Artificial intelligence, BI-RADS evaluation and morphometry: A novel combination to diagnose breast cancer using ultrasonography, results from multi-center cohorts.

PURPOSE: To develop and validate a machine learning (ML) model for the classification of breast lesions on ultrasound images. METHOD: In the present study, three separate data cohorts containing 1288 breast lesions from three countries (Malaysia, Iran, and Turkey) were utilized for MLmodel development and external validation. The model was trained on ultrasound images of 725 breast lesions, and validation was done separately on the remaining data. An expert radiologist and a radiology resident classified the lesions based on the BI-RADS lexicon. Thirteen morphometric features were selected from a contour of the lesion and underwent a three-step feature selection process. Five features were chosen to be fed into the model separately and combined with the imaging signs mentioned in the BI-RADS reference guide. A support vector classifier was trained and optimized. RESULTS: The diagnostic profile of the model with various input data was compared to the expert radiologist and radiology resident. The agreement of each approach with histopathologic specimens was also determined. Based on BI-RADS and morphometric features, the model achieved an area under the receiver operating characteristic (ROC) curve (AUC) of 0.885, which is higher than the expert radiologist and radiology resident performances with AUC of 0.814 and 0.632, respectively in all cohorts. DeLong's test also showed that the AUC of the ML protocol was significantly different from that of the expert radiologist (ΔAUCs = 0.071, 95%CI: (0.056, 0.086), P = 0.005). CONCLUSIONS: These results support the possible role of morphometric features in enhancing the already well-excepted classification schemes.

Clinical and radiologic characteristics of symptomatic pregnant women with COVID-19 pneumonia

Objective: To describe the radiological features, diagnostic accuracy and features of imaging studies and their relation with clinical course of Coronavirus disease-2019 (COVID-19) pneumonia in pregnant women. Material and Methods: The clinical, laboratory and radiological features of symptomatic pregnant women suspected of COVID-19 were retrospectively reviewed. Chest radiography (CXR) and chest computed tomography (CT) findings of COVID-19 in pregnant women were identified. Results: Fifty-five of eighty-one pregnant women were included in the final analysis. The most common admission symptoms were dry cough (45.4%), fever (29.1%) and dyspnea (34.5%). Radiological imaging studies were performed in 34 (61.8%) patients. Fourteen (66.7%) of the laboratory-confirmed COVID-19 patients had parenchymal abnormalities on CXR, and most common abnormalities were airspace opacities (61.9%) and prominent bronchovascular shadows (28.6%). Seventeen (85.0%) of the patients had parenchymal abnormalities consistent with COVID-19 on their chest CT. Chest CT most commonly showed bilateral (88.2%), multilobe (100%) involvement; peripheral and central distribution (70.6%); patchy-shape (94.1%) and ground-glass opacity (94.1%). The sensitivity of CXR and chest CT was calculated as 66.7% and 83.3%, respectively. Preterm birth rate was 41.2% (n=7/17). Five (9.1%) of the 55 pregnant women admitted to the intensive care unit, three of those developed acute respiratory distress syndrome and one died. Conclusion: This study describes the main radiological features of symptomatic pregnant women infected with COVID-19. The refusal rate among pregnant women for the imaging modalities involving ionizing radiation was high but these had high sensitivity for COVID-19 diagnosis. The preterm birth and cesarean section rates were observed as remarkably increased.

A comparison of clinical, laboratory and chest CT findings of laboratory-confirmed and clinically diagnosed COVID-19 patients at first admission.

PURPOSE: This study aims to identify chest computed tomography (CT) characteristics of coronavirus disease 2019 (COVID-19), investigate the association between CT findings and laboratory or demographic findings, and compare the accuracy of chest CT with reverse transcription-polymerase chain reaction (RT-PCR). METHODS: Overall, 120 of 159 consecutive cases isolated due to suspected COVID-19 at our hospital between 17 and 25 March 2020 were included in this retrospective study. All patients underwent both chest CT and RT-PCR at first admission. The patients were divided into two groups: laboratory-confirmed COVID-19 and clinically diagnosed COVID-19. Clinical findings, laboratory findings, radiologic features and CT severity index (CT-SI) of the patients were noted. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of chest CT were calculated for the diagnosis of COVID-19, using RT-PCR as reference. RESULTS: The laboratory-confirmed and clinically diagnosed COVID-19 groups consisted of 69 (M/F 43/26, mean age 50.9±14.0 years) and 51 patients (M/F 24/27, mean age 50.9±18.8 years), respectively. Dry cough (62.3% vs. 52.9%), fever (30.4% vs. 25.5%) and dyspnea (23.2% vs. 27.5%) were the most common admission symptoms in the laboratory-confirmed and clinically diagnosed COVID-19 groups, respectively. Bilateral multilobe involvement (83.1% vs. 57.5%), peripheral distribution (96.9% vs. 97.5%), patchy shape (75.4% vs. 70.0%), ground-glass opacities (GGO) (96.9% vs. 100.0%), vascular enlargement (56.9% vs. 50.0%), intralobular reticular density (40.0% vs. 40.0%) and bronchial wall thickening (27.7% vs. 45.0%) were the most common CT findings in the laboratory-confirmed and clinically diagnosed COVID-19 subgroups, respectively. Except for the bilateral involvement and white blood cell (WBC) count, no difference was found between the clinical, laboratory, and parenchymal findings of the two groups. Positive correlation was found between CT-SI and, lactate dehydrogenase (LDH) and C-reactive protein (CRP) values in the laboratory-confirmed COVID-19 subgroup. Chest CT and RT-PCR positivity rates among patients with suspected COVID-19 were 87.5% (105/120) and 57.5% (69/120), respectively. The sensitivity, specificity, PPV, NPV and accuracy rates of chest CT were determined as 94.2% (95% confidence interval [CI], 85.8-98.4), 21.57% (95% CI, 11.3-35.3), 61.90% (95% CI, 58.2-65.5), 73.3% (95% CI, 48.2-89.1) and 63.3% (95% CI, 54.1-71.9), respectively. CONCLUSION: Chest CT has high sensitivity and low specificity in the diagnosis of COVID-19. The clinical, laboratory, and CT findings of laboratory-confirmed and clinically diagnosed COVID-19 patients are similar.