No abstract

No abstract

Case: A 33-year-old woman presented with a six-month history of spontaneous radial nerve palsy and no identified lesion on imaging. She underwent operative exploration where an hourglass deformity was seen and resected. Pathology returned as a rare tumor, a myofibroma. The patient regained full radial nerve function. Conclusions: A trial of observation is often indicated in the cases of isolated nerve palsy where anatomic lesions have been eliminated. This case highlights that imaging studies can miss a tumor involving nerve and that painless, spontaneous nerve palsy may be a time where early surgical intervention offers a better chance of recovery.

We present a case in which intraocular silicone injection for complex retinal detachment resulted in migration and distribution of silicone along the intracranial visual pathway, and ultimately throughout the ventricular system. Misinterpretation of this material as intracranial hemorrhage on outside computed tomography imaging delayed emergent repair of a Type A aortic dissection until the diagnosis was made on repeat imaging. A discussion of this case and salient computed tomography and magnetic resonance imaging characteristics of silicone is provided.

Deep learning (DL) has been proposed to automate image segmentation and provide accuracy, consistency, and efficiency. Accurate segmentation of lipomatous tumors (LTs) is critical for correct tumor radiomics analysis and localization. The major challenge of this task is data heterogeneity, including tumor morphological characteristics and multicenter scanning protocols. To mitigate the issue, we aimed to develop a DL-based Super Learner (SL) ensemble framework with different data correction and normalization methods. Pathologically proven LTs on pre-operative T1-weighted/proton-density MR images of 185 patients were manually segmented. The LTs were categorized by tumor locations as distal upper limb (DUL), distal lower limb (DLL), proximal upper limb (PUL), proximal lower limb (PLL), or Trunk (T) and grouped by 80%/9%/11% for training, validation and testing. Six configurations of correction/normalization were applied to data for fivefold-cross-validation trainings, resulting in 30 base learners (BLs). A SL was obtained from the BLs by optimizing SL weights. The performance was evaluated by dice-similarity-coefficient (DSC), sensitivity, specificity, and Hausdorff distance (HD95). For predictions of the BLs, the average DSC, sensitivity, and specificity from the testing data were 0.72 $$\pm$$ ± 0.16, 0.73 $$\pm$$ ± 0.168, and 0.99 $$\pm$$ ± 0.012, respectively, while for SL predictions were 0.80 $$\pm$$ ± 0.184, 0.78 $$\pm$$ ± 0.193, and 1.00 $$\pm$$ ± 0.010. The average HD95 of the BLs were 11.5 (DUL), 23.2 (DLL), 25.9 (PUL), 32.1 (PLL), and 47.9 (T) mm, whereas of SL were 1.7, 8.4, 15.9, 2.2, and 36.6 mm, respectively. The proposed method could improve the segmentation accuracy and mitigate the performance instability and data heterogeneity aiding the differential diagnosis of LTs in real clinical situations.