Radiomics Predicts Survival in Colorectal Liver Metastases Treated with TACE
Machine learning models using baseline imaging intensity and longitudinal changes stratify survival risk and identify lesion response.
Spectral CT Reconstructions Improve Acute Bowel Ischemia Detection
Adding iodine maps and monoenergetic images to portal venous CT increases diagnostic accuracy from 73% to 86% across scanner platforms.
Imaging Score Predicts Survival in Node-Negative Esophageal Cancer
A temporal lymph node stratification score identifies high-risk patients who may benefit from adjuvant therapy after neoadjuvant treatment.
Ultrasound-Derived Fat Fraction Accurately Grades Steatosis in Chronic Liver Disease
A multicenter study shows ultrasound-derived fat fraction provides high diagnostic accuracy across diverse etiologies and fibrosis stages.
CT Vascular Pruning Markers Identify Chronic Vasculopathy in Sickle Cell Disease
Automated quantification of small pulmonary vessels correlates with impaired gas exchange and history of pulmonary hypertension.
Deep Learning Synthesizes Contrast-Enhanced CT and MRI Scans
A scoping review of 56 studies shows high image fidelity but limited evidence for diagnostic interchangeability in clinical practice.
Photon-Counting CT Matches V/Q SPECT for CTEPH Perfusion Mapping
A retrospective study shows high quantitative concordance between photon-counting CT and scintigraphy for lobar lung perfusion assessment.
MR Neurography Adds Diagnostic Value in Proximal and Multi-Nerve Neuropathies
An analysis of 800 patients identifies lesion location and multi-nerve involvement as key indicators for magnetic resonance neurography.
![Splenic Uptake Variability Challenges Standardized [18F]PSMA-1007 PET Interpretation](/images/splenic-uptake-variability-challenges-standardized-April-23-2026.jpg)
Splenic Uptake Variability Challenges Standardized [18F]PSMA-1007 PET Interpretation
A multicenter study finds that 15% of patients show higher tracer uptake in the spleen than parotid glands, complicating lesion scoring.
Training-Free AI Framework Automates Radiographic Morphometry Across Multiple Joints
A generalist landmark matching model achieves measurement accuracy comparable to radiologists without requiring anatomy-specific datasets.
Most Validated Radiomics Models Lack Sufficient Training Data
A systematic review of high-impact journals finds 90% of machine learning models are undertrained, risking unreliable clinical predictions.
Contrast-Enhanced MRI Improves Mesorectal Nodal Staging Accuracy in Rectal Cancer
The Avocado Sign on T1-weighted imaging shows higher sensitivity and specificity than standard T2-weighted morphological criteria.