Background: Pathologic diagnosis of central nervous system (CNS) neoplasms is made by comparing light microscopic, immunohistochemical, and molecular cytogenetic findings with clinicoradiologic observations. Intraoperative frozen cytology smears can improve the diagnostic accuracy for CNS neoplasms. Here, we evaluate the diagnostic value of cytology in frozen diagnoses of CNS neoplasms. Methods: Cases were selected from patients undergoing both frozen cytology and frozen sections. Diagnostic accuracy was evaluated. Results: Four hundred and fifty-four cases were included in this retrospective single-center review study covering a span of 10 years. Five discrepant cases (1.1%) were found after excluding 53 deferred cases (31 cases of tentative diagnosis, 22 cases of inadequate frozen sampling). A total of 346 cases of complete concordance and 50 cases of partial concordance were classified as not discordant cases in the present study. Diagnostic accuracy of intraoperative frozen diagnosis was 87.2%, and the accuracy was 98.8% after excluding deferred cases. Discrepancies between frozen and permanent diagnoses (n = 5, 1.1%) were found in cases of nonrepresentative sampling (n = 2) and misinterpretation (n = 3). High concordance was observed more frequently in meningeal tumors (97/98, 99%), metastatic brain tumors (51/52, 98.1%), pituitary adenomas (86/89, 96.6%), schwannomas (45/47, 95.8%), high-grade astrocytic tumors (47/58, 81%), low grade astrocytic tumors (10/13, 76.9%), non-neoplastic lesions (23/36, 63.9%), in decreasing frequency. Conclusions: Using intraoperative cytology and frozen sections of CNS tumors is a highly accurate diagnostic ancillary method, providing subtyping of CNS neoplasms, especially in frequently encountered entities.
1 January 2023 in Journal of Korean Neurosurgical Society
Objective : Deep learning is a machine learning approach based on artificial neural network training, and object detection algorithm using deep learning is used as the most powerful tool in image analysis. We analyzed and evaluated the diagnostic performance of a deep learning algorithm to identify skull fractures in plain radiographic images and investigated its clinical applicability.Methods : A total of 2026 plain radiographic images of the skull (fracture, 991; normal, 1035) were obtained from 741 patients. The RetinaNet architecture was used as a deep learning model. Precision, recall, and average precision were measured to evaluate the deep learning algorithm’s diagnostic performance.Results : In ResNet-152, the average precision for intersection over union (IOU) 0.1, 0.3, and 0.5, were 0.7240, 0.6698, and 0.3687, respectively. When the intersection over union (IOU) and confidence threshold were 0.1, the precision was 0.7292, and the recall was 0.7650. When the IOU threshold was 0.1, and the confidence threshold was 0.6, the true and false rates were 82.9% and 17.1%, respectively. There were significant differences in the true/false and false-positive/false-negative ratios between the anteriorposterior, towne, and both lateral views (p=0.032 and p=0.003). Objects detected in false positives had vascular grooves and suture lines. In false negatives, the detection performance of the diastatic fractures, fractures crossing the suture line, and fractures around the vascular grooves and orbit was poor.Conclusion : The object detection algorithm applied with deep learning is expected to be a valuable tool in diagnosing skull fractures.
Objective : Traumatic brain injury (TBI) is one of the most common injuries in patients with multiple trauma, and it associates with high post-traumatic mortality and morbidity. A trauma center was established to provide optimal treatment for patients with severe trauma. This study aimed to compare the treatment outcomes of patients with severe TBI between non-trauma and trauma centers based on data from the Korean Neuro-Trauma Data Bank System (KNTDBS).Methods : From January 2018 to June 2021, 1122 patients were enrolled in the KNTDBS study. Among them, 253 patients from non-traumatic centers and 253 from trauma centers were matched using propensity score analysis. We evaluated baseline characteristics, the time required from injury to hospital arrival, surgery-related factors, neuromonitoring, and outcomes.Results : The time from injury to hospital arrival was shorter in the non-trauma centers (110.2 vs. 176.1 minutes, p=0.012). The operation time was shorter in the trauma centers (156.7 vs. 128.1 minutes, p0.003). Neuromonitoring was performed in nine patients (3.6%) in the non-trauma centers and 67 patients (26.5%) in the trauma centers (p<0.001). Mortality rates were lower in trauma centers than in non-trauma centers (58.5% vs. 47.0%, p=0.014). The average Glasgow coma scale (GCS) at discharge was higher in the trauma centers (4.3 vs. 5.7, p=0.011). For the Glasgow outcome scale-extended (GOSE) at discharge, the favorable outcome (GOSE 5–8) was 17.4% in the non-trauma centers and 27.3% in the trauma centers (p=0.014).Conclusion : This study showed lower mortality rates, higher GCS scores at discharge, and higher rates of favorable outcomes in trauma centers than in non-trauma centers. The regional trauma medical system seems to have a positive impact in treating patients with severe TBI.