Search

Background and objectivesGlioblastoma (GBM) and brain metastasis (MET) are the two most common intracranial tumors. However, the different pathogenesis of the two tumors leads to completely different treatment options. In terms of magnetic resonance imaging (MRI), GBM and MET are extremely similar, which makes differentiation by imaging extremely challenging. Therefore, this study explores an improved deep learning algorithm to assist in the differentiation of GBM and MET.Materials and methodsFor this study, axial contrast-enhanced T1 weight (ceT1W) MRI images from 321 cases of high-grade gliomas and solitary brain metastasis were collected. Among these, 251 out of 270 cases were selected for the experimental dataset (127 glioblastomas and 124 metastases), 207 cases were chosen as the training dataset, and 44 cases as the testing dataset. We designed a new deep learning algorithm called SCAT-inception (Spatial Convolutional Attention inception) and used five-fold cross-validation to verify the results.ResultsBy employing the newly designed SCAT-inception model to predict glioblastomas and brain metastasis, the prediction accuracy reached 92.3%, and the sensitivity and specificity reached 93.5 and 91.1%, respectively. On the external testing dataset, our model achieved an accuracy of 91.5%, which surpasses other model performances such as VGG, UNet, and GoogLeNet.ConclusionThis study demonstrated that the SCAT-inception architecture could extract more subtle features from ceT1W images, provide state-of-the-art performance in the differentiation of GBM and MET, and surpass most existing approaches.

ObjectiveThis study aimed to investigate the reliability of a deep neural network (DNN) model trained only on contrast-enhanced T1 (T1CE) images for predicting intraoperative cerebrospinal fluid (ioCSF) leaks in endoscopic transsphenoidal surgery (EETS).Methods396 pituitary adenoma (PA) cases were reviewed, only primary PAs with Hardy suprasellar Stages A, B, and C were included in this study. The T1CE images of these patients were collected, and sagittal and coronal T1CE slices were selected for training the DNN model. The model performance was evaluated and tested, and its interpretability was explored.ResultsA total of 102 PA cases were enrolled in this study, 51 from the ioCSF leakage group, and 51 from the non-ioCSF leakage group. 306 sagittal and 306 coronal T1CE slices were collected as the original dataset, and data augmentation was applied before model training and testing. In the test dataset, the DNN model provided a single-slice prediction accuracy of 97.29%, a sensitivity of 98.25%, and a specificity of 96.35%. In clinical test, the accuracy of the DNN model in predicting ioCSF leaks in patients reached 84.6%. The feature maps of the model were visualized and the regions of interest for prediction were the tumor roof and suprasellar region.ConclusionIn this study, the DNN model could predict ioCSF leaks based on preoperative T1CE images, especially in PAs in Hardy Stages A, B, and C. The region of interest in the model prediction-making process is similar to that of humans. DNN models trained with preoperative MRI images may provide a novel tool for predicting ioCSF leak risk for PA patients.

Objective To compare the efficacy of augmented reality assisted neuroendoscope of evacuation of hematoma and small bone window craniotomy in the treatment of hypertensive cerebral hemorrhage. Methods From July 2016 to December 2017, 82 patients with hypertensive cerebral hemorrhage were treated with small bone window craniotomy (41 cases) or augmented reality assisted neuroendoscopic surgery (41 cases) of evacuation of hematoma. The operation time, intraoperative blood loss, hematomas clearance rate and hospitalization time, as well as the incidence of postoperative rebleeding, intracranial infection, pulmonary infection, gastrointestinal bleeding and other complications, and Glasgow Outcome Scale (GOS) were recorded to evaluate prognosis of the 2 groups. Results Compared with the small bone window craniotomy group, in the neuroendoscopic group, the operation time [(72.66 ± 5.21) min vs. (137.51 ± 6.25) min; t = -51.065, P = 0.000], the hospitalization time [(12.21 ± 1.81) d vs. (13.12 ± 1.41) d; t = -4.035, P = 0.035] shortened, intraoperative blood loss decreased [(91.88 ± 8.89) ml vs. (400.24 ± 36.57) ml; t = -52.464, P = 0.000], hematoma clearance rate increased [(98.29 ± 1.37)% vs. (72.83 ± 9.37)%; t = -17.224, P = 0.002], and prognosis was better at 6 months after operation (Z = -3.303, P = 0.001). The complication rate was no significant difference between the 2 groups [12.20% (5/41) vs.26.83% (11/41); χ2 = 2.795, P = 0.095]. Conclusions Augmented reality assisted neuroendoscope of evacuation of hematoma is superior than small bone window craniotomy. DOI:10.3969/j.issn.1672-6731.2019.09.009

Objective To assess the safety and efficacy of microsurgical resection of cerebral arteriovenous malformation (AVM). Methods A total of 65 patients with cerebral AVMs were treated with microsurgical resections from April to August 2010 in our hospital. Of the 65 patients, 26 were male and 39 were female with age ranging from 4 to 72 years (average 42 years). Initial symptoms included cerebral hemorrhage in 32 cases, seizures in 10 cases, headache in 6 cases, neurological dysfunction in 14 cases, and the left 3 cases were diagnosed in health examination. According to Spetzler-Martin grades, there were 7 cases in Grade Ⅰ, 20 in Grade Ⅱ, 23 in Grade Ⅲ, 10 in Grade Ⅳ and 5 in Grade Ⅴ. The diameter of nidi ranged from 2 to 7 cm (on average of 3.50 cm). The lesions were located in the surface of cerebral cortex (31 cases) and deep brain (34 cases), including frontal lobe (19 cases), temporal lobe (12 cases), parietal lobe (16 cases), occipital lobe (9 cases), cerebellum (3 cases), basal ganglia (2 cases) and lateral cleft (4 cases). Feeding arteries included anterior cerebral artery (ACA) and its branches in 11 cases, middle cerebral artery (MCA) in 29 cases, posterior cerebral artery (PCA) in 15 cases, both ACA and MCA in 2 cases, both ACA and PCA in 5 cases, and vertebral artery (VA) in 3 cases. There were superficial vein drainage in 29 cases, deep vein drainage in 15 cases, and both superficial and deep vein drainage in 21 cases. All of those patients underwent surgical resections, including simple excisions in 45 cases and excisions combined with evacuation of hematoma in 20 cases. Among all of those operations, 6 were performed through single-frontal approach, 3 orbitozygomatic approach, 2 transcortical approach, 8 frontal-parietal approach, 10 pterional approach, 3 subtemporal approach, 4 temporal-parietal approach, 7 single-parietal approach, 7 interhemispheric approach, 8 parietal-occipital approach, 3 suboccipital approach, and 4 transcortical-transventricular approach. Results Among these patients, 4 cases (6.15%) had undergone prior endovascular embolization. Total resection was obtained in 60 cases (92.31%) and 5 cases (7.69%) took postoperative stereotactic radiosurgery. On discharge, modified Rankin Scale (mRS) scores were 0 in 21 cases (32.31%), 1 in 23 cases (35.38%), 2 in 9 cases (13.85%), 3 in 7 cases (10.77%), 4 in 3 cases (4.62% ) and 5 in 2 cases (3.08% ). During the follow-up (12 months on average), 6 patients got good recovery, while 3 patients presented with limb weakness, one visual field defect, 2 aphasia, one underwent ventriculoperitoneal shunt (VPS) due to hydrocephalus, 2 newly occurred seizures, and 2 deaths (one died of respiratory failure after one year, and the other was agnogenic). Conclusions Microsurgical resection of AVMs is highly efficient and can be undertaken with low rates of mortality. Adequate patient selection and careful preoperative planning are essential to improve the prognosis of patients with complicated AVM lesions. DOI: 10.3969/j.issn.1672-6731.2015.04.008

Objective To discuss the therapeutic effect of carotid endarterectomy (CEA) for patients with carotid atherosclerotic stenosis. Methods A total of 65 patients with carotid atherosclerotic stenosis were enrolled. There were 35 cases of left stenosis (53.85%) and 30 cases of right stenosis (46.15%). According to North American Symptomatic Carotid Endarterectomy Trial (NASCET), there were 17 cases of moderate stenosis (26.15%) and 48 cases of severe stenosis (73.85%). All of them underwent CEA, and the postoperative complications and prognosis were observed and assessed. Results CEA was successfully performed in 65 patients with a mean operation time of 90 min and a mean carotid blocking time of 30 min. There was no use of carotid patch and no intraoperative shunting. After CEA, the clinical symptoms of all patients were improved. Postoperative complications were as follows: one patient presented muscle weakness and numbness of contralateral extremities, and recovered after one-week symptomatic treatment; cranial nerve injury occurred in 2 cases: one recurrent laryngeal nerve injury and the other glossopharyngeal nerve injury, however, all injuries were transient and were cured 6 months after the operation; one case presented restenosis of ipsilateral carotid without clinical symptom and thus received conservative treatment; one case suffered from acute myocardial infarction 4 months after the operation, and recovered after symptomatic treatment; one case died without clear cause of death 9 months after the operation. None of these patients occurred with ischemic stroke. Conclusions Full preoperative preparation, close monitoring and meticulous manipulation during operation, frequent postoperative observation and treatment can ensure the treatment effect of CEA. DOI: 10.3969/j.issn.1672-6731.2015.03.009