Two-Step Machine Learning to Diagnose and Predict Involvement of Lungs in COVID-19 and Pneumonia using CT Radiomics

ObjectiveWe aimed to develop a two-step machine learning (ML) based model to diagnose and predict involvement of lungs in COVID-19 and non COVID-19 pneumonia patients using CT chest radiomic features. MethodsThree hundred CT scans (3-classes: 100 COVID-19, 100 pneumonia, and 100 healthy subjects) were enrolled in this study. Diagnostic task included 3-class classification. For severity prediction, two radiologists scored involvement of lungs in COVID-19 and pneumonia scans based on percentage of involvement in all 5 lobes. Datasets were classified into mild (0-25%), moderate (26-50%), and severe (>50%). Whole lungs were segmented utilizing deep learning-based segmentation method. Altogether, 107 features including shape, first-order histogram, second and high order texture features were extracted. For both tasks, datasets were randomly divided into 90% training sets (70% and 30% for training and validation, respectively) and 10% test sets. Pearson correlation coefficient (PCC[≥]90%) was performed to exclude highly correlated features. Subsequently, different feature selection algorithms (Correlation attribute evaluation, Information gain attribute, Wrapper Subset selection algorithm, Relief method, and Correlation-based feature selection) were assessed. The most pertinent features were finally selected using voting method based on the evaluation of all algorithms. Several ML-based supervised algorithms were utilized, namely Naive Bays, Support Vector Machine, Bagging, Random Forest, K-nearest neighbors, Decision Tree and Ensemble Meta voting. The synthetic minority oversampling technique (SMOTE) was used to balance the three classes in training sets. The optimal model was first selected based on precision, recall and area-under-curve (AUC) by randomizing the training/validation sets 20 times, followed by testing using the test set. To ensure the repeatability of the results, the entire process was repeated 50 times. ResultsNine pertinent features (2 shape, 1 first-order, and 6 second-order features) were obtained after feature selection for both phases. In diagnostic task, the performance of 3-class classification using Random Forest was 0.909{+/-}0.026, 0.907{+/-}0.056, 0.902{+/-}0.044, 0.939{+/-}0.031, and 0.982{+/-}0.010 for precision, recall, F1-score, accuracy, and AUC, respectively. The severity prediction task using Random Forest achieved 0.868{+/-}0.123 precision, 0.865{+/-}0.121 recall, 0.853{+/-}0.139 F1-score, 0.934{+/-}0.024 accuracy, and 0.969{+/-}0.022 AUC. ConclusionThe two-phase ML-based model accurately classified COVID-19 and pneumonia patients using CT radiomics, and adequately predicted severity of lungs involvement. This 2-steps model showed great potential in assessing COVID-19 CT images towards improved management of patients.

Radiologic evaluation of congenital anomalies of anterior and posterior arch of atlas in Omani subjects / 대한해부학회지

The atlas (C1) is known to present congenital anomalies in its anterior and posterior arches. The reported incidence of C1 anomalies is varied among the ethnic groups. We sought to determine the prevalence and various existing variations of C1 arch congenital anomalies in Omani subjects. This study was carried out by reviewing the cervical spine computed tomography scans of all the patients who had been referred to the Radiology Department, Sultan Qaboos University Hospital.Descriptive statistics and chi-square test were employed to analyse the data. A total of 663 subjects aged ≥18 years were included in the present study. Overall prevalence of C1 arch anomalies was 4.37% with 4.07% of isolated posterior arch anomalies, 0.3% of combined anterior and posterior arch anomalies. Among isolated posterior arch anomalies, type A and type B posterior arch defects were found in 3.77% and 0.3% of cases, respectively. Atlanto-occipital assimilation was noted in one case of total study subjects. The prevalence rate of C1 arch anomalies is relatively high in Omani subjects. The baseline data of C1 arch anomalies reported in the present study has a great impact on clinical practice, due to the fact that studying and evaluating the types of congenital anomalies helps in their accurate diagnosis and early intervention.

Radiological study of fabella in Omani subjects at a tertiary care center / 대한해부학회지

Ethnic diversity is associated with variability in the prevalence rates of fabella. We aimed to evaluate the prevalence and the radiological features of fabella in Omani patients. This is a retrospective analysis of hospital electronic database of patients referred for radiological investigations (radiographs and magnetic resonance imaging) of the knee, at a tertiary care referral center. Descriptive statistics were performed to determine the prevalence of fabella. Chi-square test was used to determine the association between sex or age with respect to the presence of fabella. A total of 813 knee radiographs were reviewed for the presence of fabella. Fabella was found in 24.1% of total cases. A statistically significant sex difference was observed with respect to the presence of fabella in left knees in males (P<0.01). The presence of fabella was significantly associated with age groups for the right (P<0.05) and left knees (P<0.01). In magnetic resonance imaging film reviews, all the identified fabellae (20.2%) were bony structures and were located within the lateral head of the gastrocnemius muscle. There were no cartilaginous fabellae detected. The current study revealed a prevalence of 24.1% of fabella in Omani subjects which is almost similar to the results as seen in Caucasian ethnic populations.

Radiological study of fabella in Omani subjects at a tertiary care center / 대한해부학회지

Ethnic diversity is associated with variability in the prevalence rates of fabella. We aimed to evaluate the prevalence and the radiological features of fabella in Omani patients. This is a retrospective analysis of hospital electronic database of patients referred for radiological investigations (radiographs and magnetic resonance imaging) of the knee, at a tertiary care referral center. Descriptive statistics were performed to determine the prevalence of fabella. Chi-square test was used to determine the association between sex or age with respect to the presence of fabella. A total of 813 knee radiographs were reviewed for the presence of fabella. Fabella was found in 24.1% of total cases. A statistically significant sex difference was observed with respect to the presence of fabella in left knees in males (P<0.01). The presence of fabella was significantly associated with age groups for the right (P<0.05) and left knees (P<0.01). In magnetic resonance imaging film reviews, all the identified fabellae (20.2%) were bony structures and were located within the lateral head of the gastrocnemius muscle. There were no cartilaginous fabellae detected. The current study revealed a prevalence of 24.1% of fabella in Omani subjects which is almost similar to the results as seen in Caucasian ethnic populations.

Prevalence of diffuse idiopathic skeletal hyperostosis among elderly subjects referred for radiological investigation in tertiary hospital at Oman / 대한해부학회지

Diffuse idiopathic skeletal hyperostosis (DISH) is characterized by calcification of different entheseal sites including the anterior longitudinal ligament of the spine. There is no documented information about DISH in Oman. This study determined the prevalence of DISH and associated factors among elderly subjects in a national tertiary care referral hospital in Oman. This retrospective study reviewed chest X-rays of all patients aged more than 50 years, referred to the radiology department of Sultan Qaboos University Hospital in the year 2016, based on the Resnick's criteria. The prevalence was expressed as proportions across age groups and sex. Chi-square test and logistic regression analysis was done to determine the association of the age and sex with DISH. A total of 1305 chest X-rays of patients were reviewed. The overall prevalence of DISH was 10%, with male to female ratio of 1.56:1. The odds ratio for males and increasing age were 1.63 (95% confidence interval [CI], 1.12–2.3; P<0.05) and 1.34 (95% CI, 1.14–1.58; P<0.001) respectively. The prevalence increased with age to maximum of 13% in the age group of ≤80. The pre-stage DISH prevalence was 9.3% and more frequently observed among males. Prevalence of DISH in the national tertiary care referral center in Oman is lower than in Jewish population, almost similar to Japanese, but higher than in Koreans. DISH prevalence is positively associated with age and sex. It is necessary to take appropriate precautionary measures to target the ageing population in Oman, especially elderly males.