Automatic Urinary Stone Detection System for Abdominal Non-Enhanced CT Images Reduces the Burden on Radiologists.

To develop a fully automatic urinary stone detection system (kidney, ureter, and bladder) and to test it in a real clinical environment. The local institutional review board approved this retrospective single-center study that used non-enhanced abdominopelvic CT scans from patients admitted urology (uPatients) and emergency (ePatients). The uPatients were randomly divided into training and validation sets in a ratio of 3:1. We designed a cascade urinary stone map location-feature pyramid networks (USm-FPNs) and innovatively proposed a ureter distance heatmap method to estimate the ureter position on non-enhanced CT to further reduce the false positives. The performances of the system were compared using the free-response receiver operating characteristic curve and the precision-recall curve. This study included 811 uPatients and 356 ePatients. At stone level, the cascade detector USm-FPNs has the mean of false positives per scan (mFP) 1.88 with the sensitivity 0.977 in validation set, and mFP was further reduced to 1.18 with the sensitivity 0.977 after combining the ureter distance heatmap. At patient level, the sensitivity and precision were as high as 0.995 and 0.990 in validation set, respectively. In a real clinical set of ePatients (27.5% of patients contain stones), the mFP was 1.31 with as high as sensitivity 0.977, and the diagnostic time reduced by > 20% with the system help. A fully automatic detection system for entire urinary stones on non-enhanced CT scans was proposed and reduces obviously the burden on junior radiologists without compromising sensitivity in real emergency data.

Multitask prediction models for serous ovarian cancer by preoperative CT image assessments based on radiomics.

Objective: High-grade serous ovarian cancer (HGSOC) has the highest mortality rate among female reproductive system tumors. Accurate preoperative assessment is crucial for treatment planning. This study aims to develop multitask prediction models for HGSOC using radiomics analysis based on preoperative CT images. Methods: This study enrolled 112 patients diagnosed with HGSOC. Laboratory findings, including serum levels of CA125, HE-4, and NLR, were collected. Radiomic features were extracted from manually delineated ROI on CT images by two radiologists. Classification models were developed using selected optimal feature sets to predict R0 resection, lymph node invasion, and distant metastasis status. Model evaluation was conducted by quantifying receiver operating curves (ROC), calculating the area under the curve (AUC), De Long's test. Results: The radiomics models applied to CT images demonstrated superior performance in the testing set compared to the clinical models. The area under the curve (AUC) values for the combined model in predicting R0 resection were 0.913 and 0.881 in the training and testing datasets, respectively. De Long's test indicated significant differences between the combined and clinical models in the testing set (p = 0.003). For predicting lymph node invasion, the AUCs of the combined model were 0.868 and 0.800 in the training and testing datasets, respectively. The results also revealed significant differences between the combined and clinical models in the testing set (p = 0.002). The combined model for predicting distant metastasis achieved AUCs of 0.872 and 0.796 in the training and test datasets, respectively. The combined model displayed excellent agreement between observed and predicted results in predicting R0 resection, while the radiomics model demonstrated better calibration than both the clinical model and combined model in predicting lymph node invasion and distant metastasis. The decision curve analysis (DCA) for predicting R0 resection favored the combined model over both the clinical and radiomics models, whereas for predicting lymph node invasion and distant metastasis, DCA favored the radiomics model over both the clinical model and combined model. Conclusion: The identified radiomics signature holds potential value in preoperatively evaluating the R0, lymph node invasion and distant metastasis in patients with HGSC. The radiomics nomogram demonstrated the incremental value of clinical predictors for surgical outcome and metastasis estimation.

Artificial intelligence-based non-invasive tumor segmentation, grade stratification and prognosis prediction for clear-cell renal-cell carcinoma.

Due to the complicated histopathological characteristics of clear-cell renal-cell carcinoma (ccRCC), non-invasive prognosis before operative treatment is crucial in selecting the appropriate treatment. A total of 126 345 computerized tomography (CT) images from four independent patient cohorts were included for analysis in this study. We propose a V Bottleneck multi-resolution and focus-organ network (VB-MrFo-Net) using a cascade framework for deep learning analysis. The VB-MrFo-Net achieved better performance than VB-Net in tumor segmentation, with a Dice score of 0.87. The nuclear-grade prediction model performed best in the logistic regression classifier, with area under curve values from 0.782 to 0.746. Survival analysis revealed that our prediction model could significantly distinguish patients with high survival risk, with a hazard ratio (HR) of 2.49 [95% confidence interval (CI): 1.13-5.45, P = 0.023] in the General cohort. Excellent performance had also been verified in the Cancer Genome Atlas cohort, the Clinical Proteomic Tumor Analysis Consortium cohort, and the Kidney Tumor Segmentation Challenge cohort, with HRs of 2.77 (95%CI: 1.58-4.84, P = 0.0019), 3.83 (95%CI: 1.22-11.96, P = 0.029), and 2.80 (95%CI: 1.05-7.47, P = 0.025), respectively. In conclusion, we propose a novel VB-MrFo-Net for the renal tumor segmentation and automatic diagnosis of ccRCC. The risk stratification model could accurately distinguish patients with high tumor grade and high survival risk based on non-invasive CT images before surgical treatments, which could provide practical advice for deciding treatment options.

A Comprehensive Model Based on Dynamic Contrast-Enhanced Magnetic Resonance Imaging Can Better Predict the Preoperative Histological Grade of Breast Cancer Than a Radiomics Model.

Background: Histological grade is an important prognostic factor for patients with breast cancer and can affect clinical decision-making. From a clinical perspective, developing an efficient and non-invasive method for evaluating histological grading is desirable, facilitating improved clinical decision-making by physicians. This study aimed to develop an integrated model based on radiomics and clinical imaging features for preoperative prediction of histological grade invasive breast cancer. Methods: In this retrospective study, we recruited 211 patients with invasive breast cancer and randomly assigned them to either a training group (n=147) or a validation group (n=64) with a 7:3 ratio. Patients were classified as having low-grade tumors, which included grade I and II tumors, or high-grade tumors, which included grade III tumors. Three models were constructed based on basic clinical features, radiomics features, and the sum of the two. To assess diagnostic performance of the radiomics models, we employed measures such as receiver operating characteristic (ROC) curve, decision curve analysis (DCA), accuracy, sensitivity, and specificity, and the predictive performance of the three models was compared using the DeLong test and net reclassification improvement (NRI). Results: The area under the curve (AUC) of the clinical model, radiomics model, and comprehensive model was 0.682, 0.833, and 0.882 in the training set and 0.741, 0.751, and 0.836 in the validation set, respectively. NRI analysis confirmed that the combined model was better than the other two models in predicting the histological grade of breast cancer (NRI=21.4% in the testing cohort). Conclusion: Compared with the other models, the comprehensive model based on the combination of basic clinical features and radiomics features exhibits more significant potential for predicting histological grade and can better assist clinicians in optimal decision-making.

Automated Rib Fracture Detection on Chest X-Ray Using Contrastive Learning.

To develop a deep learning-based model for detecting rib fractures on chest X-Ray and to evaluate its performance based on a multicenter study. Chest digital radiography (DR) images from 18,631 subjects were used for the training, testing, and validation of the deep learning fracture detection model. We first built a pretrained model, a simple framework for contrastive learning of visual representations (simCLR), using contrastive learning with the training set. Then, simCLR was used as the backbone for a fully convolutional one-stage (FCOS) objective detection network to identify rib fractures from chest X-ray images. The detection performance of the network for four different types of rib fractures was evaluated using the testing set. A total of 127 images from Data-CZ and 109 images from Data-CH with the annotations for four types of rib fractures were used for evaluation. The results showed that for Data-CZ, the sensitivities of the detection model with no pretraining, pretrained ImageNet, and pretrained DR were 0.465, 0.735, and 0.822, respectively, and the average number of false positives per scan was five in all cases. For the Data-CH test set, the sensitivities of three different pretraining methods were 0.403, 0.655, and 0.748. In the identification of four fracture types, the detection model achieved the highest performance for displaced fractures, with sensitivities of 0.873 and 0.774 for the Data-CZ and Data-CH test sets, respectively, with 5 false positives per scan, followed by nondisplaced fractures, buckle fractures, and old fractures. A pretrained model can significantly improve the performance of the deep learning-based rib fracture detection based on X-ray images, which can reduce missed diagnoses and improve the diagnostic efficacy.

uRP: An integrated research platform for one-stop analysis of medical images.

Introduction: Medical image analysis is of tremendous importance in serving clinical diagnosis, treatment planning, as well as prognosis assessment. However, the image analysis process usually involves multiple modality-specific software and relies on rigorous manual operations, which is time-consuming and potentially low reproducible. Methods: We present an integrated platform - uAI Research Portal (uRP), to achieve one-stop analyses of multimodal images such as CT, MRI, and PET for clinical research applications. The proposed uRP adopts a modularized architecture to be multifunctional, extensible, and customizable. Results and Discussion: The uRP shows 3 advantages, as it 1) spans a wealth of algorithms for image processing including semi-automatic delineation, automatic segmentation, registration, classification, quantitative analysis, and image visualization, to realize a one-stop analytic pipeline, 2) integrates a variety of functional modules, which can be directly applied, combined, or customized for specific application domains, such as brain, pneumonia, and knee joint analyses, 3) enables full-stack analysis of one disease, including diagnosis, treatment planning, and prognosis assessment, as well as full-spectrum coverage for multiple disease applications. With the continuous development and inclusion of advanced algorithms, we expect this platform to largely simplify the clinical scientific research process and promote more and better discoveries.

Automated Prediction of Early Recurrence in Advanced Sinonasal Squamous Cell Carcinoma With Deep Learning and Multi-parametric MRI-based Radiomics Nomogram.

RATIONALE AND OBJECTIVES: Preoperative prediction of the recurrence risk in patients with advanced sinonasal squamous cell carcinoma (SNSCC) is critical for individualized treatment. To evaluate the predictive ability of radiomics signature (RS) based on deep learning and multiparametric MRI for the risk of 2-year recurrence in advanced SNSCC. MATERIALS AND METHODS: Preoperative MRI datasets were retrospectively collected from 265 SNSCC patients (145 recurrences) who underwent preoperative MRI, including T2-weighted (T2W), contrast-enhanced T1-weighted (T1c) sequences and diffusion-weighted (DW). All patients were divided into 165 training cohort and 70 test cohort. A deep learning segmentation model based on VB-Net was used to segment regions of interest (ROIs) for preoperative MRI and radiomics features were extracted from automatically segmented ROIs. Least absolute shrinkage and selection operator (LASSO) and logistic regression (LR) were applied for feature selection and radiomics score construction. Combined with meaningful clinicopathological predictors, a nomogram was developed and its performance was evaluated. In addition, X-title software was used to divide patients into high-risk or low-risk early relapse (ER) subgroups. Recurrence-free survival probability (RFS) was assessed for each subgroup. RESULTS: The radiomics score, T stage, histological grade and Ki-67 predictors were independent predictors. The segmentation models of T2WI, T1c, and apparent diffusion coefficient (ADC) sequences achieved Dice coefficients of 0.720, 0.727, and 0.756, respectively, in the test cohort. RS-T2, RS-T1c and RS-ADC were derived from single-parameter MRI. RS-Combined (combined with T2WI, T1c, and ADC features) was derived from multiparametric MRI and reached area under curve (AUC) and accuracy of 0.854 (0.749-0.927) and 74.3% (0.624-0.840), respectively, in the test cohort. The calibration curve and decision curve analysis (DCA) illustrate its value in clinical practice. Kaplan-Meier analysis showed that the 2-year RFS rate for low-risk patients was significantly greater than that for high-risk patients in both the training and testing cohorts (p < 0.001). CONCLUSION: Automated nomograms based on multi-sequence MRI help to predict ER in SNSCC patients preoperatively.

Fully Automatic Knee Joint Segmentation and Quantitative Analysis for Osteoarthritis from Magnetic Resonance (MR) Images Using a Deep Learning Model.

BACKGROUND We aimed to develop and evaluate a deep learning-based method for fully automatic segmentation of knee joint MR imaging and quantitative computation of knee osteoarthritis (OA)-related imaging biomarkers. MATERIAL AND METHODS This retrospective study included 843 volumes of proton density-weighted fat suppression MR imaging. A convolutional neural network segmentation method with multiclass gradient harmonized Dice loss was trained and evaluated on 500 and 137 volumes, respectively. To assess potential morphologic biomarkers for OA, the volumes and thickness of cartilage and meniscus, and minimal joint space width (mJSW) were automatically computed and compared between 128 OA and 162 control data. RESULTS The CNN segmentation model produced reasonably high Dice coefficients, ranging from 0.948 to 0.974 for knee bone compartments, 0.717 to 0.809 for cartilage, and 0.846 for both lateral and medial menisci. The OA-related biomarkers computed from automatic knee segmentation achieved strong correlation with those from manual segmentation: average intraclass correlations of 0.916, 0.899, and 0.876 for volume and thickness of cartilage, meniscus, and mJSW, respectively. Volume and thickness measurements of cartilage and mJSW were strongly correlated with knee OA progression. CONCLUSIONS We present a fully automatic CNN-based knee segmentation system for fast and accurate evaluation of knee joint images, and OA-related biomarkers such as cartilage thickness and mJSW were reliably computed and visualized in 3D. The results show that the CNN model can serve as an assistant tool for radiologists and orthopedic surgeons in clinical practice and basic research.

Value of Whole-Thyroid CT-Based Radiomics in Predicting Benign and Malignant Thyroid Nodules.

The objective of this research is to explore the value of whole-thyroid CT-based radiomics in predicting benign (noncancerous) and malignant thyroid nodules. The imaging and clinical data of 161 patients with thyroid nodules that were confirmed by pathology were retrospectively analyzed. The entire thyroid regions of interest (ROIs) were manually sketched for all 161 cases. After extracting CT radiomic features, the patients were divided into a training group (128 cases) and a test group (33 cases) according to the 4:1 ratio with stratified random sampling (fivefold cross validation). All the data were normalized by the maximum absolute value and screened using selection operator regression analysis and K best. The data generation model was trained by logistic regression. The effectiveness of the model in differentiating between benign and malignant thyroid nodules was validated by a receiver operating characteristic (ROC) curve. After data grouping, eigenvalue screening, and data training, the logistic regression model with the maximum absolute value normalized was constructed. For the training group, the area under the ROC curve (AUC) was 94.4% (95% confidence interval: 0.941-0.977); the sensitivity and specificity were 89.7% and 86.7%, respectively; and the diagnostic accuracy was 87.6%. For the test group, the AUC was 94.2% (95% confidence interval: 0.881-0.999); the sensitivity and specificity were 89.4% and 86.8%, respectively; and the diagnostic accuracy was 87.6%. The CT radiomic model of the entire thyroid gland is highly efficient in differentiating between benign and malignant thyroid nodules.

Comparable Performance of Deep Learning-Based to Manual-Based Tumor Segmentation in KRAS/NRAS/BRAF Mutation Prediction With MR-Based Radiomics in Rectal Cancer.

Radiomic features extracted from segmented tumor regions have shown great power in gene mutation prediction, while deep learning-based (DL-based) segmentation helps to address the inherent limitations of manual segmentation. We therefore investigated whether deep learning-based segmentation is feasible in predicting KRAS/NRAS/BRAF mutations of rectal cancer using MR-based radiomics. In this study, we proposed DL-based segmentation models with 3D V-net architecture. One hundred and eight patients' images (T2WI and DWI) were collected for training, and another 94 patients' images were collected for validation. We evaluated the DL-based segmentation manner and compared it with the manual-based segmentation manner through comparing the gene prediction performance of six radiomics-based models on the test set. The performance of the DL-based segmentation was evaluated by Dice coefficients, which are 0.878 ± 0.214 and 0.955 ± 0.055 for T2WI and DWI, respectively. The performance of the radiomics-based model in gene prediction based on DL-segmented VOI was evaluated by AUCs (0.714 for T2WI, 0.816 for DWI, and 0.887 for T2WI+DWI), which were comparable to that of corresponding manual-based VOI (0.637 for T2WI, P=0.188; 0.872 for DWI, P=0.181; and 0.906 for T2WI+DWI, P=0.676). The results showed that 3D V-Net architecture could conduct reliable rectal cancer segmentation on T2WI and DWI images. All-relevant radiomics-based models presented similar performances in KRAS/NRAS/BRAF prediction between the two segmentation manners.

Peritumoral Dilation Radiomics of Gadoxetate Disodium-Enhanced MRI Excellently Predicts Early Recurrence of Hepatocellular Carcinoma without Macrovascular Invasion After Hepatectomy.

BACKGROUND: Whether peritumoral dilation radiomics can excellently predict early recrudescence (≤2 years) in hepatocellular carcinoma (HCC) remains unclear. METHODS: Between March 2012 and June 2018, 323 pathologically confirmed HCC patients without macrovascular invasion, who underwent liver resection and preoperative gadoxetate disodium (Gd-EOB-DTPA) MRI, were consecutively recruited into this study. Multivariate logistic regression identified independent clinicoradiologic predictors of 2-year recrudescence. Peritumoral dilation (tumor and peritumoral zones within 1cm) radiomics extracted features from 7-sequence images for modeling and achieved average but robust predictive performance through 5-fold cross validation. Independent clinicoradiologic predictors were then incorporated with the radiomics model for constructing a comprehensive nomogram. The predictive discrimination was quantified with the area under the receiver operating characteristic curve (AUC) and net reclassification improvement (NRI). RESULTS: With the median recurrence-free survival (RFS) reaching 60.43 months, 28.2% (91/323) and 16.4% (53/323) patients suffered from early and delay relapse, respectively. Microvascular invasion, tumor size >5 cm, alanine aminotransferase >50 U/L, γ-glutamyltransferase >60 U/L, prealbumin ≤250 mg/L, and peritumoral enhancement independently impaired 2-year RFS in the clinicoradiologic model with AUC of 0.694 (95% CI 0.628-0.760). Nevertheless, these indexes were paucity of robustness (P >0.05) when integrating with 38 most recurrence-related radiomics signatures for developing the comprehensive nomogram. The peritumoral dilation radiomics-the ultimate prediction model yielded satisfactory mean AUCs (training cohort: 0.939, 95% CI 0.908-0.973; validation cohort: 0.842, 95% CI 0.736-0.951) after 5-fold cross validation and fitted well with the actual relapse status in the calibration curve. Besides, our radiomics model obtained the best clinical net benefits, with significant improvements of NRI (35.9%-66.1%, P <0.001) versus five clinical algorithms: the clinicoradiologic model, the tumor-node-metastasis classification, the Barcelona Clinic Liver Cancer stage, the preoperative and postoperative risks of Early Recurrence After Surgery for Liver tumor. CONCLUSION: Gd-EOB-DTPA MRI-based peritumoral dilation radiomics is a potential preoperative biomarker for early recurrence of HCC patients without macrovascular invasion.

A cascade and heterogeneous neural network for CT pulmonary nodule detection and its evaluation on both phantom and patient data.

Screening of pulmonary nodules in computed tomography (CT) is crucial for early diagnosis and treatment of lung cancer. Although computer-aided diagnosis (CAD) systems have been designed to assist radiologists to detect nodules, fully automated detection is still challenging due to variations in nodule size, shape, and density. In this paper, we first propose a fully automated nodule detection method using a cascade and heterogeneous neural network trained on chest CT images of 12155 patients, then evaluate the performance by using phantom (828 CT images) and clinical datasets (2640 CT images) scanned with different imaging parameters. The nodule detection network employs two feature pyramid networks (FPNs) and a classification network (BasicNet). The first FPN is trained to achieve high sensitivity for nodule detection, and the second FPN refines the candidates for false positive reduction (FPR). Then, a BasicNet is combined with the second FPR to classify the candidates into either nodules or non-nodules for the final refinement. This study investigates the performance of nodule detection of solid and ground-glass nodules in phantom and patient data scanned with different imaging parameters. The results show that the detection of the solid nodules is robust to imaging parameters, and for GGO detection, reconstruction methods "iDose4-YA" and "STD-YA" achieve better performance. For thin-slice images, higher performance is achieved across different nodule sizes with reconstruction method "iDose4-STD". For 5 mm slice thickness, the best choice is the reconstruction method "iDose4-YA" for larger nodules (>5 mm). Overall, the reconstruction method "iDose4-YA" is suggested to achieve the best balanced results for both solid and GGO nodules.