An 80 year old lady with a history of metastatic sigmoid carcinoma presented with expressive dysphasia and unsteady gait 4 days after commencement of adjuvant capecitabine chemotherapy. MRI demonstrated restricted diffusion and T2/FLAIR hyperintensity involving the course of the bilateral corticospinal tracts, the corpus callosum and the middle cerebellar peduncles. Discontinuation of chemotherapy lead to symptom resolution in 2 days; repeat MRI at 2 months demonstrated reversal of the diffusion changes and improvement of the previous T2W/FLAIR hyperintensity. This report describes the first case of capecitabine induced leukoencephalopathy causing restricted diffusion along the corticospinal tracts, which should be differentiated from other entities that involve the corticospinal tracts (i.e. amyotrophic lateral sclerosis (ALS), primary lateral sclerosis (PLS), hypoglycemic coma, etc.)

Introduction: 3D-printed temporal bone models enable the training and rehearsal of complex otological procedures. To date, there has been no consolidation of the literature regarding the developmental process of 3D-printed temporal bone models. A brief review of the current literature shows that many of the key surgical landmarks of the temporal bone are poorly represented in models. This study aims to propose a novel design and production workflow to produce high-fidelity 3D-printed temporal bone models for surgical simulation. Methods: Developmental phases for data extraction, 3D segmentation and Computer Aided Design (CAD), and fabrication are outlined. The design and fabrication considerations for key anatomical regions, such as the mastoid air cells and course of the facial nerve, are expounded on with the associated strategy and design methods employed. To validate the model, radiological measurements were compared and a senior otolaryngologist performed various surgical procedures on the model. Results: Measurements between the original scans and scans of the model demonstrate sub-millimetre accuracy of the model. Assessment by the senior otologist found that the model was satisfactory in simulating multiple surgical procedures. Conclusion: This study offers a systematic method for creating accurate 3D-printed temporal bone models for surgical training. Results show high accuracy and effectiveness in simulating surgical procedures, promising improved training and patient outcomes.

Introduction: In this study, we aimed to compare the performance of a convolutional neural network (CNN)-based deep learning model that was trained on a dataset of normal and abnormal paediatric elbow radiographs with that of paediatric emergency department (ED) physicians on a binomial classification task. Methods: A total of 1,314 paediatric elbow lateral radiographs (patient mean age 8.2 years) were retrospectively retrieved and classified based on annotation as normal or abnormal (with pathology). They were then randomly partitioned to a development set (993 images); first and second tuning (validation) sets (109 and 100 images, respectively); and a test set (112 images). An artificial intelligence (AI) model was trained on the development set using the EfficientNet B1 network architecture. Its performance on the test set was compared to that of five physicians (inter-rater agreement: fair). Performance of the AI model and the physician group was tested using McNemar test. Results: The accuracy of the AI model on the test set was 80.4% (95% confidence interval [CI] 71.8%–87.3%), and the area under the receiver operating characteristic curve (AUROC) was 0.872 (95% CI 0.831–0.947). The performance of the AI model vs. the physician group on the test set was: sensitivity 79.0% (95% CI: 68.4%–89.5%) vs. 64.9% (95% CI: 52.5%–77.3%; P = 0.088); and specificity 81.8% (95% CI: 71.6%–92.0%) vs. 87.3% (95% CI: 78.5%–96.1%; P = 0.439). Conclusion: The AI model showed good AUROC values and higher sensitivity, with the P-value at nominal significance when compared to the clinician group.

Background: Pediatric craniocervical junction (CCJ) anomalies consist of a unique subset of anatomically complex spine conditions. The aims of intervention are to achieve long-term stability, correct existing deformity, and prevent neurological compromise. However, surgery is challenging due to critical neurovascular and musculoskeletal structures in the limited operative space of a young child. Recently, the use of three-dimensional (3D) printed models has been demonstrated to be valuable neurosurgical adjuncts. We therein report the application of a 3D-printed model for a pediatric case with a complex CCJ condition. A systematic review of the related literature is concurrently performed. Case description: A 10-year-old male presented with torticollis associated with neck pain and progressive thoracic kyphosis. Neuroimaging reported an unfused os odontoideum inferior to the basion and anterior half of the C2 vertebral body and anteriorly angulated with the C1 anterior arch. Of note, there was a large vertebral vein coursing over the left C2 lamina that was predominantly draining into the CCJ venous plexus. A radiologically derived 3D model of the patient’s CCJ was printed and used for pre-operative planning, multi-disciplinary team discussion, and detailed counseling with the patient and caregivers. The patient underwent an uneventful C1–C2 posterior screw fixation and has recovered well since. Separately, we observed there is a paucity of publications specific to this topic. Conclusions: As demonstrated, a custom-made 3D model was useful for clinicians work through technical difficulties and improve the perioperative discussion process in an otherwise difficult case.

Background and Objective: To compare the performance of an AI model developed based on strategies designed to accommodate small sized development sets to pediatric ER physicians on a binomial classification task of pediatric elbow radiographs.Materials and Methods: 1,314 pediatric elbow lateral radiographs (mean age: 8.2 years) were retrospectively retrieved, binomially classified based on annotation as normal or abnormal (with pathology), and randomly partitioned into a development set (993 images), tuning set (109 images), second tuning set (100 images) and test set (112 images). An AI model was trained on the development set utilizing the EfficientNet B1 compound scaling network architecture and augmentations. Its performance on the test set was compared to a group of five physicians (inter-rater agreement: fair). Statistical analysis: AUC of AI model - DeLong method. Performance of AI model and physician groups - McNemar test.Results: Accuracy of the model on the test set - 0.804 (95% CI, 0.718 - 0.873), Area under Receiver Operating Characteristic (AUROC) - 0.872 (95% CI, 0.831 - 0.947). AI model performance compared to the physician group on the test set - sensitivity 0.790 (95% CI 0.684 to 0.895) vs 0.649 (95% CI 0.525 to 0.773), p value 0.088; specificity 0.818 (95% CI 0.716 to 0.920) vs 0.873 (95% CI 0.785 to 0.961), p value 0.439.Conclusions: An AI model for elbow radiograph triage designed with strategies to optimize performance for a small sized development set performed comparably to physicians.

BackgroundTo compare the performance of an AI model based on strategies designed to overcome small sized development sets to pediatric ER physicians at a classification triage task of pediatric elbow radiographs. Methods1,314 pediatric elbow lateral radiographs (mean age: 8.2 years) were retrospectively retrieved, binomially classified based on their annotation as normal or abnormal (with pathology), and randomly partitioned into a development set (993 images), tuning set (109 images), second tuning set (100 images) and test set (112 images). The AI model was trained on the development set and utilized the EfficientNet B1 compound scaling network architecture and online augmentations. Its performance on the test set was compared to a group of five physicians (inter-rater agreement: fair). Statistical analysis: AUC of AI model - DeLong method. Performance of AI model and physician groups - McNemar test. ResultsAccuracy of the model on the test set - 0.804 (95% CI, 0.718 - 0.873), AUROC - 0.872 (95% CI, 0.831 - 0.947). AI model performance compared to the physician group on the test set - sensitivity 0.790 (95% CI 0.684 to 0.895) vs 0.649 (95% CI 0.525 to 0.773), p value 0.088; specificity 0.818 (95% CI 0.716 to 0.920) vs 0.873 (95% CI 0.785 to 0.961), p value 0.439.ConclusionsThe AI model for elbow radiograph triage designed with strategies to optimize performance for a small sized development set showed comparable performance to physicians.