The rapid spread of COVID-19 has resulted in the shortage of medical resources, which necessitates accurate prognosis prediction to triage patients effectively. This study used the nationwide cohort ...of South Korea to develop a machine learning model to predict prognosis based on sociodemographic and medical information. Of 10,237 COVID-19 patients, 228 (2.2%) died, 7772 (75.9%) recovered, and 2237 (21.9%) were still in isolation or being treated at the last follow-up (April 16, 2020). The Cox proportional hazards regression analysis revealed that age > 70, male sex, moderate or severe disability, the presence of symptoms, nursing home residence, and comorbidities of diabetes mellitus (DM), chronic lung disease, or asthma were significantly associated with increased risk of mortality (p ≤ 0.047). For machine learning, the least absolute shrinkage and selection operator (LASSO), linear support vector machine (SVM), SVM with radial basis function kernel, random forest (RF), and k-nearest neighbors were tested. In prediction of mortality, LASSO and linear SVM demonstrated high sensitivities (90.7% 95% confidence interval: 83.3, 97.3 and 92.0% 85.9, 98.1, respectively) and specificities (91.4% 90.3, 92.5 and 91.8%, 90.7, 92.9, respectively) while maintaining high specificities > 90%, as well as high area under the receiver operating characteristics curves (0.963 0.946, 0.979 and 0.962 0.945, 0.979, respectively). The most significant predictors for LASSO included old age and preexisting DM or cancer; for RF they were old age, infection route (cluster infection or infection from personal contact), and underlying hypertension. The proposed prediction model may be helpful for the quick triage of patients without having to wait for the results of additional tests such as laboratory or radiologic studies, during a pandemic when limited medical resources must be wisely allocated without hesitation.
We evaluated the diagnostic performance and generalizability of traditional machine learning and deep learning models for distinguishing glioblastoma from single brain metastasis using radiomics. The ...training and external validation cohorts comprised 166 (109 glioblastomas and 57 metastases) and 82 (50 glioblastomas and 32 metastases) patients, respectively. Two-hundred-and-sixty-five radiomic features were extracted from semiautomatically segmented regions on contrast-enhancing and peritumoral T2 hyperintense masks and used as input data. For each of a deep neural network (DNN) and seven traditional machine learning classifiers combined with one of five feature selection methods, hyperparameters were optimized through tenfold cross-validation in the training cohort. The diagnostic performance of the optimized models and two neuroradiologists was tested in the validation cohort for distinguishing glioblastoma from metastasis. In the external validation, DNN showed the highest diagnostic performance, with an area under receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy of 0.956 (95% confidence interval CI, 0.918-0.990), 90.6% (95% CI, 80.5-100), 88.0% (95% CI, 79.0-97.0), and 89.0% (95% CI, 82.3-95.8), respectively, compared to the best-performing traditional machine learning model (adaptive boosting combined with tree-based feature selection; AUC, 0.890 (95% CI, 0.823-0.947)) and human readers (AUC, 0.774 95% CI, 0.685-0.852 and 0.904 95% CI, 0.852-0.951). The results demonstrated deep learning using radiomic features can be useful for distinguishing glioblastoma from metastasis with good generalizability.
Objectives
To evaluate whether a deep learning (DL) model using both three-dimensional (3D) black-blood (BB) imaging and 3D gradient echo (GRE) imaging may improve the detection and segmentation ...performance of brain metastases compared to that using only 3D GRE imaging.
Methods
A total of 188 patients with brain metastases (917 lesions) who underwent a brain metastasis MRI protocol including contrast-enhanced 3D BB and 3D GRE were included in the training set. DL models based on 3D U-net were constructed. The models were validated in the test set consisting of 45 patients with brain metastases (203 lesions) and 49 patients without brain metastases.
Results
The combined 3D BB and 3D GRE model yielded better performance than the 3D GRE model (sensitivities of 93.1% vs 76.8%,
p
< 0.001), and this effect was significantly stronger in subgroups with small metastases (
p
interaction < 0.001). For metastases < 3 mm, ≥ 3 mm and < 10 mm, and ≥ 10 mm, the sensitivities were 82.4%, 93.2%, and 100%, respectively. The combined 3D BB and 3D GRE model showed a false-positive per case of 0.59 in the test set. The combined 3D BB and 3D GRE model showed a Dice coefficient of 0.822, while 3D GRE model showed a lower Dice coefficient of 0.756.
Conclusions
The combined 3D BB and 3D GRE DL model may improve the detection and segmentation performance of brain metastases, especially in detecting small metastases.
Key Points
• The combined 3D BB and 3D GRE model yielded better performance for the detection of brain metastases than the 3D GRE model (p < 0.001), with sensitivities of 93.1% and 76.8%, respectively.
• The combined 3D BB and 3D GRE model showed a false-positive rate per case of 0.59 in the test set.
• The combined 3D BB and 3D GRE model showed a Dice coefficient of 0.822, while the 3D GRE model showed a lower Dice coefficient of 0.756.
To identify magnetic resonance (MR) imaging features that enable prediction of early recurrence (<2 years) after curative resection of hepatocellular carcinoma (HCC) and to derive a preoperative ...prediction model.
This retrospective study was approved by the institutional review board. The requirement to obtain written informed consent was waived. A total of 268 patients who underwent hepatic resection for a single HCC from January 2008 to August 2011 were divided into two cohorts: a training cohort, which was used to derive a prediction model (n = 187), and a validation cohort (n = 81). All MR images from the training cohort were reviewed by two radiologists. A prediction model was constructed by using MR imaging features that were independently associated with early recurrence with use of multiple logistic regression analysis. The performance of the prediction model in the validation cohort was evaluated with respect to discrimination (ie, whether the relative ranking of individual predictions of subsequent early recurrence is in the correct order).
In the training cohort, four MR imaging features were independently associated with early recurrence: rim enhancement (odds ratio OR = 3.83; 95% confidence interval CI: 1.39, 10.52), peritumoral parenchymal enhancement in the arterial phase (OR = 2.64; 95% CI: 1.27, 5.46), satellite nodule (OR = 4.07; 95% CI: 1.09, 15.21), and tumor size (OR = 1.66; 95% CI: 1.31, 2.09). A prediction model derived from these variables showed an area under the receiver operating characteristic curve (AUC) of 0.788 in the prediction of the risk of early recurrence in the training cohort. When applied to the validation cohort, this model showed good discrimination (AUC, 0.783).
The prediction model derived from rim enhancement, peritumoral parenchymal enhancement, satellite nodule, and tumor size can be used preoperatively to estimate the risk of early recurrence after resection of a single HCC.
This study aims to determine how randomly splitting a dataset into training and test sets affects the estimated performance of a machine learning model and its gap from the test performance under ...different conditions, using real-world brain tumor radiomics data. We conducted two classification tasks of different difficulty levels with magnetic resonance imaging (MRI) radiomics features: (1) "Simple" task, glioblastomas n = 109 vs. brain metastasis n = 58 and (2) "difficult" task, low- n = 163 vs. high-grade n = 95 meningiomas. Additionally, two undersampled datasets were created by randomly sampling 50% from these datasets. We performed random training-test set splitting for each dataset repeatedly to create 1,000 different training-test set pairs. For each dataset pair, the least absolute shrinkage and selection operator model was trained and evaluated using various validation methods in the training set, and tested in the test set, using the area under the curve (AUC) as an evaluation metric. The AUCs in training and testing varied among different training-test set pairs, especially with the undersampled datasets and the difficult task. The mean (±standard deviation) AUC difference between training and testing was 0.039 (±0.032) for the simple task without undersampling and 0.092 (±0.071) for the difficult task with undersampling. In a training-test set pair with the difficult task without undersampling, for example, the AUC was high in training but much lower in testing (0.882 and 0.667, respectively); in another dataset pair with the same task, however, the AUC was low in training but much higher in testing (0.709 and 0.911, respectively). When the AUC discrepancy between training and test, or generalization gap, was large, none of the validation methods helped sufficiently reduce the generalization gap. Our results suggest that machine learning after a single random training-test set split may lead to unreliable results in radiomics studies especially with small sample sizes.
Objectives
To evaluate the diagnostic performance of the LI-RADS (v2014) on gadoxetate-enhanced MRI prospectively applied in actual practice.
Methods
We retrospectively reviewed the prospectively ...written radiology reports of 143 treatment-naïve at-risk patients who underwent gadoxetate-enhanced liver MRI from January to December 2014, and identified 202 hepatic observations categorized using the LI-RADS. The diagnostic performances of LI-RADS categories for hepatocellular carcinoma (HCC) and hepatic malignancy were calculated.
Results
Twenty (69.0 %) of 29 LR-4, 73 (97.3 %) of 75 LR-5, and all of five (100 %) LR-5V observations were HCCs. The remaining two (2.7 %) LR-5 observations were combined hepatocellular-cholangiocarcinomas, while 10 (76.9 %) of 13 LR-M observations were HCCs. The sensitivity and specificity of LR-5/5V for HCC were 60.5 % and 97.3 %, respectively. Including LR-M in the diagnostic criteria for HCC increased sensitivity (68.2 %,
p
= 0.002) but decreased specificity without statistical significance (93.2 %,
p
= 0.154). LR-5/5V/M yielded sensitivity of 68.9 % and specificity of 100.0 % for hepatic malignancy.
Conclusions
LI-RADS v2014 was successfully applied on gadoxetate-enhanced MRI in clinical practice. LR-5/5V was the most specific diagnostic measure for HCC, but most LR-M observations were HCCs and a considerable portion of non-HCC malignancies were categorized as LR-4 or LR-5.
Key Points
•
LR-5/5V provided a highly specific diagnosis for HCC
.
•
Half of non-HCC malignancies were categorized as LR-4 or LR-5
.
•
The majority of LR-M observations were finally diagnosed as HCCs
.
•
More sensitive diagnosis of HCC was feasible with LR-5/5V/M on gadoxetate-enhanced MRI
.
•
Observations in either LR-5/5V or LR-M categories were definitely malignant
.
Purpose
To assess whether the radiomic features of diffusion tensor imaging (DTI) and conventional postcontrast T1-weighted (T1C) images can differentiate the epidermal growth factor receptor (EGFR) ...mutation status in brain metastases from non-small cell lung cancer (NSCLC).
Methods
A total of 99 brain metastases in 51 patients who underwent surgery or biopsy with underlying NSCLC and known EGFR mutation statuses (57 from EGFR wild type, 42 from EGFR mutant) were allocated to the training (57 lesions in 31 patients) and test (42 lesions in 20 patients) sets. Radiomic features (
n
= 526) were extracted from preoperative MR images including T1C and DTI. Radiomics classifiers were constructed by combinations of five feature selectors and four machine learning algorithms. The trained classifiers were validated on the test set, and the classifier performance was assessed by determining the area under the curve (AUC).
Results
EGFR mutation status showed an overall discordance rate of 12% between the primary tumors and corresponding brain metastases. The best performing classifier was a combination of the tree-based feature selection and linear discriminant algorithm and 5 features were selected (1 from ADC, 2 from fractional anisotropy, and 2 from T1C images), resulting in an AUC, accuracy, sensitivity, and specificity of 0.73, 78.6%, 81.3%, and 76.9% in the test set, respectively.
Conclusions
Radiomics classifiers integrating multiparametric MRI parameters may have potential in differentiating the EGFR mutation status in brain metastases from NSCLC.
Abstract
There is a growing need to develop novel strategies for the diagnosis of schizophrenia using neuroimaging biomarkers. We investigated the robustness of the diagnostic model for schizophrenia ...using radiomic features from T1-weighted and diffusion tensor images of the corpus callosum (CC). A total of 165 participants 86 schizophrenia and 79 healthy controls (HCs) were allocated to training (
N
= 115) and test (
N
= 50) sets. Radiomic features of the CC subregions were extracted from T1-weighted, apparent diffusion coefficient (ADC), and fractional anisotropy (FA) images (
N
= 1605). Following feature selection, various combinations of classifiers were trained, and Bayesian optimization was adopted in the best performing classifier. Discrimination, calibration, and clinical utility of the model were assessed. An online calculator was constructed to offer the probability of having schizophrenia. SHapley Additive exPlanations (SHAP) was applied to explore the interpretability of the model. We identified 30 radiomic features to differentiate participants with schizophrenia from HCs. The Bayesian optimized model achieved the highest performance, with an area under the curve (AUC), accuracy, sensitivity, and specificity of 0.89 (95% confidence interval: 0.81–0.98), 80.0, 83.3, and 76.9%, respectively, in the test set. The final model offers clinical probability in an online calculator. The model explanation by SHAP suggested that second-order features from the posterior CC were highly associated with the risk of schizophrenia. The multiparametric radiomics model focusing on the CC shows its robustness for the diagnosis of schizophrenia. Radiomic features could be a potential source of biomarkers that support the biomarker-based diagnosis of schizophrenia and improve the understanding of its neurobiology.
Abstract
We developed a tool to guide decision-making for early triage of COVID-19 patients based on a predicted prognosis, using a Korean national cohort of 5,596 patients, and validated the ...developed tool with an external cohort of 445 patients treated in a single institution. Predictors chosen for our model were older age, male sex, subjective fever, dyspnea, altered consciousness, temperature ≥ 37.5 °C, heart rate ≥ 100 bpm, systolic blood pressure ≥ 160 mmHg, diabetes mellitus, heart disease, chronic kidney disease, cancer, dementia, anemia, leukocytosis, lymphocytopenia, and thrombocytopenia. In the external validation, when age, sex, symptoms, and underlying disease were used as predictors, the AUC used as an evaluation metric for our model’s performance was 0.850 in predicting whether a patient will require at least oxygen therapy and 0.833 in predicting whether a patient will need critical care or die from COVID-19. The AUCs improved to 0.871 and 0.864, respectively, when additional information on vital signs and blood test results were also used. In contrast, the protocols currently recommended in Korea showed AUCs less than 0.75. An application for calculating the prognostic score in COVID-19 patients based on the results of this study is presented on our website (
https://nhimc.shinyapps.io/ih-psc/
), where the results of the validation ongoing in our institution are periodically updated.
PDF
