Integrating intratumoral and peritumoral radiomics with deep transfer learning for DCE-MRI breast lesion differentiation: A multicenter study comparing performance with radiologists.

PURPOSE: To conduct the fusion of radiomics and deep transfer learning features from the intratumoral and peritumoral areas in breast DCE-MRI images to differentiate between benign and malignant breast tumors, and to compare the diagnostic accuracy of this fusion model against the assessments made by experienced radiologists. MATERIALS AND METHODS: This multi-center study conducted a retrospective analysis of DCE-MRI images from 330 women diagnosed with breast cancer, with 138 cases categorized as benign and 192 as malignant. The training and internal testing sets comprised 270 patients from center 1, while the external testing cohort consisted of 60 patients from center 2. A fusion feature set consisting of radiomics features and deep transfer learning features was constructed from both intratumoral (ITR) and peritumoral (PTR) areas. The Least absolute shrinkage and selection operator (LASSO) based support vector machine was chosen as the classifier by comparing its performance with five other machine learning models. The diagnostic performance and clinical usefulness of fusion model were verified and assessed through the area under the receiver operating characteristics (ROC) and decision curve analysis. Additionally, the performance of the fusion model was compared with the diagnostic assessments of two experienced radiologists to evaluate its relative accuracy. The study strictly adhered to CLEAR and METRICS guidelines for standardization to ensure rigorous and reproducible methods. RESULTS: The findings show that the fusion model, utilizing radiomics and deep transfer learning features from the ITR and PTR, exhibited exceptional performance in classifying breast tumors, achieving AUCs of 0.950 in the internal testing set and 0.921 in the external testing set. This performance significantly surpasses that of models relying on singular regional radiomics or deep transfer learning features alone. Moreover, the fusion model demonstrated superior diagnostic accuracy compared to the evaluations conducted by two experienced radiologists, thereby highlighting its potential to support and enhance clinical decision-making in the differentiation of benign and malignant breast tumors. CONCLUSION: The fusion model, combining multi-regional radiomics with deep transfer learning features, not only accurately differentiates between benign and malignant breast tumors but also outperforms the diagnostic assessments made by experienced radiologists. This underscores the model's potential as a valuable tool for improving the accuracy and reliability of breast tumor diagnosis.

Preoperative CT-based radiomics and deep learning model for predicting risk stratification of gastric gastrointestinal stromal tumors.

BACKGROUND: Gastrointestinal stromal tumors (GISTs) are clinically heterogeneous with various malignant potential in different individuals. It is crucial to explore a reliable method for preoperative risk stratification of gastric GISTs noninvasively. PURPOSE: To establish and evaluate a machine learning model using the combination of computed tomography (CT) morphology, radiomics, and deep learning features to predict the risk stratification of primary gastric GISTs preoperatively. METHODS: The 193 gastric GISTs lesions were randomly divided into training set, validation set, and test set in a ratio of 6:2:2. The qualitative and quantitative CT morphological features were assessed by two radiologists. The tumors were segmented manually, and then radiomic features were extracted using PyRadiomics and the deep learning features were extracted using pre-trained Resnet50 from arterial phase and venous phase CT images, respectively. Pearson correlation analysis and recursive feature elimination were used for feature selection. Support vector machines were employed to build a classifier for predicting the risk stratification of GISTs. This study compared the performance of models using different pre-trained convolutional neural networks (CNNs) to extract deep features for classification, as well as the performance of modeling features from single-phase and dual-phase images. The arterial phase, venous phase and dual-phase machine learning models were built, respectively, and the morphological features were added to the dual-phase machine learning model to construct a combined model. Receiver operating characteristic (ROC) curves were used to evaluate the efficacy of each model. The clinical application value of the combined model was determined through the decision curve analysis (DCA) and the net reclassification index (NRI) was analyzed. RESULTS: The area under the curve (AUC) of the dual-phase machine learning model was 0.876, which was higher than that of the arterial phase model or venous phase model (0.813, 0.838, respectively). The combined model had best predictive performance than the above models with an AUC of 0.941 (95% CI: 0.887-0.974) (p = 0.012, Delong test). DCA demonstrated that the combined model had good clinical application value with an NRI of 0.575 (95% CI: 0.357-0.891). CONCLUSION: In this study, we established a combined model that incorporated dual-phase morphology, radiomics, and deep learning characteristics, which can be used to predict the preoperative risk stratification of gastric GISTs.

CTR > 0.7 predicts the subgroup of lung adenocarcinomas ≤ 2 cm at risk of poor outcome treated by sublobar resection compared to lobar resection.

BACKGROUND: A standard surgical procedure for patients with small early-stage lung adenocarcinomas remains unknown. Hence, we aim in this study to assess the clinical utility of the consolidation-to-tumor ratio (CTR) when treating patients with small (2 cm) early stage lung cancers. METHODS: A retrospective cohort of 298 sublobar resection and 266 lobar resection recipients for early stage lung adenocarcinoma ≤ 2 cm was assembled from the First Affiliated Hospital of Chongqing Medical University between 2016 and 2019. To compare survival rates among the different groups, Kaplan-Meier curves were calculated, and the log-rank test was used. A multivariate Cox proportional hazard model was constructed utilizing variables that were significant in univariate analysis of survival. RESULTS: In the study, 564 patients were included, with 298 patients (52.8%) undergoing sublobar resection and 266 patients (47.2%) undergoing lobar resection. Regarding survival results, there was no significant difference in the 5-year overall survival (OS, P = 0.674) and 5-year recurrence-free survival (RFS, P = 0.253) between the two groups. Cox regression analyses showed that CTR ≥ 0.75(P < 0.001), age > 56 years (P = 0.007), and sublobar resection(P = 0.001) could predict worse survival. After examining survival results based on CTR categorization, we segmented the individuals into three categories: CTR<0.7, 0.7 ≤ CTR<1, and CTR = 1.The lobar resection groups had more favorable clinical outcomes than the sublobar resection groups in both the 0.7 ≤ CTR < 1(RFS: P < 0.001, OS: P = 0.001) and CTR = 1(RFS: P = 0.001, OS: P = 0.125). However, for patients with 0 ≤ CTR < 0.7, no difference in either RFS or OS was found between the lobar resection and sublobar resection groups, all of which had no positive events. Patients with a CTR between 0.7 and 1 who underwent lobar resection had similar 5-year RFS and OS rates compared to those with a CTR between 0 and 0.7 who underwent sublobar resection (100% vs. 100%). Nevertheless, a CTR of 1 following lobar resection resulted in notably reduced RFS and OS when compared to a CTR between 0.7 and 1 following lobar resection (P = 0.005 and P = 0.016, respectively). CONCLUSION: Lobar resection is associated with better long-term survival outcomes than sublobar resection for small lung adenocarcinomas ≤ 2 cm and CTR ≥ 0.7.

Nomogram for preoperative differentiation of benign and malignant breast tumors using contrast-enhanced cone-beam breast CT (CE CB-BCT) quantitative imaging and assessment features.

PURPOSE: Breast cancer's impact necessitates refined diagnostic approaches. This study develops a nomogram using radiology quantitative features from contrast-enhanced cone-beam breast CT for accurate preoperative classification of benign and malignant breast tumors. MATERIAL AND METHODS: A retrospective study enrolled 234 females with breast tumors, split into training and test sets. Contrast-enhanced cone-beam breast CT-images were acquired using Koning Breast CT-1000. Quantitative assessment features were extracted via 3D-slicer software, identifying independent predictors. The nomogram was constructed to preoperative differentiation benign and malignant breast tumors. Calibration curve was used to assess whether the model showed favorable correspondence with pathological confirmation. Decision curve analysis confirmed the model's superiority. RESULTS: The study enrolled 234 female patients with a mean age of 50.2 years (SD ± 9.2). The training set had 164 patients (89 benign, 75 malignant), and the test set had 70 patients (29 benign, 41 malignant). The nomogram achieved excellent predictive performance in distinguishing benign and malignant breast lesions with an AUC of 0.940 (95% CI 0.900-0.940) in the training set and 0.970 (95% CI 0.940-0.970) in the test set. CONCLUSION: This study illustrates the effectiveness of quantitative radiology features derived from contrast-enhanced cone-beam breast CT in distinguishing between benign and malignant breast tumors. Incorporating these features into a nomogram-based diagnostic model allows for breast tumor diagnoses that are objective and possess good accuracy. The application of these insights could substantially increase reliability and efficacy in the management of breast tumors, offering enhanced diagnostic capability.

Boundary attention with multi-task consistency constraints for semi-supervised 2D echocardiography segmentation.

The 2D echocardiography semantic automatic segmentation technique is important in clinical applications for cardiac function assessment and diagnosis of cardiac diseases. However, automatic segmentation of 2D echocardiograms also faces the problems of loss of image boundary information, loss of image localization information, and limitations in data acquisition and annotation. To address these issues, this paper proposes a semi-supervised echocardiography segmentation method. It consists of two models: (1) a boundary attention transformer net (BATNet) and (2) a multi-task level semi-supervised model with consistency constraints on boundary features (semi-BATNet). BATNet is able to capture the location and spatial information of the input feature maps by using the self-attention mechanism. The multi-task level semi-supervised model with boundary feature consistency constraints (semi-BATNet) encourages consistent predictions of boundary features at different scales from the student and teacher networks to calculate the multi-scale consistency loss for unlabeled data. The proposed semi-BATNet was extensively evaluated on the dataset of cardiac acquisitions for multi-structure ultrasound segmentation (CAMUS) and self-collected echocardiography dataset from the First Affiliated Hospital of Chongqing Medical University. Experimental results on the CAMUS dataset showed that when only 25% of the images are labeled, the proposed method greatly improved the segmentation performance by utilizing unlabeled images, and it also outperformed five state-of-the-art semi-supervised segmentation methods. Moreover, when only 50% of the images labeled, semi-BATNet achieved the Dice coefficient values of 0.936, the Jaccard similarity of 0.881 on self-collected echocardiography dataset. Semi-BATNet can complete a more accurate segmentation of cardiac structures in 2D echocardiograms, indicating that it has the potential to accurately and efficiently assist cardiologists.

Detection of Coronary Artery Disease Based on Clinical Phonocardiogram and Multiscale Attention Convolutional Compression Network.

Heart sound is an important physiological signal that contains rich pathological information related with coronary stenosis. Thus, some machine learning methods are developed to detect coronary artery disease (CAD) based on phonocardiogram (PCG). However, current methods lack sufficient clinical dataset and fail to achieve efficient feature utilization. Besides, the methods require complex processing steps including empirical feature extraction and classifier design. To achieve efficient CAD detection, we propose the multiscale attention convolutional compression network (MACCN) based on clinical PCG dataset. Firstly, PCG dataset including 102 CAD subjects and 82 non-CAD subjects was established. Then, a multiscale convolution structure was developed to catch comprehensive heart sound features and a channel attention module was developed to enhance key features in multiscale attention convolutional block (MACB). Finally, a separate downsampling block was proposed to reduce feature losses. MACCN combining the blocks can automatically extract features without empirical and manual feature selection. It obtains good classification results with accuracy 93.43%, sensitivity 93.44%, precision 93.48%, and F1 score 93.42%. The study implies that MACCN performs effective PCG feature mining aiming for CAD detection. Further, it integrates feature extraction and classification and provides a simplified PCG processing case.

Analysis and recognition of post-exercise cardiac state based on heart sound features and cardiac troponin I.

PURPOSE: Excessive intensity exercises can bring irreversible damage to the heart. We explore whether heart sounds can evaluate cardiac function after high-intensity exercise and hope to prevent overtraining through the changes of heart sound in future training. METHODS: The study population consisted of 25 male athletes and 24 female athletes. All subjects were healthy and had no history of cardiovascular disease or family history of cardiovascular disease. The subjects were required to do high-intensity exercise for 3 days, with their blood sample and heart sound (HS) signals being collected and analysed before and after exercise. We then developed a Kernel extreme learning machine (KELM) model that can distinguish the state of heart by using the pre- and post-exercise data. RESULTS: There was no significant change in serum cardiac troponin I after 3 days of load cross-country running, which indicates that there was no myocardial injury after the race. The statistical analysis of time-domain characteristics and multi-fractal characteristic parameters of HS showed that the cardiac reserve capacity of the subjects was enhanced after the cross-country running, and the KELM is an effective classifier to recognize HS and the state of the heart after exercise. CONCLUSION: Through the results, we can draw the conclusion that this intensity of exercise will not cause profound damage to the athlete's heart. The findings of this study are of great significance for evaluating the condition of the heart with the proposed index of heart sound and prevention of excessive training that causes damage to the heart.

Comparison of different noncontrast computed tomographic markers for predicting early perihematomal edema expansion in patients with intracerebral hemorrhage.

OBJECTIVES: Noncontrast computed tomography (NCCT) imaging markers are associated with early perihematomal edema (PHE) growth. The aim of this study was to compare the predictive value of different NCCT markers in predicting early PHE expansion. METHODS: ICH patients who underwent baseline CT scan within 6 h of symptoms onset and follow-up CT scan within 36 h between July 2011 and March 2017 were included in this study. The predictive value of hypodensity, satellite sign, heterogeneous density, irregular shape, blend sign, black hole sign, island sign and expansion-prone hematoma for early perihematomal edema expansion were assessed, separately. RESULTS: 214 patients were included in our final analysis. After adjusting for ICH characteristics, hypodensity, blend sign, island sign and expansion-prone hematoma are still predictors of early perihematomal edema expansion in multivariable logistics regression analysis (all P < 0.05). The area under the receiver operating characteristic (ROC) curve of expansion-prone hematoma was significantly larger than the area under the ROC curve of hypodensity, blend sign and island sign in predicting PHE expansion (P = 0.003, P < 0.001 and P = 0.002, respectively). CONCLUSION: Compared with single NCCT imaging markers, expansion-prone hematoma seems to be optimal predictor for early PHE expansion than any single NCCT imaging marker.

Prediction of Early Perihematomal Edema Expansion Based on Noncontrast Computed Tomography Radiomics and Machine Learning in Intracerebral Hemorrhage.

OBJECTIVES: To investigate the predictive value of noncontrast computed tomography (NCCT) models based on radiomics features and machine learning for early perihematomal edema (PHE) expansion in patients with spontaneous intracerebral hemorrhage (ICH). METHODS: We retrospectively reviewed NCCT data from 214 patients with spontaneous ICH. All radiomics features were extracted from volume of interest of hematomas on admission scans. A total of 8 machine learning methods were applied for constructing models in the training and the test set. Receiver operating characteristic analysis and the areas under the curve were used to evaluate the predictive value. RESULTS: A total of 23 features were finally selected to establish models of early PHE expansion after feature screening. Patients were randomly assigned into training (n = 171) and test (n = 43) sets. The accuracy, sensitivity, and specificity in the test set were 72.1%, 90.0%, and 66.7% for the support vector machine model; 79.1%, 70.0%, and 84.4% for the k-nearest neighbor model; 88.4%, 90.0%, and 87.9% for the logistic regression model; 74.4%, 90.0%, and 69.7% for the extra tree model; 74.4%, 90.0%, and 69.7% for the extreme gradient boosting model; 83.7%, 100%, and 78.8% for the multilayer perceptron (MLP) model; 72.1%, 100%, and 65.6% for the light gradient boosting machine model; and 60.5%, 90.0%, and 53.1% for the random forest model, respectively. CONCLUSIONS: The MLP model seemed to be the best model for prediction of PHE expansion in patients with ICH. NCCT models based on radiomics features and machine learning could predict early PHE expansion and improve the discrimination of identify spontaneous intracerebral hemorrhage patients at risk of early PHE expansion.

Phonocardiogram transfer learning-based CatBoost model for diastolic dysfunction identification using multiple domain-specific deep feature fusion.

Left ventricular diastolic dyfunction detection is particularly important in cardiac function screening. This paper proposed a phonocardiogram (PCG) transfer learning-based CatBoost model to detect diastolic dysfunction noninvasively. The Short-Time Fourier Transform (STFT), Mel Frequency Cepstral Coefficients (MFCCs), S-transform and gammatonegram were utilized to perform four different representations of spectrograms for learning the representative patterns of PCG signals in two-dimensional image modality. Then, four pre-trained convolutional neural networks (CNNs) such as VGG16, Xception, ResNet50 and InceptionResNetv2 were employed to extract multiple domain-specific deep features from PCG spectrograms using transfer learning, respectively. Further, principal component analysis and linear discriminant analysis (LDA) were applied to different feature subsets, respectively, and then these different selected features are fused and fed into CatBoost for classification and performance comparison. Finally, three typical machine learning classifiers such as multilayer perceptron, support vector machine and random forest were employed to compared with CatBoost. The hyperparameter optimization of the investigated models was determined through grid search. The visualized result of the global feature importance showed that deep features extracted from gammatonegram by ResNet50 contributed most to classification. Overall, the proposed multiple domain-specific feature fusion based CatBoost model with LDA achieved the best performance with an area under the curve of 0.911, accuracy of 0.882, sensitivity of 0.821, specificity of 0.927, F1-score of 0.892 on the testing set. The PCG transfer learning-based model developed in this study could aid in diastolic dysfunction detection and could contribute to non-invasive evaluation of diastolic function.

DARU-Net: A dual attention residual U-Net for uterine fibroids segmentation on MRI.

PURPOSE: Uterine fibroid is the most common benign tumor in female reproductive organs. In order to guide the treatment, it is crucial to detect the location, shape, and size of the tumor. This study proposed a deep learning approach based on attention mechanisms to segment uterine fibroids automatically on preoperative Magnetic Resonance (MR) images. METHODS: The proposed method is based on U-Net architecture and integrates two attention mechanisms: channel attention of squeeze-and-excitation (SE) blocks with residual connections, spatial attention of pyramid pooling module (PPM). We did the ablation study to verify the performance of these two attention mechanisms module and compared DARU-Net with other deep learning methods. All experiments were performed on a clinical dataset consisting of 150 cases collected from our hospital. Among them, 120 cases were used as the training set, and 30 cases are used as the test set. After preprocessing and data augmentation, we trained the network and tested it on the test dataset. We evaluated segmentation performance through the Dice similarity coefficient (DSC), precision, recall, and Jaccard index (JI). RESULTS: The average DSC, precision, recall, and JI of DARU-Net reached 0.8066 ± 0.0956, 0.8233 ± 0.1255, 0.7913 ± 0.1304, and 0.6743 ± 0.1317. Compared with U-Net and other deep learning methods, DARU-Net was more accurate and stable. CONCLUSION: This work proposed an optimized U-Net with channel and spatial attention mechanisms to segment uterine fibroids on preoperative MR images. Results showed that DARU-Net was able to accurately segment uterine fibroids from MR images.

A combined non-enhanced CT radiomics and clinical variable machine learning model for differentiating benign and malignant sub-centimeter pulmonary solid nodules.

BACKGROUND: Radiomics has been used to predict pulmonary nodule (PN) malignancy. However, most of the studies focused on pulmonary ground-glass nodules. The use of computed tomography (CT) radiomics in pulmonary solid nodules, particularly sub-centimeter solid nodules, is rare. PURPOSE: This study aims to develop a radiomics model based on non-enhanced CT images that can distinguish between benign and malignant sub-centimeter pulmonary solid nodules (SPSNs, <1 cm). METHODS: The clinical and CT data of 180 SPSNs confirmed by pathology were analyzed retrospectively. All SPSNs were divided into two groups: training set (n = 144) and testing set (n = 36). From non-enhanced chest CT images, over 1000 radiomics features were extracted. Radiomics feature selection was performed using the analysis of variance and principal component analysis. The selected radiomics features were fed into a support vector machine (SVM) to develop a radiomics model. The clinical and CT characteristics were used to develop a clinical model. Associating non-enhanced CT radiomics features with clinical factors were used to develop a combined model using SVM. The performance was evaluated using the area under the receiver-operating characteristic curve (AUC). RESULTS: The radiomics model performed well in distinguishing between benign and malignant SPSNs, with an AUC of 0.913 (95% confidence interval [CI], 0.862-0.954) in the training set and an AUC of 0.877 (95% CI, 0.817-0.924) in the testing set. The combined model outperformed the clinical and radiomics models with an AUC of 0.940 (95% CI, 0.906-0.969) in the training set and an AUC of 0.903 (95% CI, 0.857-0.944) in the testing set. CONCLUSIONS: Radiomics features based on non-enhanced CT images can be used to differentiate SPSNs. The combined model, which included radiomics and clinical factors, had the best discrimination power between benign and malignant SPSNs.

Heart function grading evaluation based on heart sounds and convolutional neural networks.

Accurate and rapid cardiac function assessment is critical for disease diagnosis and treatment strategy. However, the current cardiac function assessment methods have their adaptability and limitations. Heart sounds (HS) can reflect changes in heart function. Therefore, HS signals were proposed to assess cardiac function, and a specially designed pruning convolutional neural network (CNN) was applied to recognize subjects' cardiac function at different levels in this paper. Firstly, the adaptive wavelet denoising algorithm and logistic regression based hidden semi-Markov model were utilized for signal denoising and segmentation. Then, the continuous wavelet transform (CWT) was employed to convert the preprocessed HS signals into spectra as input to the convolutional neural network, which can extract features automatically. Finally, the proposed method was compared with AlexNet, Resnet50, Xception, GhostNet and EfficientNet to verify the superiority of the proposed method. Through comprehensive comparison, the proposed approach achieves the best classification performance with an accuracy of 94.34%. The study indicates HS analysis is a non-invasive and effective method for cardiac function classification, which has broad research prospects.

Stratification of malignant renal neoplasms from cystic renal lesions using deep learning and radiomics features based on a stacking ensemble CT machine learning algorithm.

Using nephrographic phase CT images combined with pathology diagnosis, we aim to develop and validate a fusion feature-based stacking ensemble machine learning model to distinguish malignant renal neoplasms from cystic renal lesions (CRLs). This retrospective research includes 166 individuals with CRLs for model training and 47 individuals with CRLs in another institution for model testing. Histopathology results are adopted as diagnosis criterion. Nephrographic phase CT scans are selected to build the fusion feature-based machine learning algorithms. The pretrained 3D-ResNet50 CNN model and radiomics methods are selected to extract deep features and radiomics features, respectively. Fivefold cross-validated least absolute shrinkage and selection operator (LASSO) regression methods are adopted to identify the most discriminative candidate features in the development cohort. Intraclass correlation coefficients and interclass correlation coefficients are employed to evaluate feature's reproducibility. Pearson correlation coefficients for normal distribution features and Spearman's rank correlation coefficients for non-normal distribution features are used to eliminate redundant features. After that, stacking ensemble machine learning models are developed in the training cohort. The area under the receiver operator characteristic curve (ROC), calibration curve, and decision curve analysis (DCA) are adopted in the testing cohort to evaluate the performance of each model. The stacking ensemble machine learning algorithm reached excellent diagnostic performance in the testing dataset. The calibration plot shows good stability when using the stacking ensemble model. Net benefits presented by DCA are higher than the Bosniak 2019 version classification when employing any machine learning algorithm. The fusion feature-based machine learning algorithm accurately distinguishes malignant renal neoplasms from CRLs, which outperformed the Bosniak 2019 version classification, and proves to be more applicable for clinical decision-making.

3D island sign on computed tomography predicts early perihematomal edema expansion and poor outcome in patients with intracerebral hemorrhage.

BACKGROUND AND PURPOSE: To determine the prognostics significance of the computed tomography (CT) 3D island sign for predicting early perihematomal edema (PHE) expansion and poor functional outcome in patients presenting with intracerebral hemorrhage (ICH). METHODS: Between July 2011 and March 2017, patients with intracerebral hemorrhage who had undergone baseline CT within 6 h after ICH symptom onsets and follow-up CT in our hospital were included. Two different readers independently assessed the presence of 3D island sign on admission CT scan of each patient. Multivariable logistic regression analysis was used to analyze association between 3D island sign and early perihematomal edema expansion and poor functional outcome, separately. RESULTS: A total of 214 patients who met the inclusion criteria were included in our study, 3D island sign was observed in 60 patients (28.0 %) on admission CT scan. The multivariate logistic regression analysis demonstrated that baseline hematoma volume, time to baseline and follow-up CT scans and the presence of 3D island sign were predictors of early PHE expansion. After adjusting for age, baseline hematoma and edema volume, time to baseline and follow-up CT scans, GCS on admission, presence of intraventricular hemorrhage (IVH) and systolic blood pressure, the 3D island sign was an independently imaging marker for poor outcome (OR, 2.803; 95 % confidence interval, 1.189-6.609; P = 0.018). CONCLUSION: The 3D island sign in patients with intracerebral hemorrhage was a reliable predictor for early perihematomal edema expansion and poor functional outcome. It may serve as a potential therapeutic target for intervention.

Comparison of Different Machine Models Based on Multi-Phase Computed Tomography Radiomic Analysis to Differentiate Parotid Basal Cell Adenoma From Pleomorphic Adenoma.

Objective: This study explored the value of different radiomic models based on multiphase computed tomography in differentiating parotid pleomorphic adenoma (PA) and basal cell tumor (BCA) concerning the predominant phase and the optimal radiomic model. Methods: This study enrolled 173 patients with pathologically confirmed parotid tumors (training cohort: n=121; testing cohort: n=52). Radiomic features were extracted from the nonenhanced, arterial, venous, and delayed phases CT images. After dimensionality reduction and screening, logistic regression (LR), K-nearest neighbor (KNN) and support vector machine (SVM) were applied to develop radiomic models. The optimal radiomic model was selected by using ROC curve analysis. Univariate and multivariable logistic regression was performed to analyze clinical-radiological characteristics and to identify variables for developing a clinical model. A combined model was constructed by integrating clinical and radiomic features. Model performances were assessed by ROC curve analysis. Results: A total of 1036 radiomic features were extracted from each phase of CT images. Sixteen radiomic features were considered valuable by dimensionality reduction and screening. Among radiomic models, the SVM model of the arterial and delayed phases showed superior predictive efficiency and robustness (AUC, training cohort: 0.822, 0.838; testing cohort: 0.752, 0.751). The discriminatory capability of the combined model was the best (AUC, training cohort: 0.885; testing cohort: 0.834). Conclusions: The diagnostic performance of the arterial and delayed phases contributed more than other phases. However, the combined model demonstrated excellent ability to distinguish BCA from PA, which may provide a non-invasive and efficient method for clinical decision-making.

A multi-scale and multi-domain heart sound feature-based machine learning model for ACC/AHA heart failure stage classification.

Objective.Heart sounds can reflect detrimental changes in cardiac mechanical activity that are common pathological characteristics of chronic heart failure (CHF). The ACC/AHA heart failure (HF) stage classification is essential for clinical decision-making and the management of CHF. Herein, a machine learning model that makes use of multi-scale and multi-domain heart sound features was proposed to provide an objective aid for ACC/AHA HF stage classification.Approach.A dataset containing phonocardiogram (PCG) signals from 275 subjects was obtained from two medical institutions and used in this study. Complementary ensemble empirical mode decomposition and tunable-Q wavelet transform were used to construct self-adaptive sub-sequences and multi-level sub-band signals for PCG signals. Time-domain, frequency-domain and nonlinear feature extraction were then applied to the original PCG signal, heart sound sub-sequences and sub-band signals to construct multi-scale and multi-domain heart sound features. The features selected via the least absolute shrinkage and selection operator were fed into a machine learning classifier for ACC/AHA HF stage classification. Finally, mainstream machine learning classifiers, including least-squares support vector machine (LS-SVM), deep belief network (DBN) and random forest (RF), were compared to determine the optimal model.Main results. The results showed that the LS-SVM, which utilized a combination of multi-scale and multi-domain features, achieved better classification performance than the DBN and RF using multi-scale or/and multi-domain features alone or together, with average sensitivity, specificity, and accuracy of 0.821, 0.955 and 0.820 on the testing set, respectively.Significance.PCG signal analysis provides efficient measurement information regarding CHF severity and is a promising noninvasive method for ACC/AHA HF stage classification.

An automatic approach for heart failure typing based on heart sounds and convolutional recurrent neural networks.

Heart failure (HF) is a complex clinical syndrome that poses a major hazard to human health. Patients with different types of HF have great differences in pathogenesis and treatment options. Therefore, HF typing is of great significance for timely treatment of patients. In this paper, we proposed an automatic approach for HF typing based on heart sounds (HS) and convolutional recurrent neural networks, which provides a new non-invasive and convenient way for HF typing. Firstly, the collected HS signals were preprocessed with adaptive wavelet denoising. Then, the logistic regression based hidden semi-Markov model was utilized to segment HS frames. For the distinction between normal subjects and the HF patients with preserved ejection fraction or reduced ejection fraction, a model based on convolutional neural network and recurrent neural network was built. The model can automatically learn the spatial and temporal characteristics of HS signals. The results show that the proposed model achieved a superior performance with an accuracy of 97.64%. This study suggests the proposed method could be a useful tool for HF recognition and as a supplement for HF typing.

A CT-Based Radiomics Nomogram Integrated With Clinic-Radiological Features for Preoperatively Predicting WHO/ISUP Grade of Clear Cell Renal Cell Carcinoma.

OBJECTIVE: This study aims to develop and validate a CT-based radiomics nomogram integrated with clinic-radiological factors for preoperatively differentiating high-grade from low-grade clear cell renal cell carcinomas (CCRCCs). METHODS: 370 patients with complete clinical, pathological, and CT image data were enrolled in this retrospective study, and were randomly divided into training and testing sets with a 7:3 ratio. Radiomics features were extracted from nephrographic phase (NP) contrast-enhanced images, and then a radiomics model was constructed by the selected radiomics features using a multivariable logistic regression combined with the most suitable feature selection algorithm determined by the comparison among least absolute shrinkage and selection operator (LASSO), recursive feature elimination (RFE) and ReliefF. A clinical model was established using clinical and radiological features. A radiomics nomogram was constructed by integrating the radiomics signature and independent clinic-radiological features. Performance of these three models was assessed using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA). RESULTS: Using multivariate logistic regression analysis, three clinic-radiological features including intratumoral necrosis (OR=3.00, 95% CI=1.30-6.90, p=0.049), intratumoral angiogenesis (OR=3.28, 95% CI=1.22-8.78, p=0.018), and perinephric metastasis (OR=2.90, 95% CI=1.03-8.17, p=0.044) were found to be independent predictors of WHO/ISUP grade in CCRCC. Incorporating the above clinic-radiological predictors and radiomics signature constructed by LASSO, a CT-based radiomics nomogram was developed, and presented better predictive performance than clinic-radiological model and radiomics signature model, with an AUC of 0.891 (95% CI=0.832-0.962) and 0.843 (95% CI=0.718-0.975) in the training and testing sets, respectively. DCA indicated that the nomogram has potential clinical usefulness. CONCLUSION: The CT-based radiomics nomogram is a promising tool to predict WHO/ISUP grade of CCRCC preoperatively and noninvasively.

Quantitative Susceptibility Mapping-Derived Radiomic Features in Discriminating Multiple Sclerosis From Neuromyelitis Optica Spectrum Disorder.

Objectives: To implement a machine learning model using radiomic features extracted from quantitative susceptibility mapping (QSM) in discriminating multiple sclerosis (MS) from neuromyelitis optica spectrum disorder (NMOSD). Materials and Methods: Forty-seven patients with MS (mean age = 40.00 ± 13.72 years) and 36 patients with NMOSD (mean age = 42.14 ± 12.34 years) who underwent enhanced gradient-echo T2*-weighted angiography (ESWAN) sequence in 3.0-T MRI were included between April 2017 and October 2019. QSM images were reconstructed from ESWAN, and QSM-derived radiomic features were obtained from seven regions of interest (ROIs), including bilateral putamen, globus pallidus, head of the caudate nucleus, thalamus, substantia nigra, red nucleus, and dentate nucleus. A machine learning model (logistic regression) was applied to classify MS and NMOSD, which combined radiomic signatures and demographic information to assess the classification accuracy using the area under the receiver operating characteristic (ROC) curve (AUC). Results: The radiomics-only models showed better discrimination performance in almost all deep gray matter (DGM) regions than the demographic information-only model, with the highest AUC in DN of 0.902 (95% CI: 0.840-0.955). Moreover, the hybrid model combining radiomic signatures and demographic information showed the highest discrimination performance which achieved the AUC of 0.927 (95% CI: 0.871-0.984) with fivefold cross-validation. Conclusion: The hybrid model based on QSM and powered with machine learning has the potential to discriminate MS from NMOSD.