Magnetic resonance imaging (MRI) has become a crucial tool for evaluating mediastinal masses considering that several lesions that appear indeterminate on computed tomography and radiography can be differentiated on MRI. Using a three-compartment model to localize the mass and employing a basic knowledge of MRI, radiologists can easily diagnose mediastinal masses. Here, we review the use of MRI in evaluating mediastinal masses and present the images of various mediastinal masses categorized using the International Thymic Malignancy Interest Group's three-compartment classification system. These masses include thymic hyperplasia, thymic cyst, pericardial cyst, thymoma, mediastinal hemangioma, lymphoma, mature teratoma, bronchogenic cyst, esophageal duplication cyst, mediastinal thyroid carcinoma originating from ectopic thyroid tissue, mediastinal liposarcoma, mediastinal pancreatic pseudocyst, neurogenic tumor, meningocele, and plasmacytoma.

Conventional deep learning (DL) algorithm requires full supervision of annotating the region of interest (ROI) that is laborious and often biased. We aimed to develop a weakly-supervised DL algorithm that diagnosis breast cancer at ultrasound without image annotation. Weakly-supervised DL algorithms were implemented with three networks (VGG16, ResNet34, and GoogLeNet) and trained using 1000 unannotated US images (500 benign and 500 malignant masses). Two sets of 200 images (100 benign and 100 malignant masses) were used for internal and external validation sets. For comparison with fully-supervised algorithms, ROI annotation was performed manually and automatically. Diagnostic performances were calculated as the area under the receiver operating characteristic curve (AUC). Using the class activation map, we determined how accurately the weakly-supervised DL algorithms localized the breast masses. For internal validation sets, the weakly-supervised DL algorithms achieved excellent diagnostic performances, with AUC values of 0.92–0.96, which were not statistically different (all Ps > 0.05) from those of fully-supervised DL algorithms with either manual or automated ROI annotation (AUC, 0.92–0.96). For external validation sets, the weakly-supervised DL algorithms achieved AUC values of 0.86–0.90, which were not statistically different (Ps > 0.05) or higher (P = 0.04, VGG16 with automated ROI annotation) from those of fully-supervised DL algorithms (AUC, 0.84–0.92). In internal and external validation sets, weakly-supervised algorithms could localize 100% of malignant masses, except for ResNet34 (98%). The weakly-supervised DL algorithms developed in the present study were feasible for US diagnosis of breast cancer with well-performing localization and differential diagnosis.

Response assessments based on interim C-MET PET/CT could predict the therapeutic outcome of PCNSL.

No abstract

Chordoid glioma is a rare low grade tumor typically located in the third ventricle. Although a chordoid glioma can arise from ventricle with tumor cells having features of ependymal differentiation, intraventricular dissemination has not been reported. Here we report a case of a patient with third ventricular chordoid glioma and intraventricular dissemination in the lateral and fourth ventricles. We described the perfusion MR imaging features of our case different from a previous report.