Purpose To develop and evaluate an attention-based multimodal model for improving the prediction of overall survival in patients with glioblastoma by integrating clinical and imaging features. Material and methods Data from 358 glioblastoma patients were obtained from The Cancer Imaging Archive. The dataset included clinical data and three magnetic resonance imaging (MRI) sequences. Three categories of features – clinical, radiomics, and deep features – were used for survival prediction. Radiomics features were extracted using PyRadiomics, while deep features were extracted from MRI images using a Swin UNEt TRansformers. An attention-based multimodal predictive model was designed to integrate these three feature sets. The DeepHit framework was used for survival modeling. Model performance was evaluated using the concordance index (C-index) and the Brier score. In addition, accuracy, precision, recall, area under the curve (AUC), and F1-score were calculated to predict mortality risk at different time intervals. Results The proposed model achieved a C-index of 0.91 and a Brier score of 0.14. For identifying mortality risk within 6 months, the model achieved an accuracy of 0.93, precision of 0.87, recall of 0.92, AUC of 0.96, and an F1-score of 0.89. For mortality risk within 12 months, the corresponding values were 0.91, 0.87, 0.85, 0.92, and 0.88, respectively. For mortality risk within 24 months, the model achieved an accuracy of 0.75, precision of 0.71, recall of 0.70, AUC of 0.94, and an F1-score of 0.80. Conclusions The proposed model, leveraging a combined radiomics-based approach, demonstrated the capability to predict overall survival in patients with glioblastoma. However, to ensure generalizability, further validation using multicenter datasets is required.
Purpose To investigate the association between hepatic function and liver and renal parenchymal signal intensities (SI) during the hepatobiliary phase of gadoxetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI), and to provide imaging evidence for the combined evaluation of hepatic and renal function in cirrhotic patients. Material and methods Seventy-eight patients who underwent Gd-EOB-DTPA-enhanced MRI were retrospectively enrolled and divided into a normal control group (n = 28), Child-Pugh grade A group (n = 25), grade B group (n = 17), and grade C group (n = 8). Pre-enhancement SI of the liver and renal cortex/medulla (SIliver-pre, SIcortex-pre, SImedulla-pre) and post-enhancement SI (SIliver-post, SIcortex-post, SImedulla-post) were measured. The signal intensity differences between the renal cortex and medulla (SI[cortex-medulla]pre and SI[cortex-medulla]post) were calculated. Relative enhancement rates (RE) were calculated. One-way analysis of variance (ANOVA), Kruskal-Wallis, and Mann-Whitney U tests were used to compare signal intensity parameters among the normal liver function group and different Child-Pugh grade groups (A, B, C). Results As the Child-Pugh grade increased, SIliver-pre, SIliver-post, and REliver showed statistically significant differences (p < 0.05). SIcortex-pre and SIcortex-post exhibited significant differences in subgroup comparisons (p < 0.05), but REcortex showed no statistical significance (p = 0.859). Pre- and post-enhancement renal medullary signals and SI[cortex-medulla] differences showed no significant variations across all groups (p > 0.05). Conclusions Hepatic dysfunction was associated with lower hepatic signal intensity and enhancement in the hepatobiliary phase. Significant differences were observed in the renal cortical signal intensity before and after contrast administration, whereas the renal medullary signals and corticomedullary signal differences showed no significant variation.
Purpose Acute pulmonary embolism (APE) is a potentially life-threatening condition requiring rapid risk stratification. Although computed tomography (CT) pulmonary angiography is the diagnostic gold standard, the prognostic value of CT-derived radiological markers remains incompletely defined. Material and methods This retrospective study included 91 patients with APE. CT-derived radiological parameters – including the Qanadli obstruction index (QI), Mastora score, right ventricle/left ventricle (RV/LV) ratio, and RV wall thickness – were evaluated together with clinical severity scores (Pulmonary Embolism Severity Index [PESI], and its simplified version [sPESI]) and laboratory markers. Associations with disease severity and intensive care unit (ICU) admission were assessed using correlation and receiver operating characteristic (ROC) analyses. Results Among the patients, 52.7% were female and 27.5% required ICU admission. PESI, sPESI, QI, Mastora score, and RV/LV ratio were significantly higher in ICU-admitted patients (all p < 0.01). QI correlated significantly with PESI (r = 0.433), disease severity (r = 0.565), D-dimer (r = 0.510), and troponin levels (r = 0.536) (all p < 0.001). The Mastora score showed similar significant associations with clinical severity indices, RV/LV ratio, and laboratory markers. ROC analysis demonstrated that PESI had the highest predictive accuracy for ICU admission (area under the curve, AUC = 0.849), followed by RV/LV ratio (AUC = 0.838), sPESI (AUC = 0.791), and both QI and Mastora score (AUC = 0.782). Platelet-to-lymphocyte ratio, systemic immune-inflammation index, and RV wall thickness were not significant predictors. Conclusions CT-derived obstruction indices and RV/LV ratio provide clinically meaningful prognostic information for ICU requirement in APE, particularly when integrated with established clinical severity scores.
Purpose:Osteosarcoma is the most common primary malignant bone tumor in children and young adults and demonstrates substantial histopathologic heterogeneity that complicates diagnosis and treatment monitoring. Conventional imaging and biopsy may not adequately reflect whole-tumor biology because of spatial sampling limitations. Multiparametric magnetic resonance imaging (MRI), including dynamic contrast-enhanced MRI (DCE-MRI) and diffusion-weighted imaging (DWI), provides quantitative biomarkers related to tumor vascularity and cellularity. The aim of this study was to evaluate the association between quantitative DCE-MRI parameters and apparent diffusion coefficient (ADC) values across osteosarcoma subtypes and to explore their role in reflecting histopathologic heterogeneity. Material and methods:This retrospective exploratory study included 43 patients with histopathologically confirmed osteosarcoma who underwent pre-treatment 3.0-T MRI, including DCE-MRI and DWI, between January 2023 and February 2025. The cohort consisted of 17 osteoblastic, 9 fibroblastic, 6 chondroblastic, 5 giant cell-rich, 5 telangiectatic, and 1 extraskeletal osteosarcoma cases. Quantitative pharmacokinetic parameters (Ktrans, Kep, and Ve) were derived using the Tofts model, while ADC values were obtained from diffusion-weighted sequences. Inter-subtype differences were evaluated using the Kruskal-Wallis test and Dunn-Bonferroni post hoc analyses. Receiver operating characteristic (ROC) analysis and multivariable logistic regression were performed to assess diagnostic performance. Results:Ktrans demonstrated the strongest discriminatory performance across osteosarcoma subtypes (H = 13.852, p = 0.017), with significantly higher values observed in chondroblastic compared with telangiectatic tumors. ROC analysis demonstrated good diagnostic performance of Ktrans in identifying osteoblastic (area under the curve [AUC] = 0.806, 95% confidence interval [CI]: 0.652-0.960, p = 0.021) and chondroblastic (AUC = 0.775, 95% CI: 0.611-0.939, p = 0.033) subtypes. ADC values provided complementary information regarding tumor cellularity but demonstrated less consistent discriminatory performance. Kep and Ve did not show statistically significant differences across subtypes. Conclusions:Quantitative DCE-MRI parameters, particularly Ktrans, demonstrated significant associations with osteosarcoma subtype heterogeneity and may reflect differences in tumor vascular permeability. Combined diffusion and perfusion imaging may provide complementary information for non-invasive characterization of osteosarcoma. However, these findings should be interpreted cautiously because of the exploratory design and limited subgroup sample sizes, and further validation in larger prospective multicenter studies is required.
Purpose:Image-guided transthoracic core needle biopsy (CNB) is a minimally invasive method widely used for diagnosing thoracic lesions. While its effectiveness and safety are well established in adults, data on its performance in pediatric populations are limited, necessitating further evaluation. This study aimed to assess the efficacy, diagnostic performance, and safety of image-guided transthoracic CNB in children with thoracic lesions. Material and methods:A retrospective study was conducted on 40 pediatric patients who underwent image-guided transthoracic CNB between May 2011 and January 2025. Biopsies were performed under ultrasound (US) or computed tomography (CT) guidance, depending on lesion location. Demographics, lesion characteristics, imaging modality, histopathology, and complications were analyzed. Results:Technical success was 100%. Diagnostic accuracy was 95%, with a sensitivity of 95.2%, specificity of 100%, PPV of 100%, and NPV of 94.7%. CT was used for central intraparenchymal lung lesions (n = 11); US was used for mediastinal (n = 8), pleural (n = 9), and peripheral lung lesions (n = 12). Pneumothorax occurred in 18.2% (2/11) of CT-guided cases and 6.9% (2/29) of US-guided cases. One chest tube was required (3.4%). Pulmonary hemorrhage occurred in one CT-guided case (9.1%). There was no statistically significant difference in complication rates between the CT-guided and US-guided biopsy groups (p = 0.300). Conclusions:Image-guided transthoracic CNB is a safe and reliable diagnostic technique for thoracic lesions in pediatric patients, offering high diagnostic accuracy with low complication rates.
