Machine Learning Insights: Predicting Hepatic Encephalopathy After TIPS Placement.

PURPOSE: To develop and assess machine learning (ML) models' ability to predict post-procedural hepatic encephalopathy (HE) following transjugular intrahepatic portosystemic shunt (TIPS) placement. MATERIALS AND METHODS: In this retrospective study, 327 patients who underwent TIPS for hepatic cirrhosis between 2005 and 2019 were analyzed. Thirty features (8 clinical, 10 laboratory, 12 procedural) were collected, and HE development regardless of severity was recorded one month follow-up. Univariate statistical analysis was performed with numeric and categoric data, as appropriate. Feature selection is used with a sequential feature selection model with fivefold cross-validation (CV). Three ML models were developed using support vector machine (SVM), logistic regression (LR) and CatBoost, algorithms. Performances were evaluated with nested fivefold-CV technique. RESULTS: Post-procedural HE was observed in 105 (32%) patients. Patients with variceal bleeding (p = 0.008) and high post-porto-systemic pressure gradient (p = 0.004) had a significantly increased likelihood of developing HE. Also, patients having only one indication of bleeding or ascites were significantly unlikely to develop HE as well as Budd-Chiari disease (p = 0.03). The feature selection algorithm selected 7 features. Accuracy ratios for the SVM, LR and CatBoost, models were 74%, 75%, and 73%, with area under the curve (AUC) values of 0.82, 0.83, and 0.83, respectively. CONCLUSION: ML models can aid identifying patients at risk of developing HE after TIPS placement, providing an additional tool for patient selection and management.

Machine Learning Models in Prediction of Treatment Response After Chemoembolization with MRI Clinicoradiomics Features.

PURPOSE: To evaluate machine learning models, created with radiomics and clinicoradiomics features, ability to predict local response after TACE. MATERIALS AND METHODS: 188 treatment-naïve patients (150 responders, 38 non-responders) with HCC who underwent TACE were included in this retrospective study. Laboratory, clinical and procedural information were recorded. Local response was evaluated by European Association for the Study of the Liver criteria at 3-months. Radiomics features were extracted from pretreatment pre-contrast enhanced T1 (T1WI) and late arterial-phase contrast-enhanced T1 (CE-T1) MRI images. After data augmentation, data were split into training and test sets (70/30). Intra-class correlations, Pearson's correlation coefficients were analyzed and followed by a sequential-feature-selection (SFS) algorithm for feature selection. Support-vector-machine (SVM) models were trained with radiomics and clinicoradiomics features of T1WI, CE-T1 and the combination of both datasets, respectively. Performance metrics were calculated with the test sets. Models' performances were compared with Delong's test. RESULTS: 1128 features were extracted. In feature selection, SFS algorithm selected 18, 12, 24 and 8 features in T1WI, CE-T1, combined datasets and clinical features, respectively. The SVM models area-under-curve was 0.86 and 0.88 in T1WI; 0.76, 0.71 in CE-T1 and 0.82, 0.91 in the combined dataset, with and without clinical features, respectively. The only significant change was observed after inclusion of clinical features in the combined dataset (p = 0.001). Higher WBC and neutrophil levels were significantly associated with lower treatment response in univariant analysis (p = 0.02, for both). CONCLUSION: Machine learning models created with clinical and MRI radiomics features, may have promise in predicting local response after TACE. LEVEL OF EVIDENCE: Level 4, Case-control study.

Prediction of carcinogenic human papillomavirus types in cervical cancer from multiparametric magnetic resonance images with machine learning-based radiomics models.

PURPOSE: This study aimed to evaluate the potential of machine learning-based models for predicting carcinogenic human papillomavirus (HPV) oncogene types using radiomics features from magnetic resonance imaging (MRI). METHODS: Pre-treatment MRI images of patients with cervical cancer were collected retrospectively. An HPV DNA oncogene analysis was performed based on cervical biopsy specimens. Radiomics features were extracted from contrast-enhanced T1-weighted images (CE-T1) and T2-weighted images (T2WI). A third feature subset was created as a combined group by concatenating the CE-T1 and T2WI subsets. Feature selection was performed using Pearson's correlation coefficient and wrapper- based sequential-feature selection. Two models were built with each feature subset, using support vector machine (SVM) and logistic regression (LR) classifiers. The models were validated using a five-fold cross-validation technique and compared using Wilcoxon's signed rank and Friedman's tests. RESULTS: Forty-one patients were enrolled in the study (26 were positive for carcinogenic HPV oncogenes, and 15 were negative). A total of 851 features were extracted from each imaging sequence. After feature selection, 5, 17, and 20 features remained in the CE-T1, T2WI, and combined groups, respectively. The SVM models showed 83%, 95%, and 95% accuracy scores, and the LR models revealed 83%, 81%, and 92.5% accuracy scores in the CE-T1, T2WI, and combined groups, respectively. The SVM algorithm performed better than the LR algorithm in the T2WI feature subset (P = 0.005), and the feature sets in the T2WI and the combined group performed better than CE-T1 in the SVM model (P = 0.033 and 0.006, respectively). The combined group feature subset performed better than T2WI in the LR model (P = 0.023). CONCLUSION: Machine learning-based radiomics models based on pre-treatment MRI can detect carcinogenic HPV status with discriminative accuracy.

Visceral to subcutaneous fat ratio predicts short-term mortality in patients with Covid 19. A multicenter study.

OBJECTIVE: To evaluate the association of body composition parameters with outcomes in Covid-19. METHODS: 173 patients hospitalized for Covid-19 infection in 6 European centers were included in this retrospective study. Measurements were performed at L3-level and comprised skeletal muscle index (SMI), muscle density (MD), and adipose tissue measurements [visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), intramuscular adipose tissue (IMAT), visceral-to-subcutaneous-adipose-tissue-area-ratio (VSR)]. The association with mortality, the need for intubation (MV), and the need for admission to ICU within 30 days were evaluated. RESULTS: Higher SAT density was associated with a greater risk of MV (OR = 1.071, 95%CI=(1.034;1.110), p < 0.001). Higher VAT density was associated with admission to ICU (OR = 1.068, 95%CI=(1.029;1.109), p < 0.001). Higher MD was a protective factor for MV and ICU admission (OR = 0.914, 95%CI=(0.870;0.960), p < 0.001; OR = 0.882, 95%CI=(0.832;0.934), p = 0.028). Higher VSR was associated with mortality (OR = 2.147, 95%CI=(1.022;4.512), p = 0.044). Male sex showed the strongest influence on the risk of ICU admission and MV. SMI was not associated with either parameter. CONCLUSION: In patients hospitalized for Covid-19 infection, higher VSR seems to be a strong prognostic factor of short-term mortality. Weak associations with clinical course were found for MD and adipose tissue measurements. Male sex was the strongest prognostic factor of adverse clinical course. ADVANCES IN KNOWLEDGE: VSR is a prognostic biomarker for 30-day mortality in patients hospitalized for Covid-19 disease.

Prediction of Response of Hepatocellular Carcinoma to Radioembolization: Machine Learning Using Preprocedural Clinical Factors and MR Imaging Radiomics.

PURPOSE: To create and evaluate the ability of machine learning-based models with clinicoradiomic features to predict radiologic response after transarterial radioembolization (TARE). MATERIALS AND METHODS: 82 treatment-naïve patients (65 responders and 17 nonresponders; median age: 65 years; interquartile range: 11) who underwent selective TARE were included. Treatment responses were evaluated using the European Association for the Study of the Liver criteria at 3-month follow-up. Laboratory, clinical, and procedural information were collected. Radiomic features were extracted from pretreatment contrast-enhanced T1-weighted magnetic resonance images obtained within 3 months before TARE. Feature selection consisted of intraclass correlation, followed by Pearson correlation analysis and finally, sequential feature selection algorithm. Support vector machine, logistic regression, random forest, and LightGBM models were created with both clinicoradiomic features and clinical features alone. Performance metrics were calculated with a nested 5-fold cross-validation technique. The performances of the models were compared by Wilcoxon signed-rank and Friedman tests. RESULTS: In total, 1,128 features were extracted. The feature selection process resulted in 12 features (8 radiomic and 4 clinical features) being included in the final analysis. The area under the receiver operating characteristic curve values from the support vector machine, logistic regression, random forest, and LightGBM models were 0.94, 0.94, 0.88, and 0.92 with clinicoradiomic features and 0.82, 0.83, 0.82, and 0.83 with clinical features alone, respectively. All models exhibited significantly higher performances when radiomic features were included (P = .028, .028, .043, and .028, respectively). CONCLUSIONS: Based on clinical and imaging-based information before treatment, machine learning-based clinicoradiomic models demonstrated potential to predict response to TARE.

Classification of retinoblastoma-1 gene mutation with machine learning-based models in bladder cancer.

Purpose: This study aims to evaluate the potential of machine learning algorithms built with radiomics features from computed tomography urography (CTU) images that classify RB1 gene mutation status in bladder cancer. Method: The study enrolled CTU images of 18 patients with and 54 without RB1 mutation from a public database. Image and data preprocessing were performed after data augmentation. Feature selection steps were consisted of filter and wrapper methods. Pearson's correlation analysis was the filter, and a wrapper-based sequential feature selection algorithm was the wrapper. Models with XGBoost, Random Forest (RF), and k-Nearest Neighbors (kNN) algorithms were developed. Performance metrics of the models were calculated. Models' performances were compared by using Friedman's test. Results: 8 features were selected from 851 total extracted features. Accuracy, sensitivity, specificity, precision, recall, F1 measure and AUC were 84%, 80%, 88%, 86%, 80%, 0.83 and 0.84, for XGBoost; 72%, 80%, 65%, 67%, 80%, 0.73 and 0.72 for RF; 66%, 53%, 76%, 67%, 53%, 0.60 and 0.65 for kNN, respectively. XGBoost model had outperformed kNN model in Friedman's test (p = 0.006). Conclusions: Machine learning algorithms with radiomics features from CTU images show promising results in classifying bladder cancer by RB1 mutation status non-invasively.

Diagnostic Performance of COVID-19 Reporting and Data System Classification Across Residents and Radiologists: A Retrospective Study.

OBJECTIVE: The aim of the study was to evaluate the interobserver agreement and diagnostic accuracy of COVID-19 Reporting and Data System (CO-RADS), in patients suspected COVID-19 pneumonia. METHODS: Two hundred nine nonenhanced chest computed tomography images of patients with clinically suspected COVID-19 pneumonia were included. The images were evaluated by 2 groups of observers, consisting of 2 residents-radiologists, using CO-RADS. Reverse transcriptase-polymerase chain reaction (PCR) was used as a reference standard for diagnosis in this study. Sensitivity, specificity, area under receiver operating characteristic curve (AUC), and intraobserver/interobserver agreement were calculated. RESULTS: COVID-19 Reporting and Data System was able to distinguish patients with positive PCR results from those with negative PCR results with AUC of 0.796 in the group of residents and AUC of 0.810 in the group of radiologists. There was moderate interobserver agreement between residents and radiologist with κ values of 0.54 and 0.57. CONCLUSIONS: The diagnostic performance of CO-RADS for predicting COVID-19 pneumonia showed moderate interobserver agreement between residents and radiologists.

Analysis of potential predatory journals in radiology.

PURPOSE: The aim of this study is to determine the presence and evaluate the features of potential predatory journals in the radiology field. METHODS: The presence of the keywords related to radiology listed in the name of journals was investigated in Beall's list. We have searched and recorded the features and the information of the included journals listed under the following headings: address and location, publishing features, editorial board, indexing features, submission, and peer-review processes. RESULTS: A total of 66 radiology journals from 27 publishers were identified from the updated version of the original Beall's list. Regarding the publishers, 33 journals (50%) reported an address in the United States of America, while others were from United Kingdom, India, Hong Kong, Iran, and Canada. While 44 journals' (67%) website reported a contact address, no addresses were declared in the website of 21 journals (32%). The median time of publication activity was 3.5 years (interquartile range [IQR], 1-5 years; range, 0-16 years). Thirty-five journals (53%) indicated their publication ethics policy on the website. Forty-seven (71%) journals reported a regular editorial board (EB) list. The competency of the EB was considered as "inappropriate" in 27 (41%) journals. Only 18% of the total number of EB members had affiliations related to radiology (n=286/1566). Forty journals (61%) did not report any indexing and database coverage. We found 26 journals (39%) which had a DOI number in its latest 5 articles. Fifty-nine (89%) journals clearly reported article processing change (APC) on the webpage. The median APC value was 641.43 USD (IQR, 300-918.75 USD; range, 100-2588 USD). Considering the latest 5 articles, the number of journals with radiologic images in all of the articles was 8 (12%). Mean peer-review time was 63.5 days (IQR, 21.75-87.5 days; range, 1-237 days) for the journals which indicated the submission and acceptance dates clearly. CONCLUSION: We demonstrated the several main characteristics of potential predatory journals in the radiology field such as reliability of the reported address, APC, publication frequencies, indexing features, features of published article and peer-review time which were all found to be similar to the characteristics of potential predatory journals in other biomedical fields.

RE: Distinguishing between Renal Cell Carcinoma and Fat Poor Angiomyolipoma in Diffusion-Weighted Imaging

No abstract available.

Primary and Metastatic Liver Malignancy: Utility Low and High B Value (1600-2000) in 3 Tesla MRI.

BACKGROUND/AIMS: To determine the utility of the quantitative apparent diffusion coefficient (ADC) values at various b values, in the differentiation of malignant hepatic masses on 3.0 Tesla (T) MRI. METHODOLOGY: We evaluated 81 consecutive patients presenting with 529 malignant masses in the liver. Of those patients 27 had a primary hepatic malignancy while the other 54 patients had metastases in the liver. Quantitative ADC values of malignant hepatic masses was measured at four b values (b 400, b 800, b 1600, b 2000 mm2/s) on MR-DWI. We compared the primary and metastatic tumors within their groups and also with each other in terms of their ADC values. RESULTS: In 4 various b value measurements, the mean ADC values of the primary and metastatic hepatic masses were 1. x 10(-3), 1.06 x 10(-3), 0.87 x 10(-3), and 0.736 x 10(-3)mm2/ seconds, 1.30 x 10(-3), 1.10 x 10(-3), 0.84 x 10(-3), and 0.715 x 10(-3) mm2/seconds respectively. There was no significant difference between mean ADC values of HCCs and metastases at b 400, 800, 1600 and 2000 gradients (P > 0.05). CONCLUSIONS: The ADC values obtained at intermediate (400, 800) and high (b 1600, 2000) diffusion gradients are not helpful in differentiation between HCCs and liver metastases.

Sonographic findings of hepatobiliary fascioliasis accompanied by extrahepatic expansion and ectopic lesions.

OBJECTIVES: The aim of the study was to describe the sonographic findings of hepatobiliary fascioliasis with extrahepatic expansion and ectopic lesions. METHODS: The study included 45 patients with fascioliasis. All diagnoses were confirmed via serologic enzyme-linked immunosorbent assays. Sonographic findings in the hepatobiliary system, extrahepatic expansion, and ectopic lesions were defined. RESULTS: The most common hepatic lesions were subcapsular localized, small, confluent, multiple hypoechoic nodules with poorly defined borders. We also detected ectopic lesion in 5 patients (11.1%) and live parasites in the gallbladder and bile duct in 11 (24.4%). CONCLUSIONS: The large spectrum of entities in the differential diagnosis of hepatobiliary fascioliasis may lead to misdiagnosis and incorrect treatment. However, the diagnosis can be made when the characteristic sonographic features are seen, such as heterogeneity of the liver with multiple poorly defined hypoechoic-isoechoic lesions and multiple echogenic nonshadowing particles in the gallbladder or common bile ducts. Nonetheless, the differential diagnosis of fascioliasis versus other hepatic lesions may still be difficult. In these situations, pathologic confirmation should be performed to exclude the possibility of malignancy.

Is there any significance of lung cancer histology to compare the diagnostic accuracies of (18)F-FDG-PET/CT and (99m)Tc-MDP BS for the detection of bone metastases in advanced NSCLC?

AIM OF THE STUDY: Bone scintigraphy (BS) and fluorine-18 deoxyglucose positron emission tomography computed tomography ((18)F-FDG-PET/CT) are widely used for the detection of bone involvement. The optimal imaging modality for the detection of bone metastases in histological subgroups of non-small cell lung cancer (NSCLC) remains ambiguous. The aim of this study was to compare the efficacy of (18)F-FDG-PET/C and 99mTc-methylene diphosphonate ((99m)Tc-MDP) BS in the detection of bone metastases of patients in NSCLC. Specifically, we compared the diagnostic accuracies of these imaging techniques evaluating bone metastasis in histological subgroups of NSCLC. MATERIAL AND METHODS: Fifty-three patients with advanced NSCLC, who had undergone both (18)F-FDG-PET/CT and BS and were eventually diagnosed as having bone metastasis, were enrolled in this retrospective study. RESULTS: The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of (18)F-FDG-PET/CT and BS were 90.4%, 99.4%, 98.1%, 96.6%, 97.0% and 84.6%, 93.1%, 82.5%, 93.2, 90.8%, respectively. The κ statistics were calculated for (18)F-FDG-PET/CT and BS. The κ-value was 0.67 between (18)F-FDG-PET/CT and BS in all patients. On the other hand, the κ-value was 0.65 in adenocarcinoma, and 0.61 in squamous cell carcinoma between (18)F-FDG-PET/CT and BS. The κ-values suggested excellent agreement between all patients and histological subgroups of NSCLC. CONCLUSIONS: (18)F-FDG-PET/CT was more favorable than BS in the screening of metastatic bone lesions, but the trend did not reach statistical significance in all patients and histological subgroups of NSCLC. Our results need to be validated in prospective and larger study clinical trials to further clarify this topic.

Investigation of the retrorenal colon frequency using computed tomography in patients with advanced scoliosis.

PURPOSE: The aim of this study is to investigate the frequency of retrorenal colon in patients with advanced scoliosis. MATERIALS AND METHODS: The existence of retrorenal colon was retrospectively investigated in 550 patients with vertebral scoliosis who had undergone abdominal CT scans at our institution between January 2008 and March 2012. The investigation was also carried out on a control group of 200 patients without scoliosis. RESULTS: Among the 550 patients with scoliosis, 100 patients had advanced scoliosis necessitating treatment. Among these 100 patients with advanced scoliosis, retrorenal colon was detected in a total of 25 patients (25 %). The variation was observed on the right side in eight patients (two males, six females) (8 %), on the left side in 15 patients (five males, ten females) (15 %), and bilaterally in two patients (both females) (2 %). In the control group consisting of 200 individuals, retrorenal colon was detected in seven subjects (3.5 %), among which six were on the left and one was on the right. The difference between the incidence of retrorenal colon observed in the patients with advanced scoliosis and those without scoliosis was found to be statistically significant (p < 0.001). CONCLUSION: Since the frequency of retrorenal colon in patients with advanced scoliosis is significantly higher than the control group without scoliosis, the risk of experiencing complications during renal interventions including renal biopsy is also higher. Therefore, these patients should undergo a detailed CT examination before these procedures, and renal interventions should be planned according to findings.

Relation between electrophysiological findings and diffusion weighted magnetic resonance imaging in ulnar neuropathy at the elbow.

AIM: As only a limited number of studies have used diffusion-weighted imaging (DWI) and conventional magnetic resonance imaging (MRI) in patients with ulnar neuropathy at the elbow (UNE), the present study aimed to investigate the diagnostic value of the non-invasive DWI technique in patients with UNE. METHODS: A total of 26 elbows in 19 healthy controls (age range: 22-56 years) with no symptoms and 24 elbows in 21 symptomatic patients (age range: 21-46 years) with cubital tunnel syndrome underwent DWI. The electrophysiological and clinical criteria for the diagnosis of UNE were examined. RESULTS: No pathological signal from the ulnar nerve was detected in the healthy controls, whereas there was an increase in signals on DWI in all patients with UNE. On T2-weighted (T2W) imaging, there was increased signal intensity in 20 elbows, while low signal intensity was observed in the remaining four. A positive correlation was found between disease duration and presence of hyperintensity (P=0.044, r=0.42) on T2W images. CONCLUSION: DWI can be used together with electrophysiological methods for the diagnosis of UNE. Furthermore, DWI might be preferred in some cases, as it is non-invasive compared with the electrophysiological method for UNE diagnosis.

An incidental case of biliary fascioliasis with subtle clinical findings: US and MRCP findings.

BACKGROUND: Fascioliasis is a disease caused by the trematode Fasciola hepatica. Cholangitis is a common clinical manifestation. Although fascioliasis may show various radiological and clinical features, cases without biliary dilatation are rare. CASE REPORT: We present unique ultrasound (US) and magnetic resonance cholangiopancreatography (MRCP) findings of a biliary fascioliasis case which doesn't have biliary obstruction or cholestasis. Radiologically, curvilinear parasites compatible with juvenile and mature Fasciola hepatica within the gallbladder and common bile duct were found. The parasites appear as bright echogenic structures with no acoustic shadow on US and hypo-intense curvilinear lesions on T2 weighted MRCP images. CONCLUSIONS: Imaging studies may significantly contribute to the diagnosis of patients with subtle clinical and laboratory findings, particularly in endemic regions.