Classification of high-grade glioblastoma and single brain metastases using a new SCAT-inception model trained with MRI images.

Background and objectives: Glioblastoma (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 methods: For 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. Results: By 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. Conclusion: This 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.

Prediction of intraoperative cerebrospinal fluid leaks in endoscopic endonasal transsphenoidal pituitary surgery based on a deep neural network model trained with MRI images: a pilot study.

Objective: This 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). Methods: 396 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. Results: A 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. Conclusion: In 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.

The transfrontal isthmus approach for insular glioma surgery.

OBJECTIVE: The classic transopercular or transsylvian approach to insular gliomas removes the tumor laterally through the insular cortex. This study describes a new anteroposterior approach through the frontal isthmus for insular glioma surgery. METHODS: The authors detailed the surgical techniques for resection of insular gliomas through the transfrontal isthmus approach. Fifty-nine insular gliomas with at least Berger-Sanai zone I involvement were removed with the new approach, and extent of resection and postoperative neurological outcomes were assessed. RESULTS: Fifty-nine patients were enrolled in the study, including 35 men and 24 women, with a mean (range) age 44.3 (19-75) years. According to the Berger-Sanai classification system, the most common tumor was a giant glioma (67.8%), followed by involvement of zones I and IV (18.6%). Twenty-two cases were Yasargil type 3A/B, and 37 cases were Yasargil type 5A/B. The average angle between the lateral plane of the putamen and sagittal line was 33.53°, and the average width of the isthmus near the anterior insular point was 33.33 mm. The average angle between the lateral plane of the putamen and the sagittal line was positively correlated with the width of the isthmus near the anterior insular point (r = 0.935, p < 0.0001). The median (interquartile range [IQR]) preoperative tumor volume was 67.82 (57.64-92.19) cm3. Of 39 low-grade gliomas, 26 (66.67%) were totally resected; of 20 high-grade gliomas, 19 (95%) were totally resected. The median (IQR) extent of resection of the whole group was 100% (73.7%-100%). Intraoperative diffusion-weighted imaging showed no cases of middle cerebral artery- or lenticulostriate artery-related stroke. Extent of insular tumor resection was positively correlated with the angle of the lateral plane of the putamen and sagittal line (r = -0.329, p = 0.011) and the width of the isthmus near the anterior insular point (r = -0.267, p = 0.041). At 3 months postoperatively, muscle strength grade exceeded 4 in all cases, and all patients exhibited essentially normal speech. The median (IQR) Karnofsky performance score at 3 months after surgery was 90 (80-90). CONCLUSIONS: The transfrontal isthmus approach changes the working angle from lateral-medial to anterior-posterior, allowing for maximal safe removal of insular gliomas.

Resection of Insular Glioma Through the Transfrontal Limiting Sulcus Approach.

BACKGROUND: The current transsylvian or transopercular approaches make access difficult because of the limited exposure of insular tumors. Hence, maximal and safe removal of insular gliomas is challenging. In this article, a new approach to resect insular gliomas is presented. OBJECTIVE: To determine whether the new transfrontal limiting sulcus approach is helpful for maximal and safe removal of insular gliomas. METHODS: The authors reported surgical techniques for insular gliomas resected through the transfrontal limiting sulcus approach. The authors evaluated the surgical resections of 69 insular gliomas performed through the new approach in their department. The extents of resection and postoperative neurological outcomes were analyzed to determine the value of this new approach. RESULTS: Based on the Berger-Sanai classification, most insular gliomas were giant tumors (59.42%), followed by zone I + IV tumors (24.64%). The median (interquartile range) extent of resection of all patients was 100% (91%, 100%). The total resection rate for all gliomas was (55 of 69, 79.7%), and the total resection rate for low-grade gliomas was (28 of 40, 70%), which was significantly lower than that for high-grade gliomas (27 of 29, 93.1%) (P = .019). All patients had muscle strength greater than grade 4 3 months after surgery. Only 1 patient had a speech disorder 3 months after surgery. The median Karnofsky Performance Status score at the time of the 3-month follow-up was 90. CONCLUSION: The transfrontal limiting sulcus approach can help to achieve maximal and safe removal of insular gliomas.

Deep Learning model-based approach for preoperative prediction of Ki67 labeling index status in a noninvasive way using magnetic resonance images: A single-center study.

OBJECTIVES: Ki67 is an important biomarker of pituitary adenoma (PA) aggressiveness. In this study, PA invasion of surrounding structures is investigated and deep learning (DL) models are established for preoperative prediction of Ki67 labeling index (Ki67LI) status using conventional magnetic resonance (MR) images. METHODS: We reviewed 362 consecutive patients with PAs who underwent endoscopic transsphenoidal surgery, of which 246 patients with primary PA are selected for PA invasion analysis. MRI data from 234 of these PA patients are collected to develop DL models to predict Ki67LI status, and DL models were tested on 27 PA patients in the clinical setting. RESULTS: PA invasion is observed in 46.8% of cases in the Ki67 ≥ 3% group and 33.3% of cases in the Ki67 < 3% group. Three deep-learning models are developed using contrast-enhanced T1-weighted images (ceT1WI), T2-weighted images (T2WI), and multimodal images (ceT1WI+T2WI), respectively. On the validation dataset, the prediction accuracy of the ceT1WI model, T2WI model, and multimodal model were 87.4%, 89.4%, and 89.2%, respectively. In the clinical test, 27 MR slices with the largest tumors from 27 PA patients were tested using the ceT1WI model, T2WI model, and multimodal model, the average accuracy of Ki67LI status prediction was 63%, 77.8%, and 70.4%, respectively. CONCLUSION: Preoperative prediction of PA Ki67LI status in a noninvasive way was realized with the DL model by using MRI. T2WI model outperformed the ceT1WI model and multimodal model. This end-to-end model-based approach only requires a single slice of T2WI to predict Ki67LI status and provides a new tool to help clinicians make better PA treatment decisions.

Real-time augmented reality application in presurgical planning and lesion scalp localization by a smartphone.

OBJECTIVE: A smartphone augmented reality (AR) application (app) was explored for clinical use in presurgical planning and lesion scalp localization. METHODS: We programmed an AR App on a smartphone. The accuracy of the AR app was tested on a 3D-printed head model, using the Euclidean distance of displacement of virtual objects. For clinical validation, 14 patients with brain tumors were included in the study. Preoperative MRI images were used to generate 3D models for AR contents. The 3D models were then transferred to the smartphone AR app. Tumor scalp localization was marked, and a surgical corridor was planned on the patient's head by viewing AR images on the smartphone screen. Standard neuronavigation was applied to evaluate the accuracy of the smartphone. Max-margin distance (MMD) and area overlap ratio (AOR) were measured to quantitatively validate the clinical accuracy of the smartphone AR technique. RESULTS: In model validation, the total mean Euclidean distance of virtual object displacement using the smartphone AR app was 4.7 ± 2.3 mm. In clinical validation, the mean duration of AR app usage was 168.5 ± 73.9 s. The total mean MMD was 6.7 ± 3.7 mm, and total mean AOR was 79%. CONCLUSIONS: The smartphone AR app provides a new way of experience to observe intracranial anatomy in situ, and it makes surgical planning more intuitive and efficient. Localization accuracy is satisfactory with lesions larger than 15 mm.

A Supplemental Technique for Preoperative Evaluation of Giant Intracranial Aneurysm.

BACKGROUND: Preoperative planning mainly relies on digital subtraction angiography (DSA) and computed tomography angiography. However, neither technique can reveal thrombi in giant intracranial aneurysms (GIAs). In this study, we aimed to reconstruct the circulating and noncirculating parts of GIAs with the time-of-flight (TOF) and motion-sensitized driven-equilibrium (MSDE) sequences with 3D Slicer to reveal an integrated presentation of GIAs, compare its accuracy, and validate the usefulness for preoperative planning. MATERIAL AND METHODS: Patients with GIAs who were treated with microsurgery in our department were included in this study. Both the TOF and MSDE sequence data for each patient were loaded into 3D Slicer for reconstruction and segmentation. The parameters measured by 3D Slicer were compared with those measured by DSA. RESULTS: The mean diameter for all GIAs was 28.7 ± 1.5 mm (range, 25.9-31.9 mm). The mean diameter for all GIAs measured by DSA and 3D Slicer was 24.46 ± 5.25 and 28.66 ± 1.48 mm, respectively (t = 4.948, p < 0.01). When only the nonthrombotic GIAs were included, the mean diameter measured by DSA and 3D Slicer was 28.69 ± 2.03 and 28.97 ± 1.79 mm, respectively (t = 1.023, p = 0.323). The mean aneurysmal volume was 8,292.6 ± 1,175.1 mm3 and the mean thrombotic volume was 3,590.0 ± 1,003.7 mm3. CONCLUSION: The MSDE sequence brings diagnostic benefits as a comparison to other MRI sequences. Reconstruction of GIAs with 3D Slicer is a low-cost, dependable, and useful supplemental technique for surgical planning.