ABSTRACT
PURPOSE: Evaluation of a deep learning approach for the detection of meniscal tears and their characterization (presence/absence of migrated meniscal fragment). METHODS: A large annotated adult knee MRI database was built combining medical expertise of radiologists and data scientists' tools. Coronal and sagittal proton density fat suppressed-weighted images of 11,353 knee MRI examinations (10,401 individual patients) paired with their standardized structured reports were retrospectively collected. After database curation, deep learning models were trained and validated on a subset of 8058 examinations. Algorithm performance was evaluated on a test set of 299 examinations reviewed by 5 musculoskeletal specialists and compared to general radiologists' reports. External validation was performed using the publicly available MRNet database. Receiver Operating Characteristic (ROC) curves results and Area Under the Curve (AUC) values were obtained on internal and external databases. RESULTS: A combined architecture of meniscal localization and lesion classification 3D convolutional neural networks reached AUC values of 0.93 (95% CI 0.82, 0.95) for medial and 0.84 (95% CI 0.78, 0.89) for lateral meniscal tear detection, and 0.91 (95% CI 0.87, 0.94) for medial and 0.95 (95% CI 0.92, 0.97) for lateral meniscal tear migration detection. External validation of the combined medial and lateral meniscal tear detection models resulted in an AUC of 0.83 (95% CI 0.75, 0.90) without further training and 0.89 (95% CI 0.82, 0.95) with fine tuning. CONCLUSION: Our deep learning algorithm demonstrated high performance in knee menisci lesion detection and characterization, validated on an external database.
Subject(s)
Deep Learning , Tibial Meniscus Injuries , Adult , Humans , Magnetic Resonance Imaging , Menisci, Tibial/diagnostic imaging , Retrospective Studies , Tibial Meniscus Injuries/diagnostic imagingSubject(s)
Algorithms , Artifacts , Computed Tomography Angiography/methods , Aged, 80 and over , Humans , Male , MetalsABSTRACT
PURPOSE: To evaluate the performance and limitations of the signal intensity ratio method for quantifying liver iron overload at 3 T. METHODS: Institutional review board approval and written informed consent from all participants were obtained. One hundred and five patients were included prospectively. All patients underwent a liver biopsy with biochemical assessment of hepatic iron concentration and a 3 T MRI scan with 5 breath-hold single-echo gradient-echo sequences. Linear correlation between liver-to-muscle signal intensity ratio and liver iron concentration was calculated. The algorithm for calculating magnetic resonance hepatic iron concentration was adapted from the method described by Gandon et al. with echo times divided by 2. Sensitivity and specificity were calculated. RESULTS: Five patients were excluded (coil selection failure or missing sequence) and 100 patients were analyzed, 64 men and 36 women, 52 ± 13.3 years old, with a biochemical hepatic iron concentration range of 0-630 µmol/g. Linear correlation between biochemical hepatic iron concentration and MR-hepatic iron concentration was excellent with a correlation coefficient = 0.96, p < 0.0001. Sensitivity and specificity were, respectively, 83% (0.70-0.92) and 96% (0.85-0.99), with a pathological threshold of 36 µmol/g. CONCLUSION: Signal intensity ratio method for quantifying liver iron overload can be used at 3 T with echo times divided by 2.
Subject(s)
Iron Overload/diagnostic imaging , Liver Diseases/diagnostic imaging , Magnetic Resonance Imaging/methods , Algorithms , Biopsy , Female , Humans , Male , Middle Aged , Sensitivity and SpecificityABSTRACT
Hepatic steatosis is a common condition, the prevalence of which is increasing along with non-alcoholic hepatic steatosis. In imaging, it can present in a typical homogeneous or heterogeneous way. Some forms create traps in imaging, whether localised steatosis is concerned or areas which have been spared by steatosis, and the purpose of this paper is to explain and illustrate them. The role of different imaging methods is described while emphasizing the importance of MRI.
Subject(s)
Diagnostic Imaging , Fatty Liver/diagnosis , Aged , Diagnostic Imaging/methods , Humans , Magnetic Resonance Imaging , MaleABSTRACT
IPMN is a frequent disease involving pancreatic duct. This disease could be malignant (parenchymal invasive adenocarcinoma), particularly if the main pancreatic duct is involved (this involvement is considered present if > 6 mm), if this enlargement reaches 10 mm or more, and if the pathological phenotype is biliopancreatic or intestinal (malignancy is less frequent if gastric one). Invasiveness is suspected if hypodense parenchymal lesion is present, particularly near a cystical lesion or MPD, a mural nodule of the wall, or if MPD wall has got a contrast uptake. Mural nodules inside cystic branch duct are associated with in situ grade 3 malignancies. MPD IPMN must be resected to prevent malignancy. The follow-up of isolated branch duct cysts relies upon MDCT and MRI, every two years if lesion is less than 1cm. Every one year if bigger, particularly if more than to 3 cm.