Development and External Validation of the 3D U-Net Algorithm for Segmentation of Pelvic Lymph Nodes On Diffusion-Weighted Images

2021 ◽  
Author(s):  
Xiang Liu ◽  
Zhaonan Sun ◽  
Chao Han ◽  
Yingpu Cui ◽  
Jiahao Huang ◽  
...  
Keyword(s):  
Lymph Nodes ◽  
External Validation ◽  
Model Development ◽  
Area Under The Curve ◽  
True Positive Rate ◽  
Validation Dataset ◽  
Diffusion Weighted Images ◽  
Diffusion Weighted ◽  
Expert Radiologist ◽  
High Consistency

Abstract Background: The 3D U-Net model has been proved to perform well in the automatic organ segmentation. The aim of this study is to evaluate the feasibility of the 3D U-Net algorithm for the automated detection and segmentation of lymph nodes (LNs) on pelvic diffusion-weighted imaging (DWI) images.Methods: A total of 393 DWI images of patients suspected of having prostate cancer (PCa) between January 2019 and December 2020 were collected for model development. Seventy-seven DWI images from another group of PCa patients imaged between January 2021 and April 2021 were collected for external validation. Segmentation performance was assessed using the Dice score, positive predictive value (PPV), true positive rate (TPR), and volumetric similarity (VS) with manual annotation of pelvic LNs as the reference. The accuracy with which the suspicious metastatic LNs (short diameter > 0.8 cm) were detected was evaluated using the area under the curve (AUC) at the patient level, and the precision, recall, and F1-score were determined at the lesion level. The consistency of LN staging on an external dataset between the model and radiologist was assessed using Cohen’s kappa coefficient.Results: In the testing set used for model development, the Dice score, TPR, PPV, and VS for the segmentation of suspicious LNs were 0.85, 0.82, 0.80 and 0.86, respectively. The precision, recall, and F1-score for the detection of suspicious LNs were 0.97, 0.98 and 0.97, respectively. In the external validation dataset, the AUC of the model for identifying PCa patients with suspicious LNs was 0.963 (95% CI: 0.892-0.993). High consistency of LN staging (Kappa = 0.922) was achieved between the model and expert radiologist.Conclusion: The 3D U-Net algorithm can accurately detect and segment pelvic LNs based on DWI images.

scholarly journals Development and validation of the 3D U-Net algorithm for segmentation of pelvic lymph nodes on diffusion-weighted images

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Xiang Liu ◽  
Zhaonan Sun ◽  
Chao Han ◽  
Yingpu Cui ◽  
Jiahao Huang ◽  
...  
Keyword(s):  
Lymph Nodes ◽  
Average Distance ◽  
Model Development ◽  
Area Under The Curve ◽  
True Positive Rate ◽  
Validation Dataset ◽  
Diffusion Weighted Images ◽  
Diffusion Weighted ◽  
Expert Radiologist ◽  
Temporal Validation

Abstract Background The 3D U-Net model has been proved to perform well in the automatic organ segmentation. The aim of this study is to evaluate the feasibility of the 3D U-Net algorithm for the automated detection and segmentation of lymph nodes (LNs) on pelvic diffusion-weighted imaging (DWI) images. Methods A total of 393 DWI images of patients suspected of having prostate cancer (PCa) between January 2019 and December 2020 were collected for model development. Seventy-seven DWI images from another group of PCa patients imaged between January 2021 and April 2021 were collected for temporal validation. Segmentation performance was assessed using the Dice score, positive predictive value (PPV), true positive rate (TPR), and volumetric similarity (VS), Hausdorff distance (HD), the Average distance (AVD), and the Mahalanobis distance (MHD) with manual annotation of pelvic LNs as the reference. The accuracy with which the suspicious metastatic LNs (short diameter > 0.8 cm) were detected was evaluated using the area under the curve (AUC) at the patient level, and the precision, recall, and F1-score were determined at the lesion level. The consistency of LN staging on an hold-out test dataset between the model and radiologist was assessed using Cohen’s kappa coefficient. Results In the testing set used for model development, the Dice score, TPR, PPV, VS, HD, AVD and MHD values for the segmentation of suspicious LNs were 0.85, 0.82, 0.80, 0.86, 2.02 (mm), 2.01 (mm), and 1.54 (mm) respectively. The precision, recall, and F1-score for the detection of suspicious LNs were 0.97, 0.98 and 0.97, respectively. In the temporal validation dataset, the AUC of the model for identifying PCa patients with suspicious LNs was 0.963 (95% CI: 0.892–0.993). High consistency of LN staging (Kappa = 0.922) was achieved between the model and expert radiologist. Conclusion The 3D U-Net algorithm can accurately detect and segment pelvic LNs based on DWI images.

Combining tumor size and diffusion-weighted imaging to diagnose normal-sized metastatic pelvic lymph nodes in cervical cancers

2018 ◽  
Vol 60 (3) ◽  
pp. 388-395 ◽  
Author(s):  
Jiacheng Song ◽  
Qiming Hu ◽  
Junwen Huang ◽  
Zhanlong Ma ◽  
Ting Chen
Keyword(s):  
Cervical Cancer ◽  
Lymph Nodes ◽  
Cancer Patients ◽  
Tumor Size ◽  
Diffusion Weighted Imaging ◽  
Area Under The Curve ◽  
Adc Value ◽  
Cervical Cancers ◽  
Pelvic Lymph Nodes ◽  
Diffusion Weighted

Background Detecting normal-sized metastatic pelvic lymph nodes (LNs) in cervical cancers, although difficult, is of vital importance. Purpose To investigate the value of diffusion-weighted-imaging (DWI), tumor size, and LN shape in predicting metastases in normal-sized pelvic LNs in cervical cancers. Material and Methods Pathology confirmed cervical cancer patients with complete magnetic resonance imaging (MRI) were documented from 2011 to 2016. A total of 121 cervical cancer patients showed small pelvic LNs (<5 mm) and 92 showed normal-sized (5–10 mm) pelvic LNs (39 patients with 55 nodes that were histologically metastatic, 53 patients with 71 nodes that were histologically benign). Preoperative clinical and MRI variables were analyzed and compared between the metastatic and benign groups. Results LN apparent diffusion coefficient (ADC) values and short-to-long axis ratios were not significantly different between metastatic and benign normal-sized LNs (0.98 ± 0.15 × 10−3 vs. 1.00 ± 0.18 × 10−3 mm2/s, P = 0.45; 0.65 ± 0.16 vs. 0.64 ± 0.16, P = 0.60, respectively). Tumor ADC value of the metastatic LNs was significantly lower than the benign LNs (0.98 ± 0.12 × 10−3 vs. 1.07 ± 0.21 × 10−3 mm2/s, P = 0.01). Tumor size (height) was significantly higher in the metastatic LN group (27.59 ± 9.18 mm vs. 21.36 ± 10.40 mm, P < 0.00). Spiculated border rate was higher in the metastatic LN group (9 [16.4%] vs. 3 [4.2%], P = 0.03). Tumor (height) combined with tumor ADC value showed the highest area under the curve of 0.702 ( P < 0.00) in detecting metastatic pelvic nodes, with a sensitivity of 59.1% and specificity of 78.8%. Conclusions Tumor DWI combined with tumor height were superior to LN DWI and shape in predicting the metastatic state of normal-sized pelvic LNs in cervical cancer patients.

scholarly journals Development and external validation of a clinical prediction model for MRSA carriage at hospital admission in Southeast Lower Saxony, Germany

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Gabriele Raschpichler ◽  
Heike Raupach-Rosin ◽  
Manas K. Akmatov ◽  
Stefanie Castell ◽  
Nicole Rübsamen ◽  
...  
Keyword(s):  
Hospital Admission ◽  
External Validation ◽  
Model Development ◽  
Cost Effective ◽  
Risk Groups ◽  
Validation Dataset ◽  
Lower Saxony ◽  
Screening Programs ◽  
Lower Specificity ◽  
History Of

Abstract In countries with low endemic Methicillin-resistant Staphylococcus aureus (MRSA) prevalence, identification of risk groups at hospital admission is considered more cost-effective than universal MRSA screening. Predictive statistical models support the selection of suitable stratification factors for effective screening programs. Currently, there are no universal guidelines in Germany for MRSA screening. Instead, a list of criteria is available from the Commission for Hospital Hygiene and Infection Prevention (KRINKO) based on which local strategies should be adopted. We developed and externally validated a model for individual prediction of MRSA carriage at hospital admission in the region of Southeast Lower Saxony based on two prospective studies with universal screening in Braunschweig (n = 2065) and Wolfsburg (n = 461). Logistic regression was used for model development. The final model (simplified to an unweighted score) included history of MRSA carriage, care dependency and cancer treatment. In the external validation dataset, the score showed a sensitivity of 78.4% (95% CI: 64.7–88.7%), and a specificity of 70.3% (95% CI: 65.0–75.2%). Of all admitted patients, 25.4% had to be screened if the score was applied. A model based on KRINKO criteria showed similar sensitivity but lower specificity, leading to a considerably higher proportion of patients to be screened (49.5%).

scholarly journals Development and External Validation of a Diagnostic Model for Coronary Microvascular Obstruction

2020 ◽  
Author(s):  
Yong Li ◽  
Shuzheng Lyu
Keyword(s):  
Microvascular Obstruction ◽  
Characteristic Curve ◽  
External Validation ◽  
Model Development ◽  
University Hospital ◽  
Validation Dataset ◽  
Diagnostic Model ◽  
Total Occlusion ◽  
St Segment Elevation ◽  
Acute Stemi

AbstractBackgroundPrevention of coronary microvascular obstruction /no-reflow phenomenon(CMVO/NR) is a crucial step in improving prognosis of patients with acute ST segment elevation myocardial infarction (STEMI)during primary percutaneous coronary intervention (PPCI). We wanted to develop and externally validate a diagnostic model of CMVO/NR in patients with acute STEMI underwent PPCI.MethodsDesign: Multivariable logistic regression of a cohort of acute STEMI patients. Setting: Emergency department ward of a university hospital. Participants: Diagnostic model development: Totally 1232 acute STEMI patients who were consecutively treated with PPCI from November 2007 to December 2013. External validation: Totally 1301 acute STEMI patients who were treated with PPCI from January 2014 to June 2018. Outcomes: CMVO/NR during PPCI.Results147(11.9%)patients presented CMVO/NR in the development dataset and 120(9.2%) patients presented CMVO/NR in the validation dataset. The strongest predictors of CMVO/NR were age, periprocedural bradycardia, using thrombus aspiration devices during procedure and total occlusion of culprit vessel. We developed a diagnostic model of CMVO/NR.The area under the receiver operating characteristic curve (AUC) was 0.6833 in the development set.We constructed a nomogram using the development database.The AUC was 0.6547 in the validation set. Discrimination, calibration, and decision curve analysis were satisfactory.ConclusionsWe developed and externally validated a diagnostic model of CMVO/NR during PPCI.We registered this study with WHO International Clinical Trials Registry Platform on 16 May 2019. Registration number: ChiCTR1900023213. http://www.chictr.org.cn/edit.aspx?pid=39057&htm=4.

scholarly journals Predicting Sex from Retinal Fundus Photographs Using Automated Deep Learning

2021 ◽  
Author(s):  
Edward Korot ◽  
Nikolas Pontikos ◽  
Xiaoxuan Liu ◽  
Siegfried K Wagner ◽  
Livia Faes ◽  
...  
Keyword(s):  
Deep Learning ◽  
Characteristic Curve ◽  
External Validation ◽  
Model Development ◽  
Model Performance ◽  
Validation Dataset ◽  
Internal Validation ◽  
Fundus Photographs ◽  
The Uk ◽  
Retinal Fundus

Abstract Deep learning may transform health care, but model development has largely been dependent on availability of advanced technical expertise. Herein we present the development of a deep learning model by clinicians without coding, which predicts reported sex from retinal fundus photographs. A model was trained on 84,743 retinal fundus photos from the UK Biobank dataset. External validation was performed on 252 fundus photos from a tertiary ophthalmic referral center. For internal validation, the area under the receiver operating characteristic curve (AUROC) of the code free deep learning (CFDL) model was 0.93. Sensitivity, specificity, positive predictive value (PPV) and accuracy (ACC) were 88.8%, 83.6%, 87.3% and 86.5%, and for external validation were 83.9%, 72.2%, 78.2% and 78.6% respectively. Clinicians are currently unaware of distinct retinal feature variations between males and females, highlighting the importance of model explainability for this task. The model performed significantly worse when foveal pathology was present in the external validation dataset, ACC: 69.4%, compared to 85.4% in healthy eyes, suggesting the fovea is a salient region for model performance OR (95% CI): 0.36 (0.19, 0.70) p = 0.0022. Automated machine learning (AutoML) may enable clinician-driven automated discovery of novel insights and disease biomarkers.

scholarly journals Model to Predict Overall Survival in Patients With Hepatocellular Carcinoma After Curative Hepatectomy

2021 ◽  
Vol 10 ◽  
Author(s):  
Li-xiang Zhang ◽  
Pan-quan Luo ◽  
Lei Chen ◽  
Dong-da Song ◽  
A-man Xu ◽  
...  
Keyword(s):  
Hepatocellular Carcinoma ◽  
Scoring System ◽  
External Validation ◽  
Area Under The Curve ◽  
Serum Markers ◽  
Validation Dataset ◽  
Roc Curve Analysis ◽  
Scoring Model ◽  
New Scoring ◽  
Curative Hepatectomy

BackgroundThe prognosis of patients with hepatocellular carcinoma (HCC) remains difficult to accurately predict. The purpose of this study was to establish a prognostic model for HCC based on a novel scoring system.MethodsFive hundred and sixty patients who underwent a curative hepatectomy for treatment of HCC at our hospital between January 2007 and January 2014 were included in this study. Univariate and multivariate analyses were used to screen for prognostic risk factors. The nomogram construction was based on Cox proportional hazard regression models, and the development of the new scoring model was analyzed using receiver operating characteristic (ROC) curve analysis and then compared with other clinical indexes. The novel scoring system was then validated with an external dataset from a different medical institution.ResultsMultivariate analysis showed that tumor size, portal vein tumor thrombus (PVTT), invasion of adjacent tissues, microvascular invasion, and levels of fibrinogen and total bilirubin were independent prognostic factors. The new scoring model had higher area under the curve (AUC) values compared to other systems, and the C-index of the nomogram was highly consistent for evaluating the survival of HCC patients in the validation and training datasets, as well as the external validation dataset.ConclusionsBased on serum markers and other clinical indicators, a precise model to predict the prognosis of patients with HCC was developed. This novel scoring system can be an effective tool for both surgeons and patients.

scholarly journals Preoperative Alfa-Fetoprotein and Fibrinogen Predict Hepatocellular Carcinoma Recurrence After Liver Transplantation Regardless of the Milan Criteria: Model Development with External Validation

2018 ◽  
Vol 48 (1) ◽  
pp. 317-327 ◽  
Author(s):  
Nan Jiang ◽  
Kai-Ning Zeng ◽  
Ke-Feng Dou ◽  
Yi Lv ◽  
Jie Zhou ◽  
...  
Keyword(s):  
Hepatocellular Carcinoma ◽  
Liver Transplantation ◽  
External Validation ◽  
Model Development ◽  
Area Under The Curve ◽  
Disease Free Survival ◽  
Milan Criteria ◽  
Cutoff Score ◽  
Fibrinogen Concentration ◽  
Patient Prognosis

Background/Aims: Patient selection is critically important in improving the outcomes of liver transplantation for hepatocellular carcinoma. The aim of the current study was to identify biochemical measures that could affect patient prognosis after liver transplantation. Methods: A total of 119 patients receiving liver transplantation for hepatocellular carcinoma were used to construct a model for predicting recurrence. The results were validated using an independent sample of 109 patients from independent hospitals. All subjects in both cohorts met the Hangzhou criteria. Results: Analysis of the discovery cohort revealed an association of recurrence with preoperative fibrinogen and AFP levels. A mathematical model was developed for predicting probability of recurrence within 5 years: Y = logit(P) = -4.595 + 0.824 ×fibrinogen concentration (g/L) + 0.641 × AFP score (1 for AFP<=20ng/ml, 2 for 20<AFP<=100ng/ml, 3 for 100<AFP<=200ng/ml, 4 for 200<AFP<=400ng/ml, 5 for AFP> 400ng/ml). At a cutoff score of -0.85, the area under the curve (AUC) was 0.819 in predicting recurrence (vs. 0.655 when using the Milan criteria). In the validation cohort, this model had reasonable performance in predicting 5-year overall survival (68.8% vs. 28.1% in using the -0.85 cutoff, p< 0.001) and disease-free survival (65.7% vs. 25.9%, p< 0.001). The sensitivity and specificity were 77.0% and 62.5%, respectively. The AUC of this newly developed model was similar to that with the Milan criteria (0.698 vs. 0.678). Surprisingly, the DFS in patients with score <= -0.85 under this model but not meeting the Milan criteria was similar to that in patients meeting the Milan criteria (53.8% vs. 60.0%, p=0.380). Conclusions: Preoperative AFP and fibrinogen are useful in predicting recurrence of hepatocellular carcinoma after liver transplantation.

scholarly journals Plasma metabolites associated with homeostatic model assessment of insulin resistance: metabolite-model design and external validation

2019 ◽  
Vol 9 (1) ◽  
Author(s):  
Pablo Hernández-Alonso ◽  
Jesús García-Gavilán ◽  
Lucía Camacho-Barcia ◽  
Anders Sjödin ◽  
Thea T. Hansen ◽  
...  
Keyword(s):  
Insulin Resistance ◽  
External Validation ◽  
Area Under The Curve ◽  
Validation Dataset ◽  
Plasma Metabolites ◽  
Model Assessment ◽  
Cross Sectional ◽  
Lipid Species ◽  
Homeostatic Model Assessment ◽  
Homeostatic Model

Abstract Different plasma metabolites have been related to insulin resistance (IR). However, there is a lack of metabolite models predicting IR with external validation. The aim of this study is to identify a multi-metabolite model associated to the homeostatic model assessment (HOMA)-IR values. We performed a cross-sectional metabolomics analysis of samples collected from overweight and obese subjects from two independent studies. The training step was performed in 236 subjects from the SATIN study and validated in 102 subjects from the GLYNDIET study. Plasma metabolomics profile was analyzed using three different approaches: GC/quadrupole-TOF, LC/quadrupole-TOF, and nuclear magnetic resonance (NMR). Associations between metabolites and HOMA-IR were assessed using elastic net regression analysis with a leave-one-out cross validation (CV) and 100 CV runs. HOMA-IR was analyzed both as linear and categorical (median or lower versus higher than the median). Receiver operating characteristic curves were constructed based on metabolites’ weighted models. A set of 30 metabolites discriminating extremes of HOMA-IR were consistently selected. These metabolites comprised some amino acids, lipid species and different organic acids. The area under the curve (AUC) for the discrimination between HOMA-IR extreme categories was 0.82 (95% CI: 0.74–0.90), based on the multi-metabolite model weighted with the regression coefficients of metabolites in the validation dataset. We identified a set of metabolites discriminating between extremes of HOMA-IR and able to predict HOMA-IR with high accuracy.

Sign in / Sign up

Export Citation Format

Share Document