RESUMO
This study aims to develop a CT-based hybrid deep learning network to predict pathological subtypes of early-stage lung adenocarcinoma by integrating residual network (ResNet) with Vision Transformer (ViT). A total of 1411 pathologically confirmed ground-glass nodules (GGNs) retrospectively collected from two centers were used as internal and external validation sets for model development. 3D ResNet and ViT were applied to investigate two deep learning frameworks to classify three subtypes of lung adenocarcinoma namely invasive adenocarcinoma (IAC), minimally invasive adenocarcinoma and adenocarcinoma in situ, respectively. To further improve the model performance, four Res-TransNet based models were proposed by integrating ResNet and ViT with different ensemble learning strategies. Two classification tasks involving predicting IAC from Non-IAC (Task1) and classifying three subtypes (Task2) were designed and conducted in this study. For Task 1, the optimal Res-TransNet model yielded area under the receiver operating characteristic curve (AUC) values of 0.986 and 0.933 on internal and external validation sets, which were significantly higher than that of ResNet and ViT models (p < 0.05). For Task 2, the optimal fusion model generated the accuracy and weighted F1 score of 68.3% and 66.1% on the external validation set. The experimental results demonstrate that Res-TransNet can significantly increase the classification performance compared with the two basic models and have the potential to assist radiologists in precision diagnosis.
RESUMO
Current guidelines encourage large studies in a diverse population to establish normal reference ranges for three-dimensional (3D) echocardiography for different ethnic groups. This study was designed to establish the normal values of 3D-left ventricular (LV) and left atrial (LA) volume and function in a nationwide, population-based cohort of healthy Han Chinese adults. A total of 1117 healthy volunteers aged 18-89 years were enrolled from 28 collaborating laboratories in China. Two sets of 3D echocardiographic instruments were used, and full-volume echocardiographic images were recorded and transmitted to a core laboratory for image analysis with a vendor-independent off-line workstation. Finally, 866 volunteers (mean age of 48.4 years, 402 men) were qualified for final analysis. Most parameters exhibited substantial differences between different sex and age groups, even after indexation by body surface area. The normal ranges of 3D-LV and 3D-LA volume and function differed from those recommended by the American Society of Echocardiography and the European Association of Cardiovascular Imaging guidelines, presented by the World Alliance Societies of Echocardiography (WASE) study, and from the 2D values in the EMINCA study. The normal reference values of 3D echocardiography-derived LV and LA volume and function were established for the first time in healthy Han Chinese adults. Normal ranges of 3D-LV and 3D-LA echocardiographic measurements stratified with sex, age, and race should be recommended for clinical applications.
Assuntos
Ecocardiografia Tridimensional , Átrios do Coração , Ventrículos do Coração , Humanos , Pessoa de Meia-Idade , Masculino , Adulto , Ecocardiografia Tridimensional/métodos , Feminino , Idoso , Ventrículos do Coração/diagnóstico por imagem , Ventrículos do Coração/anatomia & histologia , Átrios do Coração/diagnóstico por imagem , Átrios do Coração/anatomia & histologia , China , Valores de Referência , Adulto Jovem , Adolescente , Idoso de 80 Anos ou mais , Povo Asiático , Função Ventricular Esquerda , Função do Átrio Esquerdo , População do Leste AsiáticoRESUMO
Three-dimensional (3D) echocardiography is an emerging technique for assessing right ventricular (RV) volume and function, but 3D-RV normal values from a large Chinese population are still lacking. The aim of the present study was to establish normal values of 3D-RV volume and function in healthy Chinese volunteers. A total of 1117 Han Chinese volunteers from 28 laboratories in 20 provinces of China were enrolled, and 3D-RV images of 747 volunteers with optimal image quality were ultimately analyzed by a core laboratory. Both vendor-dependent and vendor-independent software platforms were used to analyze the 3D-RV images. We found that men had larger RV volumes than women did in the whole population, even after indexing to body surface area, and older individuals had smaller RV volumes. The normal RV volume was significantly smaller than that recommended by the American Society of Echocardiography/European Association of Cardiovascular Imaging guidelines in both sexes. There were significant differences in 3D-RV measurements between the two vendor ultrasound systems and the different software platforms. The echocardiographic measurements in normal Chinese adults II study revealed normal 3D-RV volume and function in a large Chinese population, and there were significant differences between the sexes, ages, races, and vendor groups. Thus, normal 3D-RV values should be stratified by sex, age, race, and vendor.
RESUMO
Background: Normal hepatic functional reserve is the key to avoiding liver failure after liver surgery. This study investigated the assessment of hepatic functional reserve using liver shear wave velocity (LSWV) combined with biochemical indicators, tumor volume, and portal vein diameter. Methods: In this single-center prospective study, a total of 123 patients with hepatocellular carcinoma (HCC) were divided into a test group (n=92) and a validation group (n=31). All patients were Child-Pugh grade A. The indocyanine green retention rate at 15 min (ICG-R15), liver shear wave velocity (LSWV), portal vein diameter (Dpv), alanine aminotransferase (ALT), aspartate transaminase (AST), alkaline phosphatase (ALP), γ-glutamyl transpeptidase (γ-GGT), albumin (ALB), prothrombin time (PT), and also liver tumor volume (maximum diameter ≤5 cm) were measured. In the test group, multiple parameters were used to evaluate hepatic functional reserve, and the multiparametric model was established. Receiver operating characteristic (ROC) curve analysis was conducted to assess the diagnostic performance of the multiparametric model. In the validation group, the predictive effectiveness of the multiparametric model was analyzed using consistency tests. Results: It was revealed that LSWV, ALB, and PT were statistically significant in evaluation of the hepatic functional reserve (P<0.05). The multiparametric model was formulated as follows: Y= -18.954 + 9.726*LSWV-0.397*ALB+2.063*PT. The value of the area under the curve (AUC) for the multiparametric model was 0.913 (95% confidence interval (CI): 0.835-0.962, P< 0.01), with a cutoff value of 16.656 (sensitivity, 0.763; specificity, 0.926). The Kappa value of consistency testing was 0.655 (P<0.01). Conclusion: LSWV combined with ALB and PT exhibited a high predictive effectiveness for the assessment of hepatic functional reserve, assisting the clinical diagnosis and management of liver diseases.
RESUMO
INTRODUCTION: Parameters, such as left ventricular ejection fraction, peak strain dispersion, global longitudinal strain, etc. are influential and clinically interpretable for detection of cardiac disease, while manual detection requires laborious steps and expertise. In this study, we evaluated a video-based deep learning method that merely depends on echocardiographic videos from four apical chamber views of hypertensive cardiomyopathy detection. METHODS: One hundred eighty-five hypertensive cardiomyopathy (HTCM) patients and 112 healthy normal controls (N) were enrolled in this diagnostic study. We collected 297 de-identified subjects' echo videos for training and testing of an end-to-end video-based pipeline of snippet proposal, snippet feature extraction by a three-dimensional (3-D) convolutional neural network (CNN), a weakly-supervised temporally correlated feature ensemble, and a final classification module. The snippet proposal step requires a preliminarily trained end-systole and end-diastole timing detection model to produce snippets that begin at end-diastole, and involve contraction and dilatation for a complete cardiac cycle. A domain adversarial neural network was introduced to systematically address the appearance variability of echo videos in terms of noise, blur, transducer depth, contrast, etc. to improve the generalization of deep learning algorithms. In contrast to previous image-based cardiac disease detection architectures, video-based approaches integrate spatial and temporal information better with a more powerful 3D convolutional operator. RESULTS: Our proposed model achieved accuracy (ACC) of 92%, area under receiver operating characteristic (ROC) curve (AUC) of 0.90, sensitivity(SEN) of 97%, and specificity (SPE) of 84% with respect to subjects for hypertensive cardiomyopathy detection in the test data set, and outperformed the corresponding 3D CNN (vanilla I3D: ACC (0.90), AUC (0.89), SEN (0.94), and SPE (0.84)). On the whole, the video-based methods remarkably appeared superior to the image-based methods, while few evaluation metrics of image-based methods exhibited to be more compelling (sensitivity of 93% and negative predictive value of 100% for the image-based methods (ES/ED and random)). CONCLUSION: The results supported the possibility of using end-to-end video-based deep learning method for the automated diagnosis of hypertensive cardiomyopathy in the field of echocardiography to augment and assist clinicians. TRIAL REGISTRATION: Current Controlled Trials ChiCTR1900025325, Aug, 24, 2019. Retrospectively registered.
Assuntos
Cardiomiopatias , Função Ventricular Esquerda , Humanos , Volume Sistólico , Coração , Redes Neurais de Computação , Cardiomiopatias/diagnóstico por imagemRESUMO
BACKGROUND: The aim of this study was to investigate the ultrasound and hematological indicators, subsequently utilizing them to predict breast cancer and construct predictive models and columnar plots. METHODS: The clinical data of 200 patients with breast tumors receiving ultrasound and blood tests at Henan Provincial People's Hospital from January 2020 to January 2023 were collected. Patients were divided into training and validation sets at a 6:4 ratio using R language. Variables were screened using logistic regression, and a nomogram predicting breast cancer probability was constructed based on the training set. The predictive performance of the nomogram was evaluated in the validation set through receiver operating characteristic, calibration and decision curves. Model robustness was validated by bootstrap resampling. RESULTS: Regression analysis revealed that maximum blood flow velocity within the breast mass ≥ 16.395 m/s, perfusion index ≥ 1.505, cancer antigen 15-3 ≥ 39.620 U/m, cancer antigen 125 ≥ 42.30 U/ml, carcinoembryonic antigen ≥ 6.520 ng/ml, Adler blood flow classification II & III, breast calcification present, and diameter of the lump > 2 cm were independent risk factors for breast cancer. Based on these ultrasonic parameters and blood indicators, the developed nomogram demonstrated excellent discrimination in both the training set (AUC = 0.917) and validation set (AUC = 0.844). The calibration plot showed high consistency between the nomogram-predicted and the actual results. Decision curve analysis indicated higher net benefit of this model. CONCLUSIONS: The nomogram developed in this study demonstrated solid predictive abilities for breast malignancy, indicating potential clinical value pending further research.
RESUMO
Continuum robots have good adaptability in unstructured and complex environments. However, affected by their inherent nature of flexibility and slender structure, there are challenges in high-precision motion and load. Thus, stiffness adjustment for continuum robots has consistently attracted the attention of researchers. In this paper, a stiffness adjustment mechanism (SAM) is proposed and built in a growth-controllable continuum robot (GCCR) to improve the motion accuracy in variable scale motion. The self-stiffness adjustment is realized by antagonism through cable force transmission during the length change of the continuum robot. With a simple structure, the mechanism has a scarce impact on the weight and mass distribution of the robot and required no independent actuators for stiffness adjustment. Following this, a static model considering gravity and end load is established. The presented theoretical static model is applicable to predict the shape deformations of robots under different loads. The experimental validations showed that the maximum error ratio is within 5.65%. The stiffness of the robot can be enhanced by nearly 79.6%.
RESUMO
Background: Liver cirrhosis is closely associated with cardiac dysfunction. The aims of this study were to evaluate left ventricular systolic function in patients with hepatitis B cirrhosis by non-invasive left ventricular pressure-strain loop (LVPSL) technique, and to explore the correlation between myocardial work indices and liver function classification. Methods: According to the Child-Pugh classification, 90 patients with hepatitis B cirrhosis were further divided into three groups: Child-Pugh A group (n = 32), Child-Pugh B group (n = 31), and Child-Pugh C group (n = 27). During the same period, 30 healthy volunteers were recruited as the control (CON) group. Myocardial work parameters, which included global work index (GWI), global constructive work (GCW), global wasted work (GWW), and global work efficiency (GWE), were derived from the LVPSL and compared among the four groups. The correlation between myocardial work parameters and Child-Pugh liver function classification was evaluated, and the independent risk factors affecting left ventricular myocardial work in patients with cirrhosis were investigated by univariable and multivariable linear regression analysis. Results: GWI, GCW and GWE of Child-Pugh B and C groups were lower than those of CON group, while GWW was higher than that of CON group, and the changes were more obvious in Child-Pugh C group (P < 0.05). Correlation analysis revealed that GWI, GCW, and GWE were negatively correlated with liver function classification to various degrees (r = -0.54, -0.57, and -0.83, respectively, all P < 0.001), while GWW was positively correlated with liver function classification (r = 0.76, P < 0.001). Multivariable linear regression analysis showed that GWE was positively correlated with ALB (ß = 0.17, P < 0.001), and negatively correlated with GLS (ß = -0.24, P < 0.001). Conclusions: The changes in the left ventricular systolic function in patients with hepatitis B cirrhosis were identified using non-invasive LVPSL technology, and myocardial work parameters are significantly correlated with liver function classification. This technique may provide a new method for the evaluation of cardiac function in patients with cirrhosis.
RESUMO
COVID-19 has claimed millions of lives since its outbreak in December 2019, and the damage continues, so it is urgent to develop new technologies to aid its diagnosis. However, the state-of-the-art deep learning methods often rely on large-scale labeled data, limiting their clinical application in COVID-19 identification. Recently, capsule networks have achieved highly competitive performance for COVID-19 detection, but they require expensive routing computation or traditional matrix multiplication to deal with the capsule dimensional entanglement. A more lightweight capsule network is developed to effectively address these problems, namely DPDH-CapNet, which aims to enhance the technology of automated diagnosis for COVID-19 chest X-ray images. It adopts depthwise convolution (D), point convolution (P), and dilated convolution (D) to construct a new feature extractor, thus successfully capturing the local and global dependencies of COVID-19 pathological features. Simultaneously, it constructs the classification layer by homogeneous (H) vector capsules with an adaptive, non-iterative, and non-routing mechanism. We conduct experiments on two publicly available combined datasets, including normal, pneumonia, and COVID-19 images. With a limited number of samples, the parameters of the proposed model are reduced by 9x compared to the state-of-the-art capsule network. Moreover, our model has faster convergence speed and better generalization, and its accuracy, precision, recall, and F-measure are improved to 97.99%, 98.05%, 98.02%, and 98.03%, respectively. In addition, experimental results demonstrate that, contrary to the transfer learning method, the proposed model does not require pre-training and a large number of training samples.
Assuntos
COVID-19 , Humanos , COVID-19/diagnóstico por imagem , Teste para COVID-19 , Raios XRESUMO
Evidence of the triglyceride-glucose (TyG) index as an independent predictor of arterial stiffness in stage 1 hypertension patients is scarce. This study aimed to explore the association between TyG index and arterial stiffness in this population. A total of 1041 individuals from 32 centers with normal/elevated blood pressure (BP, <130/80 mmHg; 345 men (33%); median age, 37 years) and 585 stage 1 hypertension patients (BP ≥130/80 and <140/90 mmHg; 305 men (52%); median age, 47 years) were prospectively enrolled. Arterial stiffness was determined by measuring carotid ultrafast pulse-wave velocity (ufPWV). TyG index was calculated as ln (fasting triglyceride (TG) × fasting blood glucose/2). Patients with a higher TyG index tended to have higher ufPWV. The TyG index was positively associated with ufPWV at the end of systole in stage 1 hypertension patients after adjusting for confounding factors (ß for per unit .48), and restricted cubic spline analysis confirmed a linear association. Subgroup analyses in terms of age, sex, and body mass index yielded similar results. However, no significant relationship was observed between the TyG index and ufPWV in the population with normal/elevated BP. The fully adjusted ß between ufPWV and the TyG index was higher than the TG/high-density lipoprotein cholesterol ratio, TG, and pulse pressure. In conclusion, patients with a higher TyG index had greater arterial stiffness, and the TyG index independently and positively correlated with arterial stiffness in stage 1 hypertension patients. The TyG index may provide a simple and reliable marker to monitor arterial stiffness in hypertensive patients.
Assuntos
Hipertensão , Rigidez Vascular , Masculino , Humanos , Adulto , Pessoa de Meia-Idade , Pressão Sanguínea/fisiologia , Hipertensão/diagnóstico , Glucose , Triglicerídeos , Rigidez Vascular/fisiologia , Glicemia , Fatores de Risco , BiomarcadoresRESUMO
OBJECTIVES: To establish a breast lesion risk stratification system using ultrasound images to predict breast malignancy and assess Breast Imaging Reporting and Data System (BI-RADS) categories simultaneously. METHODS: This multicenter study prospectively collected a dataset of ultrasound images for 5012 patients at thirty-two hospitals from December 2018 to December 2020. A deep learning (DL) model was developed to conduct binary categorization (benign and malignant) and BI-RADS categories (2, 3, 4a, 4b, 4c, and 5) simultaneously. The training set of 4212 patients and the internal test set of 416 patients were from thirty hospitals. The remaining two hospitals with 384 patients were used as an external test set. Three experienced radiologists performed a reader study on 324 patients randomly selected from the test sets. We compared the performance of the DL model with that of three radiologists and the consensus of the three radiologists. RESULTS: In the external test set, the DL model achieved areas under the receiver operating characteristic curve (AUCs) of 0.980 and 0.945 for the binary categorization and six-way categorizations, respectively. In the reader study set, the DL BI-RADS categories achieved a similar AUC (0.901 vs. 0.933, p = 0.0632), sensitivity (90.98% vs. 95.90%, p = 0.1094), and accuracy (83.33% vs. 79.01%, p = 0.0541), but higher specificity (78.71% vs. 68.81%, p = 0.0012) than those of the consensus of the three radiologists. CONCLUSIONS: The DL model performed well in distinguishing benign from malignant breast lesions and yielded outcomes similar to experienced radiologists. This indicates the potential applicability of the DL model in clinical diagnosis. KEY POINTS: ⢠The DL model can achieve binary categorization for benign and malignant breast lesions and six-way BI-RADS categorizations for categories 2, 3, 4a, 4b, 4c, and 5, simultaneously. ⢠The DL model showed acceptable agreement with radiologists for the classification of breast lesions. ⢠The DL model performed well in distinguishing benign from malignant breast lesions and had promise in helping reduce unnecessary biopsies of BI-RADS 4a lesions.
Assuntos
Neoplasias da Mama , Aprendizado Profundo , Humanos , Feminino , Neoplasias da Mama/patologia , Mama/diagnóstico por imagem , Ultrassonografia , Medição de Risco , Ultrassonografia Mamária/métodos , Estudos RetrospectivosRESUMO
OBJECTIVES: We aimed to evaluate the safety of the ultrasound contrast agent sulfur hexafluoride microbubbles in a large group of patients referred for routine contrast-enhanced ultrasound (CEUS). METHODS: A retrospective assessment was made of all patients that received sulfur hexafluoride microbubbles intravenously for CEUS at 24 centers between January 2006 and April 2019. Patient demographic details, examination type, and the dose of sulfur hexafluoride microbubbles administered were recorded with specific adverse events (AEs) documentation tools at each center. All AEs were recorded as serious or non-serious. Non-serious AEs were classified by intensity as mild, moderate, or severe according to ACR criteria. The frequencies of AEs across patient subgroups were compared using the chi-square test. RESULTS: A total of 463,434 examinations were evaluated. Overall, 157 AEs (153 [0.033%] non-serious; 4 [0.001%] serious) were reported after sulfur hexafluoride microbubbles administration, giving an AE frequency of 0.034% (157/463,434). Among the non-serious AEs, 66 (0.014%) were mild, 70 (0.015%) moderate, and 17 (0.004%) severe in intensity. The liver was the most common examination site, presenting an AE frequency of 0.026%. The highest AE frequency (0.092%) was for patients undergoing CEUS for vascular disease. There were no significant gender differences in either the total number or the severity of non-serious AEs (chi-square = 2.497, p = 0.287). The onset of AEs occurred within 30 min of sulfur hexafluoride microbubbles administration in 91% of cases. CONCLUSION: The frequency of AEs to sulfur hexafluoride microbubbles is very low and severe reactions are rare, confirming that sulfur hexafluoride microbubbles are appropriate for routine CEUS applications. KEY POINT: ⢠The frequency of AEs to sulfur hexafluoride microbubbles is very low and severe reactions are rare.
Assuntos
Microbolhas , Hexafluoreto de Enxofre , Humanos , Hexafluoreto de Enxofre/efeitos adversos , Estudos Retrospectivos , Meios de Contraste/efeitos adversos , Ultrassonografia , Administração Intravenosa , FosfolipídeosRESUMO
BACKGROUND: Studies on deep learning (DL)-based models in breast ultrasound (US) remain at the early stage due to a lack of large datasets for training and independent test sets for verification. We aimed to develop a DL model for differentiating benign from malignant breast lesions on US using a large multicenter dataset and explore the model's ability to assist the radiologists. METHODS: A total of 14,043 US images from 5012 women were prospectively collected from 32 hospitals. To develop the DL model, the patients from 30 hospitals were randomly divided into a training cohort (n = 4149) and an internal test cohort (n = 466). The remaining 2 hospitals (n = 397) were used as the external test cohorts (ETC). We compared the model with the prospective Breast Imaging Reporting and Data System assessment and five radiologists. We also explored the model's ability to assist the radiologists using two different methods. RESULTS: The model demonstrated excellent diagnostic performance with the ETC, with a high area under the receiver operating characteristic curve (AUC, 0.913), sensitivity (88.84%), specificity (83.77%), and accuracy (86.40%). In the comparison set, the AUC was similar to that of the expert (p = 0.5629) and one experienced radiologist (p = 0.2112) and significantly higher than that of three inexperienced radiologists (p < 0.01). After model assistance, the accuracies and specificities of the radiologists were substantially improved without loss in sensitivities. CONCLUSIONS: The DL model yielded satisfactory predictions in distinguishing benign from malignant breast lesions. The model showed the potential value in improving the diagnosis of breast lesions by radiologists.
RESUMO
Introduction: The aim of this study was to explore the association between sagittal spinopelvic alignment and persistent low back pain (PLBP) following posterior decompression and instrumented fusion for mild L5-S1 spondylolisthesis. Methods: By retrieving medical records from January 2015 to April 2020, 200 patients following PLIF for mild L5-S1 spondylolisthesis were retrospectively reviewed. Patients were divided into two groups: PLBP group and non-PLBP group. The baseline characteristics and radiographic parameters were analyzed and compared between groups. Results: The PLBP group comprised 26 patients, and the non-PLBP group comprised 174 patients. No significant differences in preoperative spinopelvic parameters were found between the two groups (p > 0.05). There were statistically significant differences between the preoperative and postoperative variations of LL, SL, LSA, SD, and HOD spinopelvic parameters, regardless of group (p < 0.05). Postoperative PT and SS differed significantly between the PLBP group and non-PLBP group (p < 0.05). In the PLBP group, there were no significant differences between preoperative and postoperative PT; the same applied to SS. However, significant differences were found for the variations in preoperative and postoperative PT and SS between the two groups. The ΔPT was found as an independent risk factor for postoperative PLBP. Conclusion: Patients with mild L5-S1 spondylolisthesis with PLBP after posterior lumbar spinal fusion had decreased SS and increased PT.
RESUMO
BACKGROUND: Osteosarcoma is a common primary bone tumour in children and adolescents. Circular RNAs (circRNAs) exert vital functions in human diseases, including osteosarcoma. Therefore, we explored the role of circ_0016347 in osteosarcoma. METHODS: The real-time quantitative polymerase chain reaction (RT-qPCR) was used to detect the expression levels of circ_0016347, microRNA-661 (miR-661), and Interleukin-6 receptor (IL6R) in osteosarcoma tissues and cells. The proliferation of osteosarcoma cells was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazol-3-ium bromide (MTT) and EdU experiments. The migration and invasion were determined by transwell assay. The cell cycle distribution and apoptosis were assessed by flow cytometry assay. The association relationships among circ_0016347, miR-661, and IL6R were analyzed by dual-luciferase reporter assays. The western blot assay was employed to assay the protein expression. A xenograft experiment was established to clarify the functional role of circ_0016347 inhibition in vivo. RESULTS: Circ_0016347 was obviously overexpressed in osteosarcoma tissues and cells compared with control groups. The suppression of circ_0016347 impeded proliferation, migration, invasion, and cell cycle and induced apoptosis in osteosarcoma cells, which was overturned by knockdown of miR-661. Consistently, circ_0016347 knockdown repressed tumour growth in vivo. Moreover, miR-661 directly targeted and inhibited IL6R, and the upregulation of IL6R reversed miR-661-induced effects on osteosarcoma cells. Furthermore, circ_0016347 could regulate IL6R expression through miR-661. Inhibition of circ_0016347 also inactivated the Janus kinase 2 (JAK2)/Transcription 3 (STAT3) signalling pathway in osteosarcoma cells by IL6R. CONCLUSION: Circ_0016347 functioned as an oncogene in osteosarcoma at least in part by the miR-661/IL6R axis and JAK2/STAT3 signalling pathway.
Assuntos
Neoplasias Ósseas , MicroRNAs , Osteossarcoma , Adolescente , Apoptose/genética , Neoplasias Ósseas/genética , Neoplasias Ósseas/patologia , Ciclo Celular/genética , Linhagem Celular Tumoral , Movimento Celular/genética , Proliferação de Células/genética , Criança , Humanos , MicroRNAs/genética , MicroRNAs/metabolismo , Osteossarcoma/genética , Osteossarcoma/patologia , Receptores de Interleucina-6/genéticaRESUMO
BACKGROUND: To assess the performance of conventional ultrasound (US) combined with strain elastography (SE) in the Breast Imaging Reporting and Data System (BI-RADS) category 4 lesions on mammography. MATERIALS AND METHODS: Women with breast lesions identified as having mammography BI-RADS 4 lesions and underwent US examination were included in China. US features and US BI-RADS assessment were recorded in real-time and prospectively reported. The pathological result was referred to as the gold standard. The performance of US in the mammographic BI-RADS category 4 lesions was evaluated. Diagnostic performances of US BI-RADS, SE and combined both were compared. RESULTS: A total of 751 women with 751 breast lesions classified as mammographic BI-RADS category 4 were included. For mammographic findings, 530 (70.6%) were true positive and 221 (29.4%) were false positive. Conventional US achieved higher positive predictive value (PPV) than mammography (78.5% vs. 70.6%, P=.001). The specificity increased from 34.4% to 47.1% (P< .001) without any loss in sensitivity and the PPV increased to 81.9% (P = .122) when conventional US was used in combination with SE. For conventional US combined with SE, it led to a correct diagnosis of no breast cancer in 104 of the 221 false-positive findings (47.1%) and achieved higher PPV than mammography regardless of patient age and lesion size. CONCLUSION: Conventional US combined with SE is a helpful tool for the noninvasive examination of breast lesions classified as BI-RADS category 4 on mammography. It helped increase the PPV and had the potential to avoid unnecessary biopsies of BI-RADS category 4 lesions detected on mammography.
Assuntos
Neoplasias da Mama , Técnicas de Imagem por Elasticidade , Biópsia , Neoplasias da Mama/patologia , Técnicas de Imagem por Elasticidade/métodos , Feminino , Humanos , Mamografia/métodos , Estudos Retrospectivos , Ultrassonografia Mamária/métodosRESUMO
RATIONALE AND OBJECTIVES: The objective of this study was to evaluate the utility of the fifth edition of the Breast Imaging-Reporting and Data System (BI-RADS) in clinical breast radiology by using prospective multicenter real-time analyses of ultrasound (US) images. MATERIALS AND METHODS: We prospectively studied 2049 female patients (age range, 19-86 years; mean age 46.88 years) with BI-RADS category 4 breast masses in 32 tertiary hospitals. All the patients underwent B-mode, color Doppler US, and US elastography examination. US features of the mass and associated features were described and categorized according to the fifth edition of the BI-RADS US lexicon. The pathological results were used as the reference standard. The positive predictive values (PPVs) of subcategories 4a-4c were calculated. RESULTS: A total of 2094 masses were obtained, including 1124 benign masses (54.9%) and 925 malignant masses (45.1%). For BI-RADS US features of mass shape, orientation, margin, posterior features, calcifications, architectural distortion, edema, skin changes, vascularity, and elasticity assessment were significantly different for benign and malignant masses (p< 0.05). Typical signs of malignancy were irregular shape (PPV, 57.2%), spiculated margin (PPV, 83.7%), nonparallel orientation (PPV, 63.9%), and combined pattern of posterior features (PPV, 60.6%). For the changed or newly added US features, the PPVs for intraductal calcifications were 80%, 56.4% for internal vascularity, and 80% for a hard pattern on elastography. The associated features such as architectural distortion (PPV, 89.3%), edema (PPV, 69.2%), and skin changes (PPV, 76.2%) displayed high predictive value for malignancy. The rate of malignant was 7.4% (72/975) in category 4a, 61.4% (283/461) in category 4b, and 93.0% (570/613) in category 4c. The PPV for category 4b was higher than the likelihood ranges specified in BI-RADS and the PPVs for categories 4a and 4c were within the acceptable performance ranges specified in the fifth edition of BI-RADS in our study. CONCLUSION: Not only the US features of the breast mass, but also associated features, including vascularity and elasticity assessment, have become an indispensable part of the fifth edition of BI-RADS US lexicon to distinguish benign and malignant breast lesions. The subdivision of category 4 lesions into categories 4a, 4b, and 4c for US findings is helpful for further assessment of the likelihood of malignancy of breast lesions.
Assuntos
Neoplasias da Mama , Ultrassonografia Mamária , Adulto , Idoso , Idoso de 80 Anos ou mais , Neoplasias da Mama/diagnóstico por imagem , Feminino , Humanos , Pessoa de Meia-Idade , Estudos Prospectivos , Estudos Retrospectivos , Ultrassonografia , Ultrassonografia Mamária/métodos , Adulto JovemRESUMO
The outbreak of COVID-19 threatens the lives and property safety of countless people and brings a tremendous pressure to health care systems worldwide. The principal challenge in the fight against this disease is the lack of efficient detection methods. AI-assisted diagnosis based on deep learning can detect COVID-19 cases for chest X-ray images automatically, and also improve the accuracy and efficiency of doctors' diagnosis. However, large scale annotation of chest X-ray images is difficult because of limited resources and heavy burden on the medical system. To meet the challenge, we propose a capsule network model with multi-head attention routing algorithm, called MHA-CoroCapsule, to provide fast and accurate diagnostics for COVID-19 diseases from chest X-ray images. The MHA-CoroCapsule consists of convolutional layers, two capsule layers, and a non-iterative, parameterized multi-head attention routing algorithm is used to quantify the relationship between the two capsule layers. The experiments are performed on a combined dataset constituted by two publicly available datasets including normal, non-COVID pneumonia and COVID-19 images. The model achieves the accuracy of 97.28%, recall of 97.36%, and precision of 97.38% even with a limited number of samples. The experimental results demonstrate that, contrary to the transfer learning and deep feature extraction approaches, the proposed MHA-CoroCapsule has an encouraging performance with fewer trainable parameters and does not require pretraining and plenty of training samples.