Parallel CNN-Deep Learning Clinical-Imaging Signature for Assessing Pathologic Grade and Prognosis of Soft Tissue Sarcoma Patients.

BACKGROUND: Traditional biopsies pose risks and may not accurately reflect soft tissue sarcoma (STS) heterogeneity. MRI provides a noninvasive, comprehensive alternative. PURPOSE: To assess the diagnostic accuracy of histological grading and prognosis in STS patients when integrating clinical-imaging parameters with deep learning (DL) features from preoperative MR images. STUDY TYPE: Retrospective/prospective. POPULATION: 354 pathologically confirmed STS patients (226 low-grade, 128 high-grade) from three hospitals and the Cancer Imaging Archive (TCIA), divided into training (n = 185), external test (n = 125), and TCIA cohorts (n = 44). 12 patients (6 low-grade, 6 high-grade) were enrolled into prospective validation cohort. FIELD STRENGTH/SEQUENCE: 1.5 T and 3.0 T/Unenhanced T1-weighted and fat-suppressed-T2-weighted. ASSESSMENT: DL features were extracted from MR images using a parallel ResNet-18 model to construct DL signature. Clinical-imaging characteristics included age, gender, tumor-node-metastasis stage and MRI semantic features (depth, number, heterogeneity at T1WI/FS-T2WI, necrosis, and peritumoral edema). Logistic regression analysis identified significant risk factors for the clinical model. A DL clinical-imaging signature (DLCS) was constructed by incorporating DL signature with risk factors, evaluated for risk stratification, and assessed for progression-free survival (PFS) in retrospective cohorts, with an average follow-up of 23 ± 22 months. STATISTICAL TESTS: Logistic regression, Cox regression, Kaplan-Meier curves, log-rank test, area under the receiver operating characteristic curve (AUC)，and decision curve analysis. A P-value <0.05 was considered significant. RESULTS: The AUC values for DLCS in the external test, TCIA, and prospective test cohorts (0.834, 0.838, 0.819) were superior to clinical model (0.662, 0.685, 0.694). Decision curve analysis showed that the DLCS model provided greater clinical net benefit over the DL and clinical models. Also, the DLCS model was able to risk-stratify patients and assess PFS. DATA CONCLUSION: The DLCS exhibited strong capabilities in histological grading and prognosis assessment for STS patients, and may have potential to aid in the formulation of personalized treatment plans. TECHNICAL EFFICACY: Stage 2.

LNAS: a clinically applicable deep-learning system for mediastinal enlarged lymph nodes segmentation and station mapping without regard to the pathogenesis using unenhanced CT images.

BACKGROUND: The accurate identification and evaluation of lymph nodes by CT images is of great significance for disease diagnosis, treatment, and prognosis. PURPOSE: To assess the lymph nodes' segmentation, size, and station by artificial intelligence (AI) for unenhanced chest CT images and evaluate its value in clinical scenarios. MATERIAL AND METHODS: This retrospective study proposed an end-to-end Lymph Nodes Analysis System (LNAS) consisting of three models: the Lymph Node Segmentation model (LNS), the Mediastinal Organ Segmentation model (MOS), and the Lymph Node Station Registration model (LNR). We selected a healthy chest CT image as the template image and annotated 14 lymph node station masks according to the IASLC to build the lymph node station mapping template. The exact contours and stations of the lymph nodes were annotated by two junior radiologists and reviewed by a senior radiologist. Patients aged 18 and above, who had undergone unenhanced chest CT and had at least one suspicious enlarged mediastinal lymph node in imaging reports, were included. Exclusions were patients who had thoracic surgeries in the past 2 weeks or artifacts on CT images affecting lymph node observation by radiologists. The system was trained on 6725 consecutive chest CTs that from Tianjin Medical University General Hospital, among which 6249 patients had suspicious enlarged mediastinal lymph nodes. A total of 519 consecutive chest CTs from Qilu Hospital of Shandong University (Qingdao) were used for external validation. The gold standard for each CT was determined by two radiologists and reviewed by one senior radiologist. RESULTS: The patient-level sensitivity of the LNAS system reached of 93.94% and 92.89% in internal and external test dataset, respectively. And the lesion-level sensitivity (recall) reached 89.48% and 85.97% in internal and external test dataset. For man-machine comparison, AI significantly apparently shortened the average reading time (p < 0.001) and had better lesion-level and patient-level sensitivities. CONCLUSION: AI improved the sensitivity lymph node segmentation by radiologists with an advantage in reading time.

Nomogram Using CT Radiomics Features for Differentiation of Pneumonia-Type Invasive Mucinous Adenocarcinoma and Pneumonia: Multicenter Development and External Validation Study.

BACKGROUND. Pneumonia-type invasive mucinous adenocarcinoma (IMA) and pneumonia show overlapping chest CT features as well as overlapping clinical characteristics. OBJECTIVE. The purpose of our study was to develop and validate a nomogram combining clinical and CT-based radiomics features to differentiate pneumonia-type IMA and pneumonia. METHODS. This retrospective study included 314 patients (172 men, 142 women; mean age, 60.3 ± 14.5 [SD] years) from six hospitals who underwent noncontrast chest CT showing consolidation and were diagnosed with pneumonia-type IMA (n = 106) or pneumonia (n = 208). Patients from three hospitals formed a training set (n = 195) and a validation set (n = 50), and patients from the other three hospitals formed the external test set (n = 69). A model for predicting pneumonia-type IMA was built using clinical characteristics that were significant independent predictors of this diagnosis. Radiomics features were extracted from CT images by placing ROIs on areas of consolidation, and a radiomics signature of pneumonia-type IMA was constructed. A nomogram for predicting pneumonia-type IMA was constructed that combined features in the clinical model and the radiomics signature. Two cardiothoracic radiologists independently reviewed CT images in the external test set to diagnose pneumonia-type IMA. Diagnostic performance was compared among models and radiologists. Decision curve analysis (DCA) was performed. RESULTS. The clinical model included fever and family history of lung cancer. The radiomics signature included 15 radiomics features. DCA showed higher overall net benefit from the nomogram than from the clinical model. In the external test set, AUC was higher for the nomogram (0.85) than for the clinical model (0.71, p = .01), radiologist 1 (0.70, p = .04), and radiologist 2 (0.67, p = .01). In the external test set, the nomogram had sensitivity of 46.9%, specificity of 94.6%, and accuracy of 72.5%. CONCLUSION. The nomogram combining clinical variables and CT-based radiomics features outperformed the clinical model and two cardiothoracic radiologists in differentiating pneumonia-type IMA from pneumonia. CLINICAL IMPACT. The findings support potential clinical use of the nomogram for diagnosing pneumonia-type IMA in patients with consolidation on chest CT.

Glenoid segmentation from computed tomography scans based on a 2-stage deep learning model for glenoid bone loss evaluation.

BACKGROUND: The best-fitting circle drawn by computed tomography (CT) reconstruction of the en face view of the glenoid bone to measure the bone defect is widely used in clinical application. However, there are still some limitations in practical application, which can prevent the achievement of accurate measurements. This study aimed to accurately and automatically segment the glenoid from CT scans based on a 2-stage deep learning model and to quantitatively measure the glenoid bone defect. MATERIALS AND METHODS: Patients who were referred to our institution between June 2018 and February 2022 were retrospectively reviewed. The dislocation group consisted of 237 patients with a history of ≥2 unilateral shoulder dislocations within 2 years. The control group consisted of 248 individuals with no history of shoulder dislocation, shoulder developmental deformity, or other disease that may lead to abnormal morphology of the glenoid. All patients underwent CT examination with a 1-mm slice thickness and a 1-mm increment, including complete imaging of the bilateral glenoid. A residual neural network (ResNet) location model and a U-Net bone segmentation model were constructed to develop an automated segmentation model for the glenoid from CT scans. The data set was randomly divided into training (201 of 248) and test (47 of 248) data sets of control-group data and training (190 of 237) and test (47 of 237) data sets of dislocation-group data. The accuracy of the stage 1 (glenoid location) model, the mean intersection-over-union value of the stage 2 (glenoid segmentation) model, and the glenoid volume error were used to assess the performance of the model. The R2 value and Lin concordance correlation coefficient were used to assess the correlation between the prediction and the gold standard. RESULTS: A total of 73,805 images were obtained after the labeling process, and each image was composed of CT images of the glenoid and its corresponding mask. The average overall accuracy of stage 1 was 99.28%; the average mean intersection-over-union value of stage 2 was 0.96. The average glenoid volume error between the predicted and true values was 9.33%. The R2 values of the predicted and true values of glenoid volume and glenoid bone loss (GBL) were 0.87 and 0.91, respectively. The Lin concordance correlation coefficient value of the predicted and true values of glenoid volume and GBL were 0.93 and 0.95, respectively. CONCLUSION: The 2-stage model in this study showed a good performance in glenoid bone segmentation from CT scans and could quantitatively measure GBL, providing a data reference for subsequent clinical treatment.

Deep learning radiomic nomogram to predict recurrence in soft tissue sarcoma: a multi-institutional study.

OBJECTIVES: To evaluate the performance of a deep learning radiomic nomogram (DLRN) model at predicting tumor relapse in patients with soft tissue sarcomas (STS) who underwent surgical resection. METHODS: In total, 282 patients who underwent MRI and resection for STS at three independent centers were retrospectively enrolled. In addition, 113 of the 282 patients received additional contrast-enhanced MRI scans. We separated the participants into a development cohort and an external test cohort. The development cohort consisted of patients from one center and the external test cohort consisted of patients from two other centers. Two MRI-based DLRNs for prediction of tumor relapse after resection of STS were established. We universally tested the DLRNs and compared them with other prediction models constructed by using widespread adopted predictors (i.e., staging systems and Ki67) instead of radiomics features. RESULTS: The DLRN1 model incorporated plain MRI-based radiomics signature into the clinical data, and the DLRN2 model integrated radiomics signature extracted from plain and contrast-enhanced MRI with the clinical predictors. Across both study sets, the two MRI-based DLRNs had relatively better prognostic capability (C index ≥ 0.721 and median AUC ≥ 0.746; p < 0.05 compared with most other models and predictors) and less opportunity for prediction error (integrated Brier score ≤ 0.159). The decision curve analysis indicates that the DLRNs have greater benefits than staging systems, Ki67, and other models. We selected appropriate cutoff values for the DLRNs to divide STS recurrence into three risk strata (low, medium, and high) and calculated those groups' cumulative risk rates. CONCLUSION: The DLRNs were shown to be a reliable and externally validated tool for predicting STS recurrence by comparing with other prediction models. KEY POINTS: • The prediction of a high recurrence rate of STS before emergence of local recurrence can help to determine whether more active treatment should be implemented. • Two MRI-based DLRNs for prediction of tumor relapse were shown to be a reliable and externally validated tool for predicting STS recurrence. • We used the DLRNs to divide STS recurrence into three risk strata (low, medium, and high) to facilitate more targeted postoperative management in the clinic.

A multi-stage ensemble network system to diagnose adolescent idiopathic scoliosis.

OBJECTIVES: To develop a deep learning algorithm to automatically evaluate and diagnose scoliosis on full spinal X-ray images. METHODS: This retrospective study collected full spinal X-ray images (anteroposterior) from four hospital databases from January 1, 2018, to March 31, 2021. The data were divided into training and validation sets. Full spinal X-ray images for external validation were independently collected at one hospital from April 1, 2021, to June 30, 2021. Model effectiveness was validated with a public dataset. Statistical software R was used to analyze the accuracy and sensitivity of the model curvature and anatomical balance parameters and assess interrater consistency. RESULTS: This study included 788 and 185 training and test datasets, respectively. The accuracy and recall of the algorithm model for the Cobb angle, apical vertebrae (AV), upper vertebrae, and lower vertebrae were 89.36%, 85.71%, 77.2%, and 80.24% and 97.35%, 93.38%, 84.11%, and 87.42%, respectively. The symmetric mean absolute percentage error at the Cobb angle was 5.99%, and the automatic measurement time was 1.7 s. The mean absolute error values of the Cobb angle and the distances between the center sacral vertical line and AV and C7 plumb line were 1.07° and 1.12 and 1.38 mm, respectively. Statistical analysis confirmed that the Cobb angle results were in good agreement with the gold standard (interclass coefficients of 0.996, 0.978, and 0.825; p < 0.001). CONCLUSION: Our deep learning algorithm model had high sensitivity and accuracy for scoliosis, which could help radiologists improve their diagnostic efficiency. KEY POINTS: • Our deep learning algorithm model had high sensitivity and accuracy for scoliosis, which could help radiologists improve their diagnostic efficiency. • Multi-center validation data were used in this study to guarantee the reliability of the research. • Algorithmic model measures 200 times faster than radiologists.

A Facile machine learning multi-classification model for Streptococcus agalactiae clonal complexes.

BACKGROUND: The clinical significance of group B streptococcus (GBS) was different among different clonal complexes (CCs), accurate strain typing of GBS would facilitate clinical prognostic evaluation, epidemiological investigation and infection control. The aim of this study was to construct a practical and facile CCs prediction model for S. agalactiae. METHODS: A total of 325 non-duplicated GBS strains were collected from clinical samples in Xinhua Hospital, Shanghai, China. Multilocus sequence typing (MLST) method was used for molecular classification, the results were analyzed to derive CCs by Bionumeric 8.0 software. Antibiotic susceptibility test was performed using Vitek-2 Compact system combined with K-B method. Multiplex PCR method was used for serotype identification. A total of 45 virulence genes associated with adhesion, invasion, immune evasion were detected by PCR method and electrophoresis. Three types of features, including antibiotic susceptibility (A), serotypes (S) and virulence genes (V) tests, and XGBoost algorithm was established to develop multi-class CCs identification models. The performance of proposed models was evaluated by the receiver operating characteristic curve (ROC). RESULTS: The 325 GBS were divided into 47 STs, and then calculated into 7 major CCs, including CC1, CC10, CC12, CC17, CC19, CC23, CC24. A total of 18 features in three kinds of tests (A, S, V) were significantly different from each CC. The model based on all the features (S&A&V) performed best with AUC 0.9536. The model based on serotype and antibiotic resistance (S&A) only enrolled 5 weighed features, performed well in predicting CCs with mean AUC 0.9212, and had no statistical difference in predicting CC10, CC12, CC17, CC19, CC23 and CC24 when compared with S&A&V model (all p > 0.05). CONCLUSIONS: The S&A model requires least parameters while maintaining a high accuracy and predictive power of CCs prediction. The established model could be used as a promising tool to classify the GBS molecular types, and suggests a substantive improvement in clinical application and epidemiology surveillance in GBS phenotyping.

Prediction of the degree of pathological differentiation in tongue squamous cell carcinoma based on radiomics analysis of magnetic resonance images.

BACKGROUND: Tongue squamous cell carcinoma (TSCC) is one of the most difficult malignancies to control. It displays particular and aggressive behaviour even at an early stage. The purpose of this paper is to explore the value of radiomics based on magnetic resonance fat-suppressed T2-weighted images in predicting the degree of pathological differentiation of TSCC. METHODS: Retrospective analysis of 127 patients with TSCC who were randomly divided into a primary cohort and a test cohort, including well-differentiated, moderately differentiated and poorly differentiated. The tumour regions were manually labelled in fat-suppressed T2-weighted imaging (FS-T2WI), and PyRadiomics was used to extract radiomics features. The radiomics features were then selected by the least absolute shrinkage and selection operator (LASSO) method. The model was established by the logistic regression classifier using a 5-fold cross-validation method, applied to all data and evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity and specificity. RESULTS: In total, 1132 features were extracted, and seven features were selected for modelling. The AUC in the logistic regression model for well-differentiated TSCC was 0.90 with specificity and precision values of 0.92 and 0.78, respectively, and the sensitivity for poorly differentiated TSCC was 0.74. CONCLUSIONS: The MRI-based radiomics signature could discriminate between well-differentiated, moderately differentiated and poorly differentiated TSCC and might be used as a biomarker for preoperative grading.

Multiparametric MRI-based radiomic nomogram for predicting HER-2 2+ status of breast cancer.

Objective: To explore the application of multiparametric MRI-based radiomic nomogram for assessing HER-2 2+ status of breast cancer (BC). Methods: Patients with pathology-proven HER-2 2+ invasive BC, who underwent preoperative MRI were divided into training (72 patients, 21 HER-2-positive and 51 HER-2-negative) and validation (32 patients, 9 HER-2-positive and 23 HER-2-negative) sets by randomization. All were classified as HER-2 2+ FISH-positive (HER-2-positive) or -negative (HER-2-negative) according to IHC and FISH. The 3D VOI was drawn on MR images by two radiologists. ADC, T2WI, and DCE images were analyzed separately to extract features (n = 1906). L1 regularization, F-test, and other methods were used to reduce dimensionality. Binary radiomics prediction models using features from single or combined imaging sequences were constructed using logistic regression (LR) classifier then and validated on a validation dataset. To build a radiomics nomogram, multivariate LR analysis was conducted to identify independent indicators. An evaluation of the model's predictive efficacy was made using AUC. Results: On the basis of combined ADC, T2WI, and DCE images, ten radiomic features were extracted following feature dimensionality reduction. There was superior diagnostic efficiency of radiomic signature using all three sequences compared to either one or two sequences (AUC for training group: 0.883; AUC for validation group: 0.816). Based on multivariate LR analysis, radiomic signature and peritumoral edema were independent predictors for identifying HER-2 2 +. In both training and validation datasets, nomograms combining peritumoral edema and radiomics signature demonstrated an effective discrimination (AUCs were respectively 0.966 and 0. 884). Conclusion: The nomogram that incorporated peritumoral edema and multiparametric MRI-based radiomic signature can be used to effectively predict the HER-2 2+ status of BC.

Deep Learning for Chest X-ray Diagnosis: Competition Between Radiologists with or Without Artificial Intelligence Assistance.

This study aimed to assess the performance of a deep learning algorithm in helping radiologist achieve improved efficiency and accuracy in chest radiograph diagnosis. We adopted a deep learning algorithm to concurrently detect the presence of normal findings and 13 different abnormalities in chest radiographs and evaluated its performance in assisting radiologists. Each competing radiologist had to determine the presence or absence of these signs based on the label provided by the AI. The 100 radiographs were randomly divided into two sets for evaluation: one without AI assistance (control group) and one with AI assistance (test group). The accuracy, false-positive rate, false-negative rate, and analysis time of 111 radiologists (29 senior, 32 intermediate, and 50 junior) were evaluated. A radiologist was given an initial score of 14 points for each image read, with 1 point deducted for an incorrect answer and 0 points given for a correct answer. The final score for each doctor was automatically calculated by the backend calculator. We calculated the mean scores of each radiologist in the two groups (the control group and the test group) and calculated the mean scores to evaluate the performance of the radiologists with and without AI assistance. The average score of the 111 radiologists was 597 (587-605) in the control group and 619 (612-626) in the test group (P < 0.001). The time spent by the 111 radiologists on the control and test groups was 3279 (2972-3941) and 1926 (1710-2432) s, respectively (P < 0.001). The performance of the 111 radiologists in the two groups was evaluated by the area under the receiver operating characteristic curve (AUC). The radiologists showed better performance on the test group of radiographs in terms of normal findings, pulmonary fibrosis, heart shadow enlargement, mass, pleural effusion, and pulmonary consolidation recognition, with AUCs of 1.0, 0.950, 0.991, 1.0, 0.993, and 0.982, respectively. The radiologists alone showed better performance in aortic calcification (0.993), calcification (0.933), cavity (0.963), nodule (0.923), pleural thickening (0.957), and rib fracture (0.987) recognition. This competition verified the positive effects of deep learning methods in assisting radiologists in interpreting chest X-rays. AI assistance can help to improve both the efficacy and efficiency of radiologists.

Machine learning based on radiomics features combing B-mode transrectal ultrasound and contrast-enhanced ultrasound to improve peripheral zone prostate cancer detection.

PURPOSE: To construct machine learning models based on radiomics features combing conventional transrectal ultrasound (B-mode) and contrast-enhanced ultrasound (CEUS) to improve prostate cancer (PCa) detection in peripheral zone (PZ). METHODS: A prospective study of 166 men (72 benign, 94 malignant lesions) with targeted biopsy-confirmed pathology who underwent B-mode and CEUS examinations was performed. Risk factors, including age, serum total prostate-specific antigen (tPSA), free PSA (fPSA), f/t PSA, prostate volume and prostate-specific antigen density (PSAD), were collected. Time-intensity curves were obtained using SonoLiver software for all lesions in regions of interest. Four parameters were collected as risk factors: the maximum intensity (IMAX), rise time (RT), time to peak (TTP), and mean transit time (MTT). Radiomics features were extracted from the target lesions from B-mode and CEUS imaging. Multivariable logistic regression analysis was used to construct the model. RESULTS: A total of 3306 features were extracted from seven categories. Finally, 32 features were screened out from radiomics models. Five models were developed to predict PCa: the B-mode radiomics model (B model), CEUS radiomics model (CEUS model), B-CEUS combined radiomics model (B-CEUS model), risk factors model, and risk factors-radiomics combined model (combined model). Age, PSAD, tPSA, and RT were significant independent predictors in discriminating benign and malignant PZ lesions (P < 0.05). The risk factors model combing these four predictors showed better discrimination in the validation cohort (area under the curve [AUC], 0.84) than the radiomics images (AUC, 0.79 on B model; AUC, 0.78 on CEUS model; AUC, 0.83 on B-CEUS model), and the combined model (AUC: 0.89) achieved the greatest predictive efficacy. CONCLUSION: The prediction model including B-mode and CEUS radiomics signatures and risk factors represents a promising diagnostic tool for PCa detection in PZ, which may contribute to clinical decision-making.

A simple and effective machine learning model for predicting the stability of intracranial aneurysms using CT angiography.

Background: It is vital to accurately and promptly distinguish unstable from stable intracranial aneurysms (IAs) to facilitate treatment optimization and avoid unnecessary treatment. The aim of this study is to develop a simple and effective predictive model for the clinical evaluation of the stability of IAs. Methods: In total, 1,053 patients with 1,239 IAs were randomly divided the dataset into training (70%) and internal validation (30%) datasets. One hundred and ninety seven patients with 229 IAs from another hospital were evaluated as an external validation dataset. The prediction models were developed using machine learning based on clinical information, manual parameters, and radiomic features. In addition, a simple model for predicting the stability of IAs was developed, and a nomogram was drawn for clinical use. Results: Fourteen machine learning models exhibited excellent classification performance. Logistic regression Model E (clinical information, manual parameters, and radiomic shape features) had the highest AUC of 0.963 (95% CI 0.943-0.980). Compared to manual parameters, radiomic features did not significantly improve the identification of unstable IAs. In the external validation dataset, the simplified model demonstrated excellent performance (AUC = 0.950) using only five manual parameters. Conclusion: Machine learning models have excellent potential in the classification of unstable IAs. The manual parameters from CTA images are sufficient for developing a simple and effective model for identifying unstable IAs.

Corrigendum: Radiomic machine learning and external validation based on 3.0T mpMRI for prediction of intraductal carcinoma of prostate with different proportion.

[This corrects the article DOI: 10.3389/fonc.2022.934291.].

Compare deep learning model and conventional logistic regression model for the identification of unstable saccular intracranial aneurysms in computed tomography angiography.

Background: It is crucial to distinguish unstable from stable intracranial aneurysms (IAs) as early as possible to derive optimal clinical decision-making for further treatment or follow-up. The aim of this study was to investigate the value of a deep learning model (DLM) in identifying unstable IAs from computed tomography angiography (CTA) images and to compare its discriminatory ability with that of a conventional logistic regression model (LRM). Methods: From August 2011 to May 2021, a total of 1,049 patients with 681 unstable IAs and 556 stable IAs were retrospectively analyzed. IAs were randomly divided into training (64%), internal validation (16%), and test sets (20%). Convolutional neural network (CNN) analysis and conventional logistic regression (LR) were used to predict which IAs were unstable. The area under the curve (AUC), sensitivity, specificity and accuracy were calculated to evaluate the discriminating ability of the models. One hundred and ninety-seven patients with 229 IAs from Banan Hospital were used for external validation sets. Results: The conventional LRM showed 11 unstable risk factors, including clinical and IA characteristics. The LRM had an AUC of 0.963 [95% confidence interval (CI): 0.941-0.986], a sensitivity, specificity and accuracy on the external validation set of 0.922, 0.906, and 0.913, respectively, in predicting unstable IAs. In predicting unstable IAs, the DLM had an AUC of 0.771 (95% CI: 0.582-0.960), a sensitivity, specificity and accuracy on the external validation set of 0.694, 0.929, and 0.782, respectively. Conclusions: The CNN-based DLM applied to CTA images did not outperform the conventional LRM in predicting unstable IAs. The patient clinical and IA morphological parameters remain critical factors for ensuring IA stability. Further studies are needed to enhance the diagnostic accuracy.

Automated detection and classification of acute vertebral body fractures using a convolutional neural network on computed tomography.

Background: Acute vertebral fracture is usually caused by low-energy injury with osteoporosis and high-energy trauma. The AOSpine thoracolumbar spine injury classification system (AO classification) plays an important role in the diagnosis and treatment of the disease. The diagnosis and description of vertebral fractures according to the classification scheme requires a great deal of time and energy for radiologists. Purpose: To design and validate a multistage deep learning system (multistage AO system) for the automatic detection, localization and classification of acute thoracolumbar vertebral body fractures according to AO classification on computed tomography. Materials and Methods: The CT images of 1,217 patients who came to our hospital from January 2015 to December 2019 were collected retrospectively. The fractures were marked and classified by 2 junior radiology residents according to the type A standard in the AO classification. Marked fracture sites included the upper endplate, lower endplate and posterior wall. When there were inconsistent opinions on classification labels, the final result was determined by a director radiologist. We integrated different networks into different stages of the overall framework. U-net and a graph convolutional neural network (U-GCN) are used to realize the location and classification of the thoracolumbar spine. Next, a classification network is used to detect whether the thoracolumbar spine has a fracture. In the third stage, we detect fractures in different parts of the thoracolumbar spine by using a multibranch output network and finally obtain the AO types. Results: The mean age of the patients was 61.87 years with a standard deviation of 17.04 years, consisting of 760 female patients and 457 male patients. On vertebrae level, sensitivity for fracture detection was 95.23% in test dataset, with an accuracy of 97.93% and a specificity of 98.35%. For the classification of vertebral body fractures, the balanced accuracy was 79.56%, with an AUC of 0.904 for type A1, 0.945 for type A2, 0.878 for type A3 and 0.942 for type A4. Conclusion: The multistage AO system can automatically detect and classify acute vertebral body fractures in the thoracolumbar spine on CT images according to AO classification with high accuracy.

Radiomics and Ki-67 index predict survival in clear cell renal cell carcinoma.

OBJECTIVE: To develop and validate predictive models based on Ki-67 index, radiomics, and Ki-67 index combined with radiomics for survival analysis of patients with clear cell renal cell carcinoma. METHODS: This study enrolled 148 patients who were pathologically diagnosed as ccRCC between March 2010 and December 2018 at our institute. All tissue sections were collected and immunohistochemical staining was performed to calculate Ki-67 index. All patients were randomly divided into the training and validation sets in a 7:3 ratio. Regions of interests (ROIs) were segmented manually. Radiomics features were selected from ROIs in unenhanced, corticomedullary, and nephrographic phases. Multivariate Cox models based on the Ki-67 index and radiomics and univariate Cox models based on the Ki-67 index or radiomics alone were built; the predictive power was evaluated by the concordance (C)-index, integrated area under the curve, and integrated Brier Score. RESULTS: Five features were selected to establish the prediction models of radiomics and combined model. The C-indexes of Ki-67 index model, radiomics model, and combined model were 0.741, 0.718, and 0.782 for disease-free survival (DFS); 0.941, 0.866, and 0.963 for overall survival, respectively. The predictive power of combined model was the best in both training and validation sets. CONCLUSION: The survival prediction performance of combined model was better than Ki-67 model or radiomics model. The combined model is a promising tool for predicting the prognosis of patients with ccRCC in the future. ADVANCES IN KNOWLEDGE: Both Ki-67 and radiomics have showed giant potential in prognosis prediction. There are few studies to investigate the predictive ability of Ki-67 combined with radiomics. This study intended to build a combined model and provide a reliable prognosis for ccRCC in clinical practice.

Detection of Intracranial Aneurysms Using Multiphase CT Angiography with a Deep Learning Model.

RATIONALE AND OBJECTIVES: Determine the effect of a multiphase fusion deep-learning model with automatic phase selection in detection of intracranial aneurysm (IA) from computed tomography angiography (CTA) images. MATERIALS AND METHODS: CTA images of intracranial arteries from patients at Ningbo First Hospital were retrospectively analyzed. Images were randomly classified as training data, internal validation data, or test data. CTA images from cases examined by digital subtraction angiography (DSA) were examined for independent validation. A deep-learning model was constructed by automatic phase selection of multiphase fusion, and compared to the single-phase algorithm to evaluate algorithm sensitivity. RESULTS: We analyzed 1110 patients (1493 aneurysms) as training data, 139 patients (174 aneurysms) as internal validation data, and 134 patients (175 aneurysms) as test data. The sensitivity of the multiphase analysis of the internal validation data, test data, and independent validation data were greater than from the single-phase analysis. The recall of the multiphase selection was greater or equal to that of single-phase selection in the aneurysm position, shape, size, and rupture status. Use of the test data to determine the presence and absence of aneurysm rupture led to a recall from multiphase selection of 94.8% and 87.6% respectively; both of these values were greater than those from single-phase selection (89.6% and 79.4%). CONCLUSION: A multiphase fusion deep learning model with automatic phase selection provided automated detection of IAs with high sensitivity.

Development and validation of a CT-based radiomics model for differentiating pneumonia-like primary pulmonary lymphoma from infectious pneumonia: A multicenter study.

BACKGROUND: Pneumonia-like primary pulmonary lymphoma (PPL) was commonly misdiagnosed as infectious pneumonia, leading to delayed treatment. The purpose of this study was to establish a computed tomography (CT)-based radiomics model to differentiate pneumonia-like PPL from infectious pneumonia. METHODS: In this retrospective study, 79 patients with pneumonia-like PPL and 176 patients with infectious pneumonia from 12 medical centers were enrolled. Patients from center 1 to center 7 were assigned to the training or validation cohort, and the remaining patients from other centers were used as the external test cohort. Radiomics features were extracted from CT images. A three-step procedure was applied for radiomics feature selection and radiomics signature building, including the inter- and intra-class correlation coefficients (ICCs), a one-way analysis of variance (ANOVA), and least absolute shrinkage and selection operator (LASSO). Univariate and multivariate analyses were used to identify the significant clinicoradiological variables and construct a clinical factor model. Two radiologists reviewed the CT images for the external test set. Performance of the radiomics model, clinical factor model, and each radiologist were assessed by receiver operating characteristic, and area under the curve (AUC) was compared. RESULTS: A total of 144 patients (44 with pneumonia-like PPL and 100 infectious pneumonia) were in the training cohort, 38 patients (12 with pneumonia-like PPL and 26 infectious pneumonia) were in the validation cohort, and 73 patients (23 with pneumonia-like PPL and 50 infectious pneumonia) were in the external test cohort. Twenty-three radiomics features were selected to build the radiomics model, which yielded AUCs of 0.95 (95% confidence interval [CI]: 0.94-0.99), 0.93 (95% CI: 0.85-0.98), and 0.94 (95% CI: 0.87-0.99) in the training, validation, and external test cohort, respectively. The AUCs for the two readers and clinical factor model were 0.74 (95% CI: 0.63-0.83), 0.72 (95% CI: 0.62-0.82), and 0.73 (95% CI: 0.62-0.84) in the external test cohort, respectively. The radiomics model outperformed both the readers' interpretation and clinical factor model ( P <0.05). CONCLUSIONS: The CT-based radiomics model may provide an effective and non-invasive tool to differentiate pneumonia-like PPL from infectious pneumonia, which might provide assistance for clinicians in tailoring precise therapy.

Computed tomography-based radiomics for identifying pulmonary cryptococcosis mimicking lung cancer.

BACKGROUND: Pulmonary cryptococcosis (PC) is an invasive pulmonary fungal disease, and nodule/mass-type PC may mimic lung cancer (LC) in imaging appearance. Thus, an accurate diagnosis of nodule/mass-type PC is beneficial for appropriate management. However, the differentiation of nodule/mass-type PC from LC through computed tomography (CT) is still challenging. PURPOSE: To develop and externally test a CT-based radiomics model for differentiating nodule/mass-type PC from LC. METHODS: In this retrospective study, patients with nodule/mass-type PC or LC who underwent non-enhanced chest CT were included: Institution 1 was for the training set, and institutions 2 and 3 were for the external test set. Large quantities of radiomics features were extracted. The radiomics score (Rad-score) was calculated using the linear discriminant analysis, and a subsequent fivefold cross-validation was performed. A combined model was developed by incorporating Rad-score and clinical factors. Finally, the models were tested with an external test set and compared using the area under the receiver operating characteristic curve (AUC). RESULTS: A total of 168 patients (45 with PC and 123 with LC) were in the training set, and 72 (36 with PC and 36 with LC) were in the external test set. Of the 81 patients with PC, 30 were immunocompromised (37%). Rad-score, comprising 18 features, had an AUC of 0.844 after fivefold cross-validation, which was lower than that (AUC = 0.943, p = 0.003) of the combined model integrating Rad-score, age, lobulation, pleural retraction, and patches. In the external test set, Rad-score and the combined model obtained good predictive performance (AUC = 0.824 for Rad-score, and 0.869 for the combined model). Moreover, the combined model outperformed the clinical model in the cross-validation and external test (0.943 vs. 0.810, p <0.001; 0.869 vs. 0.769, p = 0.011). CONCLUSIONS: The proposed combined model exhibits a good differential diagnostic performance between nodule/mass-type PC and LC. The CT-based radiomics analysis has the potential to serve as an effective tool for the differentiation of nodule/mass-type PC from LC in clinical practice.

A Potential Prognostic Marker for Recognizing VEGF-Positive Hepatocellular Carcinoma Based on Magnetic Resonance Radiomics Signature.

Objectives: The objective of our project is to explore a noninvasive radiomics model based on magnetic resonance imaging (MRI) that could recognize the expression of vascular endothelial growth factor (VEGF) in hepatocellular carcinoma before operation. Methods: 202 patients with proven single HCC were enlisted and stochastically distributed into a training set (n = 142) and a test set (n = 60). Arterial phase, portal venous phase, balanced phase, delayed phase, and hepatobiliary phase images were used to radiomics features extraction. We retrieved 1906 radiomic features from each phase of every participant's MRI images. The F-test was applied to choose the crucial features. A logistic regression model was adopted to generate a radiomics signature. By combining independent risk indicators from the fusion radiomics signature and clinico-radiological features, we developed a multivariable logistic regression model that could predict the VEGF status preoperatively through calculating the area under the curve (AUC). Results: The entire group comprised 108 VEGF-positive individuals and 94 VEGF-negative patients. AUCs of 0.892 (95% confidence interval [CI]: 0.839 - 0.945) in the training dataset and 0.800 (95% CI: 0.682 - 0.918) in the test dataset were achieved by utilizing radiomics features from two phase images (8 features from the portal venous phase and 5 features from the hepatobiliary phase). Furthermore, the nomogram relying on a combined model that included the clinical factors α-fetoprotein (AFP), irregular tumor margin, and the fusion radiomics signature performed well in both the training (AUC = 0.936, 95% CI: 0.898-0.974) and test (AUC = 0.836, 95% CI: 0.728-0.944) datasets. Conclusions: The combined model acquired from two phase (portal venous and hepatobiliary phase) pictures of gadolinium-ethoxybenzyl-diethylenetriamine-pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI could be considered as a credible prognostic marker for the level of VEGF in HCC.