performed in 20% of the clean data (with no artefacts) in each centre. A panel of two nuclear medicine physicians conducted qualitative assessment of image quality, diagnostic confidence and image artefacts in 128 patients with artefacts (256 images for CT-ASC and FTL-ASC).

RESULTS

The three approaches investigated in this study for 68Ga-PET imaging (CeBa, CeZe and FTL) resulted in a mean absolute error (MAE) of 0.42 ± 0.21 (CI 95%: 0.38 to 0.47), 0.32 ± 0.23 (CI 95%: 0.27 to 0.37) and 0.28 ± 0.15 (CI 95%: 0.25 to 0.31), respectively. Statistical analysis using the Wilcoxon test revealed significant differences between the three approaches, with FTL outperforming CeBa and CeZe (p-value < 0.05) in the clean test set. The qualitative assessment demonstrated that FTL-ASC significantly improved image quality and diagnostic confidence and decreased image artefacts, compared to CT-ASC in 68Ga-PET imaging. In addition, mismatch and halo artefacts were successfully detected and disentangled in the chest, abdomen and pelvic regions in 68Ga-PET imaging.

CONCLUSION

The proposed approach benefits from using large datasets from multiple centres while preserving patient privacy. Qualitative assessment by nuclear medicine physicians showed that the proposed model correctly addressed two main challenging artefacts in 68Ga-PET imaging. This technique could be integrated in the clinic for 68Ga-PET imaging artefact detection and disentanglement using multicentric heterogeneous datasets.

Left Ventricular Myocardial Dysfunction Evaluation in Thalassemia Patients Using Echocardiographic Radiomic Features and Machine Learning Algorithms

Heart failure caused by iron deposits in the myocardium is the primary cause of mortality in beta-thalassemia major patients. Cardiac magnetic resonance imaging (CMRI) T2* is the primary screening technique used to detect myocardial iron overload, but inherently bears some limitations. In this study, we aimed to differentiate beta-thalassemia major patients with myocardial iron overload from those without myocardial iron overload (detected by T2*CMRI) based on radiomic features extracted from echocardiography images and machine learning (ML) in patients with normal left ventricular ejection fraction (LVEF > 55%) in echocardiography. Out of 91 cases, 44 patients with thalassemia major with normal LVEF (> 55%) and T2* ≤ 20 ms and 47 people with LVEF > 55% and T2* > 20 ms as the control group were included in the study. Radiomic features were extracted for each end-systolic (ES) and end-diastolic (ED) image. Then, three feature selection (FS) methods and six different classifiers were used. The models were evaluated using various metrics, including the area under the ROC curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE). Maximum relevance-minimum redundancy-eXtreme gradient boosting (MRMR-XGB) (AUC = 0.73, ACC = 0.73, SPE = 0.73, SEN = 0.73), ANOVA-MLP (AUC = 0.69, ACC = 0.69, SPE = 0.56, SEN = 0.83), and recursive feature elimination-K-nearest neighbors (RFE-KNN) (AUC = 0.65, ACC = 0.65, SPE = 0.64, SEN = 0.65) were the best models in ED, ES, and ED&ES datasets. Using radiomic features extracted from echocardiographic images and ML, it is feasible to predict cardiac problems caused by iron overload.

Comparison of Machine Learning Algorithms Using Manual/Automated Features on 12-Lead Signal Electrocardiogram Classification: A Large Cohort Study on Students Aged Between 6 to 18 Years Old

PROPOSE

An electrocardiogram (ECG) has been extensively used to detect rhythm disturbances. We sought to determine the accuracy of different machine learning in distinguishing abnormal ECGs from normal ones in children who were examined using a resting 12-Lead ECG machine, and we also compared the manual and automated measurement using the modular ECG Analysis System (MEANS) algorithm of ECG features.

METHODS

Altogether, 10745 ECGs were recorded for students aged 6 to 18. Manual and automatic ECG features were extracted for each participant. Features were normalized using Z-score normalization and went through the student's t-test and chi-squared test to measure their relevance. We applied the Boruta algorithm for feature selection and then implemented eight classifier algorithms. The dataset was split into training (80%) and test (20%) partitions. The performance of the classifiers was evaluated on the test data (unseen data) by 1000 bootstrap, and sensitivity (SEN), specificity (SPE), AUC, and accuracy (ACC) were reported.

RESULTS

In univariate analysis, the highest performance was heart rate and RR interval in the manual dataset and heart rate in an automated dataset with AUC of 0.72 and 0.71, respectively. The best classifiers in the manual dataset were random forest (RF) and quadratic-discriminant-analysis (QDA) with AUC, ACC, SEN, and SPE equal to 0.93, 0.98, 0.69, 0.99, and 0.90, 0.95, 0.75, 0.96, respectively. In the automated dataset, QDA (AUC: 0.89, ACC:0.92, SEN:0.71, SPE:0.93) and stack learning (SL) (AUC:0.89, ACC:0.96, SEN:0.61, SPE:0.99) reached best performances.

CONCLUSION

This study demonstrated that the manual measurement of 12-Lead ECG features had better performance than the automated measurement (MEANS algorithm), but some classifiers had promising results in discriminating between normal and abnormal cases. Further studies can help us evaluate the applicability and efficacy of machine-learning approaches for distinguishing abnormal ECGs in community-based investigations in both adults and children.

Time-to-event overall survival prediction in glioblastoma multiforme patients using magnetic resonance imaging radiomics

PURPOSE

Glioblastoma Multiforme (GBM) represents the predominant aggressive primary tumor of the brain with short overall survival (OS) time. We aim to assess the potential of radiomic features in predicting the time-to-event OS of patients with GBM using machine learning (ML) algorithms.

MATERIALS AND METHODS

One hundred nineteen patients with GBM, who had T1-weighted contrast-enhanced and T2-FLAIR MRI sequences, along with clinical data and survival time, were enrolled. Image preprocessing methods included 64 bin discretization, Laplacian of Gaussian (LOG) filters with three Sigma values and eight variations of Wavelet Transform. Images were then segmented, followed by the extraction of 1212 radiomic features. Seven feature selection (FS) methods and six time-to-event ML algorithms were utilized. The combination of preprocessing, FS, and ML algorithms (12 × 7 × 6 = 504 models) was evaluated by multivariate analysis.

RESULTS

Our multivariate analysis showed that the best prognostic FS/ML combinations are the Mutual Information (MI)/Cox Boost, MI/Generalized Linear Model Boosting (GLMB) and MI/Generalized Linear Model Network (GLMN), all of which were done via the LOG (Sigma = 1 mm) preprocessing method (C-index = 0.77). The LOG filter with Sigma = 1 mm preprocessing method, MI, GLMB and GLMN achieved significantly higher C-indices than other preprocessing, FS, and ML methods (all p values < 0.05, mean C-indices of 0.65, 0.70, and 0.64, respectively).

CONCLUSION

ML algorithms are capable of predicting the time-to-event OS of patients using MRI-based radiomic and clinical features. MRI-based radiomics analysis in combination with clinical variables might appear promising in assisting clinicians in the survival prediction of patients with GBM. Further research is needed to establish the applicability of radiomics in the management of GBM in the clinic.

A machine learning-based sonomics for prediction of thyroid nodule malignancies

OBJECTIVES

This study aims to use ultrasound derived features as biomarkers to assess the malignancy of thyroid nodules in patients who were candidates for FNA according to the ACR TI-RADS guidelines.

METHODS

Two hundred and ten patients who met the selection criteria were enrolled in the study and subjected to ultrasound-guided FNA of thyroid nodules. Different radiomics features were extracted from sonographic images, including intensity, shape, and texture feature sets. Least Absolute Shrinkage and Selection Operator (LASSO), Minimum Redundancy Maximum Relevance (MRMR), and Random Forests/Extreme Gradient Boosting Machine (XGBoost) algorithms were used for feature selection and classification of the univariate and multivariate modeling, respectively. Evaluation of models performed using accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).

RESULTS

In the univariate analysis, Gray Level Run Length Matrix - Run-Length Non-Uniformity (GLRLM-RLNU) and gray-level zone length matrix - Run-Length Non-Uniformity (GLZLM-GLNU) (both with an AUC of 0.67) were top-performing for predicting nodules malignancy. In the multivariate analysis of the training dataset, the AUC of all combinations of feature selection algorithms and classifiers was 0.99, and the highest sensitivity was for XGBoost classifier and MRMR feature selection algorithms (0.99). Finally, the test dataset was used to evaluate our model in which XGBoost classifier with MRMR and LASSO feature selection algorithms had the highest performance (AUC = 0.95).

CONCLUSIONS

Ultrasound-extracted features can be used as non-invasive biomarkers for thyroid nodules' malignancy prediction.

Dual-Centre Harmonised Multimodal Positron Emission Tomography/Computed Tomography Image Radiomic Features and Machine Learning Algorithms for Non-small Cell Lung Cancer Histopathological Subtype Phenotype Decoding

AIMS

We aimed to build radiomic models for classifying non-small cell lung cancer (NSCLC) histopathological subtypes through a dual-centre dataset and comprehensively evaluate the effect of ComBat harmonisation on the performance of single- and multimodality radiomic models.

MATERIALS AND METHODS

A public dataset of NSCLC patients from two independent centres was used. Two image fusion methods, namely guided filtering-based fusion and image fusion based on visual saliency map and weighted least square optimisation, were used. Radiomic features were extracted from each scan, including first-order, texture and moment-invariant features. Subsequently, ComBat harmonisation was applied to the extracted features from computed tomography (CT), positron emission tomography (PET) and fused images to correct the centre effect. For feature selection, least absolute shrinkage and selection operator (Lasso) and recursive feature elimination (RFE) were investigated. For machine learning, logistic regression (LR), support vector machine (SVM) and AdaBoost were evaluated for classifying NSCLC subtypes. Training and evaluation of the models were carried out in a robust framework to offset plausible errors and performance was reported using area under the curve, balanced accuracy, sensitivity and specificity before and after harmonisation. N-way ANOVA was used to assess the effect of different factors on the performance of the models.

RESULTS

Support vector machine fed with selected features by recursive feature elimination from a harmonised PET feature set achieved the highest performance (area under the curve = 0.82) in classifying NSCLC histopathological subtypes. Although the performance of the models did not significantly improve for CT images after harmonisation, the performance of PET and guided filtering-based fusion feature signatures significantly improved for almost all models. Although the selection of the image modality and feature selection methods was effective on the performance of the model (ANOVA P-values <0.001), machine learning and harmonisation did not change the performance significantly (ANOVA P-values = 0.839 and 0.292, respectively).

CONCLUSION

This study confirmed the potential of radiomic analysis on PET, CT and hybrid images for histopathological classification of NSCLC subtypes.

Machine learning based readmission and mortality prediction in heart failure patients

This study intends to predict in-hospital and 6-month mortality, as well as 30-day and 90-day hospital readmission, using Machine Learning (ML) approach via conventional features. A total of 737 patients remained after applying the exclusion criteria to 1101 heart failure patients. Thirty-four conventional features were collected for each patient. First, the data were divided into train and test cohorts with a 70-30% ratio. Then train data were normalized using the Z-score method, and its mean and standard deviation were applied to the test data. Subsequently, Boruta, RFE, and MRMR feature selection methods were utilized to select more important features in the training set. In the next step, eight ML approaches were used for modeling. Next, hyperparameters were optimized using tenfold cross-validation and grid search in the train dataset. All model development steps (normalization, feature selection, and hyperparameter optimization) were performed on a train set without touching the hold-out test set. Then, bootstrapping was done 1000 times on the hold-out test data. Finally, the obtained results were evaluated using four metrics: area under the ROC curve (AUC), accuracy (ACC), specificity (SPE), and sensitivity (SEN). The RFE-LR (AUC: 0.91, ACC: 0.84, SPE: 0.84, SEN: 0.83) and Boruta-LR (AUC: 0.90, ACC: 0.85, SPE: 0.85, SEN: 0.83) models generated the best results in terms of in-hospital mortality. In terms of 30-day rehospitalization, Boruta-SVM (AUC: 0.73, ACC: 0.81, SPE: 0.85, SEN: 0.50) and MRMR-LR (AUC: 0.71, ACC: 0.68, SPE: 0.69, SEN: 0.63) models performed the best. The best model for 3-month rehospitalization was MRMR-KNN (AUC: 0.60, ACC: 0.63, SPE: 0.66, SEN: 0.53) and regarding 6-month mortality, the MRMR-LR (AUC: 0.61, ACC: 0.63, SPE: 0.44, SEN: 0.66) and MRMR-NB (AUC: 0.59, ACC: 0.61, SPE: 0.48, SEN: 0.63) models outperformed the others. Reliable models were developed in 30-day rehospitalization and in-hospital mortality using conventional features and ML techniques. Such models can effectively personalize treatment, decision-making, and wiser budget allocation. Obtained results in 3-month rehospitalization and 6-month mortality endpoints were not astonishing and further experiments with additional information are needed to fetch promising results in these endpoints.

Machine learning-based diagnosis and risk classification of coronary artery disease using myocardial perfusion imaging SPECT: A radiomics study

This study aimed to investigate the diagnostic performance of machine learning-based radiomics analysis to diagnose coronary artery disease status and risk from rest/stress Myocardial Perfusion Imaging (MPI) single-photon emission computed tomography (SPECT). A total of 395 patients suspicious of coronary artery disease who underwent 2-day stress-rest protocol MPI SPECT were enrolled in this study. The left ventricle myocardium, excluding the cardiac cavity, was manually delineated on rest and stress images to define a volume of interest. Added to clinical features (age, sex, family history, diabetes status, smoking, and ejection fraction), a total of 118 radiomics features, were extracted from rest and stress MPI SPECT images to establish different feature sets, including Rest-, Stress-, Delta-, and Combined-radiomics (all together) feature sets. The data were randomly divided into 80% and 20% subsets for training and testing, respectively. The performance of classifiers built from combinations of three feature selections, and nine machine learning algorithms was evaluated for two different diagnostic tasks, including 1) normal/abnormal (no CAD vs. CAD) classification, and 2) low-risk/high-risk CAD classification. Different metrics, including the area under the ROC curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE), were reported for models' evaluation. Overall, models built on the Stress feature set (compared to other feature sets), and models to diagnose the second task (compared to task 1 models) revealed better performance. The Stress-mRMR-KNN (feature set-feature selection-classifier) reached the highest performance for task 1 with AUC, ACC, SEN, and SPE equal to 0.61, 0.63, 0.64, and 0.6, respectively. The Stress-Boruta-GB model achieved the highest performance for task 2 with AUC, ACC, SEN, and SPE of 0.79, 0.76, 0.75, and 0.76, respectively. Diabetes status from the clinical feature family, and dependence count non-uniformity normalized, from the NGLDM family, which is representative of non-uniformity in the region of interest were the most frequently selected features from stress feature set for CAD risk classification. This study revealed promising results for CAD risk classification using machine learning models built on MPI SPECT radiomics. The proposed models are helpful to alleviate the labor-intensive MPI SPECT interpretation process regarding CAD status and can potentially expedite the diagnostic process.

Prediction of Parkinson's disease pathogenic variants using hybrid Machine learning systems and radiomic features

PURPOSE

In Parkinson's disease (PD), 5-10% of cases are of genetic origin with mutations identified in several genes such as leucine-rich repeat kinase 2 (LRRK2) and glucocerebrosidase (GBA). We aim to predict these two gene mutations using hybrid machine learning systems (HMLS), via imaging and non-imaging data, with the long-term goal to predict conversion to active disease.

METHODS

We studied 264 and 129 patients with known LRRK2 and GBA mutations status from PPMI database. Each dataset includes 513 features such as clinical features (CFs), conventional imaging features (CIFs) and radiomic features (RFs) extracted from DAT-SPECT images. Features, normalized by Z-score, were univariately analyzed for statistical significance by the t-test and chi-square test, adjusted by Benjamini-Hochberg correction. Multiple HMLSs, including 11 features extraction (FEA) or 10 features selection algorithms (FSA) linked with 21 classifiers were utilized. We also employed Ensemble Voting (EV) to classify the genes.

RESULTS

For prediction of LRRK2 mutation status, a number of HMLSs resulted in accuracies of 0.98 ± 0.02 and 1.00 in 5-fold cross-validation (80% out of total data points) and external testing (remaining 20%), respectively. For predicting GBA mutation status, multiple HMLSs resulted in high accuracies of 0.90 ± 0.08 and 0.96 in 5-fold cross-validation and external testing, respectively. We additionally showed that SPECT-based RFs added value to the specific prediction of of GBA mutation status.

CONCLUSION

We demonstrated that combining medical information with SPECT-based imaging features, and optimal utilization of HMLS can produce excellent prediction of the mutations status in PD patients.

Myocardial Perfusion SPECT Imaging Radiomic Features and Machine Learning Algorithms for Cardiac Contractile Pattern Recognition

A U-shaped contraction pattern was shown to be associated with a better Cardiac resynchronization therapy (CRT) response. The main goal of this study is to automatically recognize left ventricular contractile patterns using machine learning algorithms trained on conventional quantitative features (ConQuaFea) and radiomic features extracted from Gated single-photon emission computed tomography myocardial perfusion imaging (GSPECT MPI). Among 98 patients with standard resting GSPECT MPI included in this study, 29 received CRT therapy and 69 did not (also had CRT inclusion criteria but did not receive treatment yet at the time of data collection, or refused treatment). A total of 69 non-CRT patients were employed for training, and the 29 were employed for testing. The models were built utilizing features from three distinct feature sets (ConQuaFea, radiomics, and ConQuaFea + radiomics (combined)), which were chosen using Recursive feature elimination (RFE) feature selection (FS), and then trained using seven different machine learning (ML) classifiers. In addition, CRT outcome prediction was assessed by different treatment inclusion criteria as the study's final phase. The MLP classifier had the highest performance among ConQuaFea models (AUC, SEN, SPE = 0.80, 0.85, 0.76). RF achieved the best performance in terms of AUC, SEN, and SPE with values of 0.65, 0.62, and 0.68, respectively, among radiomic models. GB and RF approaches achieved the best AUC, SEN, and SPE values of 0.78, 0.92, and 0.63 and 0.74, 0.93, and 0.56, respectively, among the combined models. A promising outcome was obtained when using radiomic and ConQuaFea from GSPECT MPI to detect left ventricular contractile patterns by machine learning.

Development of a patients' satisfaction analysis system using machine learning and lexicon-based methods

BACKGROUND

Patients' rights are integral to medical ethics. This study aimed to perform sentiment analysis and opinion mining on patients' messages by a combination of lexicon-based and machine learning methods to identify positive or negative comments and to determine the different ward and staff names mentioned in patients' messages.

METHODS

The level of satisfaction and observance of the rights of 250 service recipients of the hospital was evaluated through the related checklists by the evaluator. In total, 822 Persian messages, composed of 540 negative and 282 positive comments, were collected and labeled by the evaluator. Pre-processing was performed on the messages and followed by 2 feature vectors which were extracted from the messages, including the term frequency-inverse document frequency (TFIDF) vector and a combination of the multifeature (MF) (a lexicon-based method) and TFIDF (MF + TFIDF) vectors. Six feature selectors and 5 classifiers were used in this study. For the evaluations, 5-fold cross-validation with different metrics including area under the receiver operating characteristic curve (AUC), accuracy (ACC), F1 score, sensitivity (SEN), specificity (SPE) and Precision-Recall Curves (PRC) were reported. Message tag detection, which featured different hospital wards and identified staff names mentioned in the study patients' messages, was implemented by the lexicon-based method.

RESULTS

The best classifier was Multinomial Naïve Bayes in combination with MF + TFIDF feature vector and SelectFromModel (SFM) feature selection (ACC = 0.89 ± 0.03, AUC = 0.87 ± 0.03, F1 = 0.92 ± 0.03, SEN = 0.93 ± 0.04, and SPE = 0.82 ± 0.02, PRC-AUC = 0.97). Two methods of assessment by the evaluator and artificial intelligence as well as survey systems were compared.

CONCLUSION

Our results demonstrated that the lexicon-based method, in combination with machine learning classifiers, could extract sentiments in patients' comments and classify them into positive and negative categories. We also developed an online survey system to analyze patients' satisfaction in different wards and to remove conventional assessments by the evaluator.

Artificial intelligence-based analysis of whole-body bone scintigraphy: The quest for the optimal deep learning algorithm and comparison with human observer performance

PURPOSE

Whole-body bone scintigraphy (WBS) is one of the most widely used modalities in diagnosing malignant bone diseases during the early stages. However, the procedure is time-consuming and requires vigour and experience. Moreover, interpretation of WBS scans in the early stages of the disorders might be challenging because the patterns often reflect normal appearance that is prone to subjective interpretation. To simplify the gruelling, subjective, and prone-to-error task of interpreting WBS scans, we developed deep learning (DL) models to automate two major analyses, namely (i) classification of scans into normal and abnormal and (ii) discrimination between malignant and non-neoplastic bone diseases, and compared their performance with human observers.

MATERIALS AND METHODS

After applying our exclusion criteria on 7188 patients from three different centers, 3772 and 2248 patients were enrolled for the first and second analyses, respectively. Data were split into two parts, including training and testing, while a fraction of training data were considered for validation. Ten different CNN models were applied to single- and dual-view input (posterior and anterior views) modes to find the optimal model for each analysis. In addition, three different methods, including squeeze-and-excitation (SE), spatial pyramid pooling (SPP), and attention-augmented (AA), were used to aggregate the features for dual-view input models. Model performance was reported through area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, and specificity and was compared with the DeLong test applied to ROC curves. The test dataset was evaluated by three nuclear medicine physicians (NMPs) with different levels of experience to compare the performance of AI and human observers.

RESULTS

DenseNet121_AA (DensNet121, with dual-view input aggregated by AA) and InceptionResNetV2_SPP achieved the highest performance (AUC = 0.72) for the first and second analyses, respectively. Moreover, on average, in the first analysis, Inception V3 and InceptionResNetV2 CNN models and dual-view input with AA aggregating method had superior performance. In addition, in the second analysis, DenseNet121 and InceptionResNetV2 as CNN methods and dual-view input with AA aggregating method achieved the best results. Conversely, the performance of AI models was significantly higher than human observers for the first analysis, whereas their performance was comparable in the second analysis, although the AI model assessed the scans in a drastically lower time.

CONCLUSION

Using the models designed in this study, a positive step can be taken toward improving and optimizing WBS interpretation. By training DL models with larger and more diverse cohorts, AI could potentially be used to assist physicians in the assessment of WBS images.

MRI-Based Radiomics Combined with Deep Learning for Distinguishing IDH-Mutant WHO Grade 4 Astrocytomas from IDH-Wild-Type Glioblastomas

This study aimed to investigate the potential of quantitative radiomic data extracted from conventional MR images in discriminating IDH-mutant grade 4 astrocytomas from IDH-wild-type glioblastomas (GBMs). A cohort of 57 treatment-naïve patients with IDH-mutant grade 4 astrocytomas (n = 23) and IDH-wild-type GBMs (n = 34) underwent anatomical imaging on a 3T MR system with standard parameters. Post-contrast T1-weighted and T2-FLAIR images were co-registered. A semi-automatic segmentation approach was used to generate regions of interest (ROIs) from different tissue components of neoplasms. A total of 1050 radiomic features were extracted from each image. The data were split randomly into training and testing sets. A deep learning-based data augmentation method (CTGAN) was implemented to synthesize 200 datasets from the training sets. A total of 18 classifiers were used to distinguish two genotypes of grade 4 astrocytomas. From generated data using 80% training set, the best discriminatory power was obtained from core tumor regions overlaid on post-contrast T1 using the K-best feature selection algorithm and a Gaussian naïve Bayes classifier (AUC = 0.93, accuracy = 0.92, sensitivity = 1, specificity = 0.86, PR_AUC = 0.92). Similarly, high diagnostic performances were obtained from original and generated data using 50% and 30% training sets. Our findings suggest that conventional MR imaging-based radiomic features combined with machine/deep learning methods may be valuable in discriminating IDH-mutant grade 4 astrocytomas from IDH-wild-type GBMs.

COVID-19 prognostic modeling using CT radiomic features and machine learning algorithms: Analysis of a multi-institutional dataset of 14,339 patients

BACKGROUND

We aimed to analyze the prognostic power of CT-based radiomics models using data of 14,339 COVID-19 patients.

METHODS

Whole lung segmentations were performed automatically using a deep learning-based model to extract 107 intensity and texture radiomics features. We used four feature selection algorithms and seven classifiers. We evaluated the models using ten different splitting and cross-validation strategies, including non-harmonized and ComBat-harmonized datasets. The sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were reported.

RESULTS

In the test dataset (4,301) consisting of CT and/or RT-PCR positive cases, AUC, sensitivity, and specificity of 0.83 ± 0.01 (CI95%: 0.81-0.85), 0.81, and 0.72, respectively, were obtained by ANOVA feature selector + Random Forest (RF) classifier. Similar results were achieved in RT-PCR-only positive test sets (3,644). In ComBat harmonized dataset, Relief feature selector + RF classifier resulted in the highest performance of AUC, reaching 0.83 ± 0.01 (CI95%: 0.81-0.85), with a sensitivity and specificity of 0.77 and 0.74, respectively. ComBat harmonization did not depict statistically significant improvement compared to a non-harmonized dataset. In leave-one-center-out, the combination of ANOVA feature selector and RF classifier resulted in the highest performance.

CONCLUSION

Lung CT radiomics features can be used for robust prognostic modeling of COVID-19. The predictive power of the proposed CT radiomics model is more reliable when using a large multicentric heterogeneous dataset, and may be used prospectively in clinical setting to manage COVID-19 patients.

Synergistic impact of motion and acquisition/reconstruction parameters on 18 F-FDG PET radiomic features in non-small cell lung cancer: phantom and clinical studies

Objectives

This study is aimed at examining the synergistic impact of motion and acquisition/reconstruction parameters on ¹⁸F‐FDG PET image radiomic features in non‐small cell lung cancer (NSCLC) patients, and investigating the robustness of features performance in differentiating NSCLC histopathology subtypes.

Methods

An in‐house developed thoracic phantom incorporating lesions with different sizes was used with different reconstruction settings, including various reconstruction algorithms, number of subsets and iterations, full‐width at half‐maximum of post‐reconstruction smoothing filter and acquisition parameters, including injected activity and test–retest with and without motion simulation. To simulate motion, a special motor was manufactured to simulate respiratory motion based on a normal patient in two directions. The lesions were delineated semi‐automatically to extract 174 radiomic features. All radiomic features were categorized according to the coefficient of variation (COV) to select robust features. A cohort consisting of 40 NSCLC patients with adenocarcinoma (n = 20) and squamous cell carcinoma (n = 20) was retrospectively analyzed. Statistical analysis was performed to discriminate robust features in differentiating histopathology subtypes of NSCLC lesions.

Results

Overall, 29% of radiomic features showed a COV ≤5% against motion. Forty‐five percent and 76% of the features showed a COV ≤ 5% against the test–retest with and without motion in large lesions, respectively. Thirty‐three percent and 45% of the features showed a COV ≤ 5% against different reconstruction parameters with and without motion, respectively. For NSCLC histopathological subtype differentiation, statistical analysis showed that 31 features were significant (p‐value < 0.05). Two out of the 31 significant features, namely, the joint entropy of GLCM (AUC = 0.71, COV = 0.019) and median absolute deviation of intensity histogram (AUC = 0.7, COV = 0.046), were robust against the motion (same reconstruction setting).

Conclusions

Motion, acquisition, and reconstruction parameters significantly impact radiomic features, just as their synergies. Radiomic features with high predictive performance (statistically significant) in differentiating histopathological subtype of NSCLC may be eliminated due to non‐reproducibility.

Longitudinal clustering analysis and prediction of Parkinson's disease progression using radiomics and hybrid machine learning

BACKGROUND

We employed machine learning approaches to (I) determine distinct progression trajectories in Parkinson's disease (PD) (unsupervised clustering task), and (II) predict progression trajectories (supervised prediction task), from early (years 0 and 1) data, making use of clinical and imaging features.

METHODS

We studied PD-subjects derived from longitudinal datasets (years 0, 1, 2 & 4; Parkinson's Progressive Marker Initiative). We extracted and analyzed 981 features, including motor, non-motor, and radiomics features extracted for each region-of-interest (ROIs: left/right caudate and putamen) using our standardized standardized environment for radiomics analysis (SERA) radiomics software. Segmentation of ROIs on dopamine transposer - single photon emission computed tomography (DAT SPECT) images were performed via magnetic resonance images (MRI). After performing cross-sectional clustering on 885 subjects (original dataset) to identify disease subtypes, we identified optimal longitudinal trajectories using hybrid machine learning systems (HMLS), including principal component analysis (PCA) + K-Means algorithms (KMA) followed by Bayesian information criterion (BIC), Calinski-Harabatz criterion (CHC), and elbow criterion (EC). Subsequently, prediction of the identified trajectories from early year data was performed using multiple HMLSs including 16 Dimension Reduction Algorithms (DRA) and 10 classification algorithms.

RESULTS

We identified 3 distinct progression trajectories. Hotelling's t squared test (HTST) showed that the identified trajectories were distinct. The trajectories included those with (I, II) disease escalation (2 trajectories, 27% and 38% of patients) and (III) stable disease (1 trajectory, 35% of patients). For trajectory prediction from early year data, HMLSs including the stochastic neighbor embedding algorithm (SNEA, as a DRA) as well as locally linear embedding algorithm (LLEA, as a DRA), linked with the new probabilistic neural network classifier (NPNNC, as a classifier), resulted in accuracies of 78.4% and 79.2% respectively, while other HMLSs such as SNEA + Lib_SVM (library for support vector machines) and t_SNE (t-distributed stochastic neighbor embedding) + NPNNC resulted in 76.5% and 76.1% respectively.

CONCLUSIONS

This study moves beyond cross-sectional PD subtyping to clustering of longitudinal disease trajectories. We conclude that combining medical information with SPECT-based radiomics features, and optimal utilization of HMLSs, can identify distinct disease trajectories in PD patients, and enable effective prediction of disease trajectories from early year data.

COLI-Net: Deep learning-assisted fully automated COVID-19 lung and infection pneumonia lesion detection and segmentation from chest computed tomography images

We present a deep learning (DL)-based automated whole lung and COVID-19 pneumonia infectious lesions (COLI-Net) detection and segmentation from chest computed tomography (CT) images. This multicenter/multiscanner study involved 2368 (347'259 2D slices) and 190 (17 341 2D slices) volumetric CT exams along with their corresponding manual segmentation of lungs and lesions, respectively. All images were cropped, resized, and the intensity values clipped and normalized. A residual network with non-square Dice loss function built upon TensorFlow was employed. The accuracy of lung and COVID-19 lesions segmentation was evaluated on an external reverse transcription-polymerase chain reaction positive COVID-19 dataset (7'333 2D slices) collected at five different centers. To evaluate the segmentation performance, we calculated different quantitative metrics, including radiomic features. The mean Dice coefficients were 0.98 ± 0.011 (95% CI, 0.98-0.99) and 0.91 ± 0.038 (95% CI, 0.90-0.91) for lung and lesions segmentation, respectively. The mean relative Hounsfield unit differences were 0.03 ± 0.84% (95% CI, -0.12 to 0.18) and -0.18 ± 3.4% (95% CI, -0.8 to 0.44) for the lung and lesions, respectively. The relative volume difference for lung and lesions were 0.38 ± 1.2% (95% CI, 0.16-0.59) and 0.81 ± 6.6% (95% CI, -0.39 to 2), respectively. Most radiomic features had a mean relative error less than 5% with the highest mean relative error achieved for the lung for the range first-order feature (-6.95%) and least axis length shape feature (8.68%) for lesions. We developed an automated DL-guided three-dimensional whole lung and infected regions segmentation in COVID-19 patients to provide fast, consistent, robust, and human error immune framework for lung and pneumonia lesion detection and quantification.

Cardiac SPECT radiomic features repeatability and reproducibility: A multi-scanner phantom study

BACKGROUND

The aim of this work was to assess the robustness of cardiac SPECT radiomic features against changes in imaging settings, including acquisition, and reconstruction parameters.

METHODS

Four commercial SPECT and SPECT/CT cameras were used to acquire images of a static cardiac phantom mimicking typical myorcardial perfusion imaging using 185 MBq of 99mTc. The effects of different image acquisition and reconstruction parameters, including number of views, view matrix size, attenuation correction, as well as image reconstruction related parameters (algorithm, number of iterations, number of subsets, type of post-reconstruction filter, and its associated parameters, including filter order and cut-off frequency) were studied. In total, 5,063 transverse views were reconstructed by varying the aforementioned factors. Eighty-seven radiomic features including first-, second-, and high-order textures were extracted from these images. To assess reproducibility and repeatability, the coefficient of variation (COV), as a widely adopted metric, was measured for each of the radiomic features over the different imaging settings.

RESULTS

The Inverse Difference Moment Normalized (IDMN) and Inverse Difference Normalized (IDN) features from the Gray Level Co-occurrence Matrix (GLCM), Run Percentage (RP) from the Gray Level Co-occurrence Matrix (GLRLM), Zone Entropy (ZE) from the Gray Level Size Zone Matrix (GLSZM), and Dependence Entropy (DE) from the Gray Level Dependence Matrix (GLDM) feature sets were the only features that exhibited high reproducibility (COV ≤ 5%) against changes in all imaging settings. In addition, Large Area Low Gray Level Emphasis (LALGLE), Small Area Low Gray Level Emphasis (SALGLE) and Low Gray Level Zone Emphasis (LGLZE) from GLSZM, and Small Dependence Low Gray Level Emphasis (SDLGLE) from GLDM feature sets turned out to be less reproducible (COV > 20%) against changes in imaging settings. The GLRLM (31.88%) and GLDM feature set (54.2%) had the highest (COV < 5%) and lowest (COV > 20%) number of the reproducible features, respectively. Matrix size had the largest impact on feature variability as most of the features were not repeatable when matrix size was modified with 82.8% of them having a COV > 20%.

CONCLUSION

The repeatability and reproducibility of SPECT/CT cardiac radiomic features under different imaging settings is feature-dependent. Different image acquisition and reconstruction protocols have variable effects on radiomic features. The radiomic features exhibiting low COV are potential candidates for future clinical studies.

The Impact of Preprocessing on the PET-CT Radiomics Features in Non-small Cell Lung Cancer

Purpose: This study aimed to investigate the impact of image preprocessing steps, including Gray Level Discretization (GLD) and different Interpolation Algorithms (IA) on 18F-Fluorodeoxyglucose (18F-FDG) radiomics features in Non-Small Cell Lung Cancer (NSCLC). Materials and Methods: One hundred and seventy-two radiomics features from the first-, second-, and higher-order statistic features were calculated from a set of Positron Emission Tomography/Computed Tomography (PET/CT) images of 20 non-small cell lung cancer delineated tumors with volumes ranging from 10 to 418 cm3 regarding five intensity discretization schemes with the number of gray levels of 16, 32, 64, 128, and 256, and four Interpolation algorithms, including nearest neighbor, tricubic convolution and tricubic spline interpolation, and trilinear were used. Segmentation was based on 3D region growing-based. The Intraclass Correlation Coefficient (ICC), Overall Concordance Correlation Coefficient (OCCC), and Coefficient Of Variations (COV) were calculated to demonstrate the features' variability and select robust features. ICC and OCCC < 0.5 presented weak reliability, ICC and OCCC between 0.5 and 0.75 illustrated appropriate reliability, values within 0.75 and 0.9 showed satisfying reliability, and values higher than 0.90 indicate exceptional reliability. Besides, features with less than 10% COV have been selected as robust features. Results: All morphology family (except four features), statistic, and Intensity volume histogram families were not affected by GLD and IA. And the rest of them, 10 and 61 features showed COV ≤ 5% against GLD and IA, respectively. Ten and 80 features showed excellent reliability (ICC values greater than 0.90) against GLD and IA. Eight and 60 features showed OCCC≥0.90 against GLD and IA, respectively. Based on our results Inverse difference normalized and Inverse difference moment normalized from Grey Level Co-occurrence Matrix (GLCM) were the most robust features against GLD and Skewness from intensity histogram family and Inverse difference normalized and Inverse difference moment normalized from GLCM were the most robust features against IA. Conclusion: Preprocessing can substantially impact the 18F-FDG PET image radiomic features in NSCLC. The impact of gray level discretization on radiomics features is significant and more than Interpolation algorithms.

Multi-level multi-modality (PET and CT) fusion radiomics: prognostic modeling for non-small cell lung carcinoma

We developed multi-modality radiomic models by integrating information extracted from¹⁸F-FDG PET and CT images using feature- and image-level fusions, toward improved prognosis for non-small cell lung carcinoma (NSCLC) patients. Two independent cohorts of NSCLC patients from two institutions (87 and 95 patients) were cycled as training and testing datasets. Fusion approaches were applied at two levels, namely feature- and image-levels. For feature-level fusion, radiomic features were extracted individually from CT and PET images and concatenated. Alternatively, radiomic features extracted separately from CT and PET images were averaged. For image-level fusion, wavelet fusion was utilized and tuned with two parameters, namely CT weight and Wavelet Band Pass Filtering Ratio. Clinical and combined clinical + radiomic models were developed. Gray level discretization was performed at 3 different levels (16, 32 and 64) and 225 radiomics features were extracted. Overall survival (OS) was considered as the endpoint. For feature reduction, correlated (redundant) features were excluded using Spearman's correlation, and best combination of top ten features with highest concordance-indices (via univariate Cox model) were selected in each model for further multivariate Cox model. Moreover, prognostic score's median, obtained from the training cohort, was used intact in the testing cohort as a threshold to classify patients into low- versus high-risk groups, and log-rank test was applied to assess differences between the Kaplan-Meier curves. Overall, while models based on feature-level fusion strategy showed limited superiority over single-modalities, image-level fusion strategy significantly outperformed both single-modality and feature-level fusion strategies. As such, the clinical model (C-index = 0.656) outperformed all models from single-modality and feature-level strategies, but was outperformed by certain models from image-level fusion strategy. Our findings indicated that image-level fusion multi-modality radiomics models outperformed single-modality, feature-level fusion, and clinical models for OS prediction of NSCLC patients.

Next-Generation Radiogenomics Sequencing for Prediction of EGFR and KRAS Mutation Status in NSCLC Patients Using Multimodal Imaging and Machine Learning Algorithms

PURPOSE

Considerable progress has been made in the assessment and management of non-small cell lung cancer (NSCLC) patients based on mutation status in the epidermal growth factor receptor (EGFR) and Kirsten rat sarcoma viral oncogene (KRAS). At the same time, NSCLC management through KRAS and EGFR mutation profiling faces challenges. In the present work, we aimed to evaluate a comprehensive radiomics framework that enabled prediction of EGFR and KRAS mutation status in NSCLC patients based on radiomic features from low-dose computed tomography (CT), contrast-enhanced diagnostic quality CT (CTD), and positron emission tomography (PET) imaging modalities and use of machine learning algorithms.

METHODS

Our study involved NSCLC patients including 150 PET, low-dose CT, and CTD images. Radiomic features from original and preprocessed (including 64 bin discretizing, Laplacian-of-Gaussian (LOG), and Wavelet) images were extracted. Conventional clinically used standard uptake value (SUV) parameters and metabolic tumor volume (MTV) were also obtained from PET images. Highly correlated features were pre-eliminated, and false discovery rate (FDR) correction was performed with the resulting q-values reported for univariate analysis. Six feature selection methods and 12 classifiers were then used for multivariate prediction of gene mutation status (provided by polymerase chain reaction (PCR)) in patients. We performed 10-fold cross-validation for model tuning to improve robustness, and our developed models were assessed on an independent validation set with 68 patients (common in all three imaging modalities). The average area under the receiver operator characteristic curve (AUC) was utilized for performance evaluation.

RESULTS

The best predictive power for conventional PET parameters was achieved by SUV_peak (AUC 0.69, p value = 0.0002) and MTV (AUC 0.55, p value = 0.0011) for EGFR and KRAS, respectively. Univariate analysis of extracted radiomics features improved AUC performance to 0.75 (q-value 0.003, Short-Run Emphasis feature of GLRLM from LOG preprocessed image of PET with sigma value 1.5) and 0.71 (q-value 0.00005, Large Dependence Low Gray-Level Emphasis feature of GLDM in LOG preprocessed image of CTD with sigma value 5) for EGFR and KRAS, respectively. Furthermore, multivariate machine learning-based AUC performances were significantly improved to 0.82 for EGFR (LOG preprocessed image of PET with sigma 3 with variance threshold (VT) feature selector and stochastic gradient descent (SGD) classifier (q-value = 4.86E-05) and 0.83 for KRAS (LOG preprocessed image of CT with sigma 3.5 with select model (SM) feature selector and SGD classifier (q-value = 2.81E-09).

CONCLUSION

Our work demonstrated that non-invasive and reliable radiomics analysis can be successfully used to predict EGFR and KRAS mutation status in NSCLC patients. We demonstrated that radiomic features extracted from different image-feature sets could be used for EGFR and KRAS mutation status prediction in NSCLC patients and showed improved predictive power relative to conventional image-derived metrics.

Treatment response prediction using MRI-based pre-, post-, and delta-radiomic features and machine learning algorithms in colorectal cancer

OBJECTIVES

We evaluate the feasibility of treatment response prediction using MRI-based pre-, post-, and delta-radiomic features for locally advanced rectal cancer (LARC) patients treated by neoadjuvant chemoradiation therapy (nCRT).

MATERIALS AND METHODS

This retrospective study included 53 LARC patients divided into a training set (Center#1, n = 36) and external validation set (Center#2, n = 17). T2-weighted (T2W) MRI was acquired for all patients, 2 weeks before and 4 weeks after nCRT. Ninety-six radiomic features, including intensity, morphological and second- and high-order texture features were extracted from segmented 3D volumes from T2W MRI. All features were harmonized using ComBat algorithm. Max-Relevance-Min-Redundancy (MRMR) algorithm was used as feature selector and k-nearest neighbors (KNN), Naïve Bayes (NB), Random forests (RF), and eXtreme Gradient Boosting (XGB) algorithms were used as classifiers. The evaluation was performed using the area under the receiver operator characteristic (ROC) curve (AUC), sensitivity, specificity and accuracy.

RESULTS

In univariate analysis, the highest AUC in pre-, post-, and delta-radiomic features were 0.78, 0.70, and 0.71, for GLCM_IMC1, shape (surface area and volume) and GLSZM_GLNU features, respectively. In multivariate analysis, RF and KNN achieved the highest AUC (0.85 ± 0.04 and 0.81 ± 0.14, respectively) among pre- and post-treatment features. The highest AUC was achieved for the delta-radiomic-based RF model (0.96 ± 0.01) followed by NB (0.96 ± 0.04). Overall. Delta-radiomics model, outperformed both pre- and post-treatment features (P-value <0.05).

CONCLUSION

Multivariate analysis of delta-radiomic T2W MRI features using machine learning algorithms could potentially be used for response prediction in LARC patients undergoing nCRT. We also observed that multivariate analysis of delta-radiomic features using RF classifiers can be used as powerful biomarkers for response prediction in LARC.

Machine learning-based prognostic modeling using clinical data and quantitative radiomic features from chest CT images in COVID-19 patients

OBJECTIVE

To develop prognostic models for survival (alive or deceased status) prediction of COVID-19 patients using clinical data (demographics and history, laboratory tests, visual scoring by radiologists) and lung/lesion radiomic features extracted from chest CT images.

METHODS

Overall, 152 patients were enrolled in this study protocol. These were divided into 106 training/validation and 46 test datasets (untouched during training), respectively. Radiomic features were extracted from the segmented lungs and infectious lesions separately from chest CT images. Clinical data, including patients' history and demographics, laboratory tests and radiological scores were also collected. Univariate analysis was first performed (q-value reported after false discovery rate (FDR) correction) to determine the most predictive features among all imaging and clinical data. Prognostic modeling of survival was performed using radiomic features and clinical data, separately or in combination. Maximum relevance minimum redundancy (MRMR) and XGBoost were used for feature selection and classification. The receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC), sensitivity, specificity, and accuracy were used to assess the prognostic performance of the models on the test datasets.

RESULTS

For clinical data, cancer comorbidity (q-value < 0.01), consciousness level (q-value < 0.05) and radiological score involved zone (q-value < 0.02) were found to have high correlated features with outcome. Oxygen saturation (AUC = 0.73, q-value < 0.01) and Blood Urea Nitrogen (AUC = 0.72, q-value = 0.72) were identified as high clinical features. For lung radiomic features, SAHGLE (AUC = 0.70) and HGLZE (AUC = 0.67) from GLSZM were identified as most prognostic features. Amongst lesion radiomic features, RLNU from GLRLM (AUC = 0.73), HGLZE from GLSZM (AUC = 0.73) had the highest performance. In multivariate analysis, combining lung, lesion and clinical features was determined to provide the most accurate prognostic model (AUC = 0.95 ± 0.029 (95%CI: 0.95-0.96), accuracy = 0.88 ± 0.046 (95% CI: 0.88-0.89), sensitivity = 0.88 ± 0.066 (95% CI = 0.87-0.9) and specificity = 0.89 ± 0.07 (95% CI = 0.87-0.9)).

CONCLUSION

Combination of radiomic features and clinical data can effectively predict outcome in COVID-19 patients. The developed model has significant potential for improved management of COVID-19 patients.

Radiographic Texture Reproducibility: The Impact of Different Materials, their Arrangement, and Focal Spot Size

Background:

Feature reproducibility is a critical issue in quantitative radiomic studies. The aim of this study is to assess how radiographic radiomic textures behave against changes in phantom materials, their arrangements, and focal spot size.

Method:

A phantom with detachable parts was made using wood, sponge, Plexiglas, and rubber. Each material had 1 cm thickness and was imaged for consecutive time. The phantom also was imaged by change in the arrangement of its materials. Imaging was done with two focal spot sizes including 0.6 and 1.2 mm. All images were acquired with a digital radiography machine. Several texture features were extracted from the same size region of interest in all images. To assess reproducibility, coefficient of variation (COV), intraclass correlation coefficient (ICC), and Bland–Altman tests were used.

Results:

Results show that 59%, 50%, and 4.5% of all features are most reproducible (COV ≤5%) against change in focal spot size, material arrangements, and phantom's materials, respectively. Results on Bland–Altman analysis showed that there is just a nonreproducible feature against change in the focal spot size. On the ICC results, we observed that the ICCs for more features are >0.90 and there were few features with ICC lower than 0.90.

Conclusion:

We showed that radiomic textures are vulnerable against changes in materials, arrangement, and different focal spot sizes. These results suggest that a careful analysis of the effects of these parameters is essential before any radiomic clinical application.

Repeatability of radiomic features in magnetic resonance imaging of glioblastoma: Test-retest and image registration analyses

PURPOSE

To assess the repeatability of radiomic features in magnetic resonance (MR) imaging of glioblastoma (GBM) tumors with respect to test-retest, different image registration approaches and inhomogeneity bias field correction.

METHODS

We analyzed MR images of 17 GBM patients including T1- and T2-weighted images (performed within the same imaging unit on two consecutive days). For image segmentation, we used a comprehensive segmentation approach including entire tumor, active area of tumor, necrotic regions in T1-weighted images, and edema regions in T2-weighted images (test studies only; registration to retest studies is discussed next). Analysis included N3, N4 as well as no bias correction performed on raw MR images. We evaluated 20 image registration approaches, generated by cross-combination of four transformation and five cost function methods. In total, 714 images (17 patients × 2 images × ((4 transformations × 5 cost functions) + 1 test image) and 2856 segmentations (714 images × 4 segmentations) were prepared for feature extraction. Various radiomic features were extracted, including the use of preprocessing filters, specifically wavelet (WAV) and Laplacian of Gaussian (LOG), as well as discretization into fixed bin width and fixed bin count (16, 32, 64, 128, and 256), Exponential, Gradient, Logarithm, Square and Square Root scales. Intraclass correlation coefficients (ICC) were calculated to assess the repeatability of MRI radiomic features (high repeatability defined as ICC ≥ 95%).

RESULTS

In our ICC results, we observed high repeatability (ICC ≥ 95%) with respect to image preprocessing, different image registration algorithms, and test-retest analysis, for example: RLNU and GLNU from GLRLM, GLNU and DNU from GLDM, Coarseness and Busyness from NGTDM, GLNU and ZP from GLSZM, and Energy and RMS from first order. Highest fraction (percent) of repeatable features was observed, among registration techniques, for the method Full Affine transformation with 12 degrees of freedom using Mutual Information cost function (mean 32.4%), and among image processing methods, for the method Laplacian of Gaussian (LOG) with Sigma (2.5-4.5 mm) (mean 78.9%). The trends were relatively consistent for N4, N3, or no bias correction.

CONCLUSION

Our results showed varying performances in repeatability of MR radiomic features for GBM tumors due to test-retest and image registration. The findings have implications for appropriate usage in diagnostic and predictive models.

Noninvasive Fuhrman grading of clear cell renal cell carcinoma using computed tomography radiomic features and machine learning

PURPOSE

To identify optimal classification methods for computed tomography (CT) radiomics-based preoperative prediction of clear cell renal cell carcinoma (ccRCC) grade.

MATERIALS AND METHODS

Seventy-one ccRCC patients (31 low grade and 40 high grade) were included in this study. Tumors were manually segmented on CT images followed by the application of three image preprocessing techniques (Laplacian of Gaussian, wavelet filter, and discretization of the intensity values) on delineated tumor volumes. Overall, 2530 radiomics features (tumor shape and size, intensity statistics, and texture) were extracted from each segmented tumor volume. Univariate analysis was performed to assess the association between each feature and the histological condition. Multivariate analysis involved the use of machine learning (ML) algorithms and the following three feature selection algorithms: the least absolute shrinkage and selection operator, Student's t test, and minimum Redundancy Maximum Relevance. These selected features were then used to construct three classification models (SVM, random forest, and logistic regression) to discriminate high from low-grade ccRCC at nephrectomy. Lastly, multivariate model performance was evaluated on the bootstrapped validation cohort using the area under the receiver operating characteristic curve (AUC) metric.

RESULTS

The univariate analysis demonstrated that among the different image sets, 128 bin-discretized images have statistically significant different texture parameters with a mean AUC of 0.74 ± 3 (q value < 0.05). The three ML-based classifiers showed proficient discrimination between high and low-grade ccRCC. The AUC was 0.78 for logistic regression, 0.62 for random forest, and 0.83 for the SVM model, respectively.

CONCLUSION

CT radiomic features can be considered as a useful and promising noninvasive methodology for preoperative evaluation of ccRCC Fuhrman grades.