Prediction of human epidermal growth factor receptor 2 (HER2) status in breast cancer by mammographic radiomics features and clinical characteristics: a multicenter study.

OBJECTIVES: To preoperatively evaluate the human epidermal growth factor 2 (HER2) status in breast cancer using mammographic radiomics features and clinical characteristics on a multi-vendor and multi-center basis. METHODS: This multi-center study included a cohort of 1512 Chinese female with invasive ductal carcinoma of no special type (IDC-NST) from two different hospitals and five devices (1332 from Institution A, used for training and testing the models, and 180 women from Institution B, as the external validation cohort). The Gradient Boosting Machine (GBM) was employed to establish radiomics and multiomics models. Model efficacy was evaluated by the area under the curve (AUC). RESULTS: The number of HER2-positive patients in the training, testing, and external validation cohort were 245(26.3%), 105 (26.3.8%), and 51(28.3%), respectively, with no statistical differences among the three cohorts (p = 0.842, chi-square test). The radiomics model, based solely on the radiomics features, achieved an AUC of 0.814 (95% CI, 0.784-0.844) in the training cohort, 0.776 (95% CI, 0.727-0.825) in the testing cohort, and 0.702 (95% CI, 0.614-0.790) in the external validation cohort. The multiomics model, incorporated radiomics features with clinical characteristics, consistently outperformed the radiomics model with AUC values of 0.838 (95% CI, 0.810-0.866) in the training cohort, 0.788 (95% CI, 0.741-0.835) in the testing cohort, and 0.722 (95% CI, 0.637-0.811) in the external validation cohort. CONCLUSIONS: Our study demonstrates that a model based on radiomics features and clinical characteristics has the potential to accurately predict HER2 status of breast cancer patients across multiple devices and centers. CLINICAL RELEVANCE STATEMENT: By predicting the HER2 status of breast cancer reliably, the presented model built upon radiomics features and clinical characteristics on a multi-vendor and multi-center basis can help in bolstering the model's applicability and generalizability in real-world clinical scenarios. KEY POINTS: • The mammographic presentation of breast cancer is closely associated with the status of human epidermal growth factor receptor 2 (HER2). • The radiomics model, based solely on radiomics features, exhibits sub-optimal performance in the external validation cohort. • By combining radiomics features and clinical characteristics, the multiomics model can improve the prediction ability in external data.

Distinct MRI characteristics of spinal cord diffuse midline glioma, H3 K27-altered in comparison to spinal cord glioma without H3 K27-alteration and demyelination disorder.

BACKGROUND: An applicable magnetic resonance imaging (MRI) biomarker for diffuse midline glioma (DMG), H3 K27-altered of the spinal cord is important for non-invasive diagnosis. PURPOSE: To evaluate the efficacy of conventional MRI (cMRI) in distinguishing between DMGs, H3 K27-altered, gliomas without H3 K27-alteration, and demyelinating lesions in the spinal cord. MATERIAL AND METHODS: Between January 2017 and February 2023, patients with pathology-confirmed spinal cord gliomas (including ependymomas) with definite H3 K27 status and demyelinating diseases diagnosed by recognized criteria were recruited as the training set for this retrospective study. Morphologic parameter assessment was performed by two neuroradiologists on T1-weighted, T2-weighted, and contrast-enhanced T1-weighted imaging. Variables with high inter- and intra-observer agreement were included in univariable correlation analysis and multivariable logistic regression. The performance of the final model was verified by internal and external testing sets. RESULTS: The training cohort included 21 patients with DMGs (13 men; mean age = 34.57 ± 13.489 years), 21 with wild-type gliomas (10 men; mean age = 46.76 ± 17.017 years), and 20 with demyelinating diseases (5 men; mean age = 49.50 ± 18.872 years). A significant difference was observed in MRI features, including cyst(s), hemorrhage, pial thickening with enhancement, and the maximum anteroposterior diameter of the spinal cord. The prediction model, integrating age, age2, and morphological characteristics, demonstrated good performance in the internal and external testing cohort (accuracy: 0.810 and 0.800, specificity: 0.810 and 0.720, sensitivity: 0.872 and 0.849, respectively). CONCLUSION: Based on cMRI, we developed a model with good performance for differentiating among DMGs, H3 K27-altered, wild-type glioma, and demyelinating lesions in the spinal cord.

Glioblastomas with and without peritumoral fluid-attenuated inversion recovery (FLAIR) hyperintensity present morphological and microstructural differences on conventional MR images.

OBJECTIVES: Glioblastoma (GB) without peritumoral fluid-attenuated inversion recovery (FLAIR) hyperintensity is atypical and its characteristics are barely known. The aim of this study was to explore the differences in pathological and MRI-based intrinsic features (including morphologic and first-order features) between GBs with peritumoral FLAIR hyperintensity (PFH-bearing GBs) and GBs without peritumoral FLAIR hyperintensity (PFH-free GBs). METHODS: In total, 155 patients with pathologically diagnosed GBs were retrospectively collected, which included 110 PFH-bearing GBs and 45 PFH-free GBs. The pathological and imaging data were collected. The Visually AcceSAble Rembrandt Images (VASARI) features were carefully evaluated. The first-order radiomics features from the tumor region were extracted from FLAIR, apparent diffusion coefficient (ADC), and T1CE (T1-contrast enhanced) images. All parameters were compared between the two groups of GBs. RESULTS: The pathological data showed more alpha thalassemia/mental retardation syndrome X-linked (ATRX)-loss in PFH-free GBs compared to PFH-bearing ones (p < 0.001). Based on VASARI evaluation, PFH-free GBs had larger intra-tumoral enhancing proportion and smaller necrotic proportion (both, p < 0.001), more common non-enhancing tumor (p < 0.001), mild/minimal enhancement (p = 0.003), expansive T1/FLAIR ratio (p < 0.001) and solid enhancement (p = 0.009), and less pial invasion (p = 0.010). Moreover, multiple ADC- and T1CE-based first-order radiomics features demonstrated differences, especially the lower intensity heterogeneity in PFH-free GBs (for all, adjusted p < 0.05). CONCLUSIONS: Compared to PFH-bearing GBs, PFH-free ones demonstrated less immature neovascularization and lower intra-tumoral heterogeneity, which would be helpful in clinical treatment stratification. CLINICAL RELEVANCE STATEMENT: Glioblastomas without peritumoral FLAIR hyperintensity show less immature neovascularization and lower heterogeneity leading to potential higher treatment benefits due to less drug resistance and treatment failure. KEY POINTS: • The study explored the differences between glioblastomas with and without peritumoral FLAIR hyperintensity. • Glioblastomas without peritumoral FLAIR hyperintensity showed less necrosis and contrast enhancement and lower intensity heterogeneity. • Glioblastomas without peritumoral FLAIR hyperintensity had less immature neovascularization and lower tumor heterogeneity.

Oligodendrocyte Transcription Factor 2 as a Potential Prognostic Biomarker of Glioblastoma: Kaplan-Meier Analysis and the Development of a Binary Predictive Model Based on Visually Accessible Rembrandt Image and Magnetic Resonance Imaging Radiomic Features.

OBJECTIVE: Oligodendrocyte transcription factor 2 (OLIG2) is universally expressed in human glioblastoma (GB). Our study explores whether OLIG2 expression impacts GB patients' overall survival and establishes a machine learning model for OLIG2 level prediction in patients with GB based on clinical, semantic, and magnetic resonance imaging radiomic features. METHODS: Kaplan-Meier analysis was used to determine the optimal cutoff value of the OLIG2 in 168 GB patients. Three hundred thirteen patients enrolled in the OLIG2 prediction model were randomly divided into training and testing sets in a ratio of 7:3. The radiomic, semantic, and clinical features were collected for each patient. Recursive feature elimination (RFE) was used for feature selection. The random forest (RF) model was built and fine-tuned, and the area under the curve was calculated to evaluate the performance. Finally, a new testing set excluding IDH-mutant patients was built and tested in a predictive model using the fifth edition of the central nervous system tumor classification criteria. RESULTS: One hundred nineteen patients were included in the survival analysis. Oligodendrocyte transcription factor 2 was positively associated with GB survival, with an optimal cutoff of 10% ( P = 0.00093). One hundred thirty-four patients were eligible for the OLIG2 prediction model. An RFE-RF model based on 2 semantic and 21 radiomic signatures achieved areas under the curve of 0.854 in the training set, 0.819 in the testing set, and 0.825 in the new testing set. CONCLUSIONS: Glioblastoma patients with ≤10% OLIG2 expression tended to have worse overall survival. An RFE-RF model integrating 23 features can predict the OLIG2 level of GB patients preoperatively, irrespective of the central nervous system classification criteria, further guiding individualized treatment.

A Comparative and Summative Study of Radiomics-based Overall Survival Prediction in Glioblastoma Patients.

PURPOSE: This study aimed to assess different machine learning models based on radiomic features, Visually Accessible Rembrandt Images features and clinical characteristics in overall survival prediction of glioblastoma and to identify the reproducible features. MATERIALS AND METHODS: Patients with preoperative magnetic resonance scans were allocated into 3 data sets. The Least Absolute Shrinkage and Selection Operator was used for feature selection. The prediction models were built by random survival forest (RSF) and Cox regression. C-index and integrated Brier scores were calculated to compare model performances. RESULTS: Patients with cortical involvement had shorter survival times in the training set (P = 0.006). Random survival forest showed higher C-index than Cox, and the RSF model based on the radiomic features was the best one (testing set: C-index = 0.935 ± 0.023). Ten reproducible radiomic features were summarized. CONCLUSIONS: The RSF model based on radiomic features had promising potential in predicting overall survival of glioblastoma. Ten reproducible features were identified.

Performances of clinical characteristics and radiological findings in identifying COVID-19 from suspected cases.

BACKGROUND: To identify effective factors and establish a model to distinguish COVID-19 patients from suspected cases. METHODS: The clinical characteristics, laboratory results and initial chest CT findings of suspected COVID-19 patients in 3 institutions were retrospectively reviewed. Univariate and multivariate logistic regression were performed to identify significant features. A nomogram was constructed, with calibration validated internally and externally. RESULTS: 239 patients from 2 institutions were enrolled in the primary cohort including 157 COVID-19 and 82 non-COVID-19 patients. 11 features were selected by LASSO selection, and 8 features were found significant using multivariate logistic regression analysis. We found that the COVID-19 group are more likely to have fever (OR 4.22), contact history (OR 284.73), lower WBC count (OR 0.63), left lower lobe involvement (OR 9.42), multifocal lesions (OR 8.98), pleural thickening (OR 5.59), peripheral distribution (OR 0.09), and less mediastinal lymphadenopathy (OR 0.037). The nomogram developed accordingly for clinical practice showed satisfactory internal and external validation. CONCLUSIONS: In conclusion, fever, contact history, decreased WBC count, left lower lobe involvement, pleural thickening, multifocal lesions, peripheral distribution, and absence of mediastinal lymphadenopathy are able to distinguish COVID-19 patients from other suspected patients. The corresponding nomogram is a useful tool in clinical practice.

Triage Modeling for Differential Diagnosis Between COVID-19 and Human Influenza A Pneumonia: Classification and Regression Tree Analysis.

Background: The coronavirus disease 2019 (COVID-19) pandemic has lasted much longer than an influenza season, but the main signs, symptoms, and some imaging findings are similar in COVID-19 and influenza patients. The aim of the current study was to construct an accurate and robust model for initial screening and differential diagnosis of COVID-19 and influenza A. Methods: All patients in the study were diagnosed at Fuyang No. 2 People's Hospital, and they included 151 with COVID-19 and 155 with influenza A. The patients were randomly assigned to training set or a testing set at a 4:1 ratio. Predictor variables were selected based on importance, assessed by random forest algorithms, and analyzed to develop classification and regression tree models. Results: In the optimal model A, the best single predictor of COVID-19 patients was a normal or high level of low-density lipoprotein cholesterol, followed by low level of creatine kinase, then the presence of <3 respiratory symptoms, then a highest temperature on the first day of admission <38°C. In the suboptimal model B, the best single predictor of COVID-19 was a low eosinophil count, then a normal monocyte ratio, then a normal hematocrit value, then a highest temperature on the first day of admission of <37°C, then a complete lack of respiratory symptoms. Conclusions: The two models provide clinicians with a rapid triage tool. The optimal model can be used to developed countries/regions and major hospitals, and the suboptimal model can be used in underdeveloped regions and small hospitals.