Explainable hybrid vision transformers and convolutional network for multimodal glioma segmentation in brain MRI.

Accurate localization of gliomas, the most common malignant primary brain cancer, and its different sub-region from multimodal magnetic resonance imaging (MRI) volumes are highly important for interventional procedures. Recently, deep learning models have been applied widely to assist automatic lesion segmentation tasks for neurosurgical interventions. However, these models are often complex and represented as "black box" models which limit their applicability in clinical practice. This article introduces new hybrid vision Transformers and convolutional neural networks for accurate and robust glioma segmentation in Brain MRI scans. Our proposed method, TransXAI, provides surgeon-understandable heatmaps to make the neural networks transparent. TransXAI employs a post-hoc explanation technique that provides visual interpretation after the brain tumor localization is made without any network architecture modifications or accuracy tradeoffs. Our experimental findings showed that TransXAI achieves competitive performance in extracting both local and global contexts in addition to generating explainable saliency maps to help understand the prediction of the deep network. Further, visualization maps are obtained to realize the flow of information in the internal layers of the encoder-decoder network and understand the contribution of MRI modalities in the final prediction. The explainability process could provide medical professionals with additional information about the tumor segmentation results and therefore aid in understanding how the deep learning model is capable of processing MRI data successfully. Thus, it enables the physicians' trust in such deep learning systems towards applying them clinically. To facilitate TransXAI model development and results reproducibility, we will share the source code and the pre-trained models after acceptance at https://github.com/razeineldin/TransXAI .

Explainability of deep neural networks for MRI analysis of brain tumors.

PURPOSE: Artificial intelligence (AI), in particular deep neural networks, has achieved remarkable results for medical image analysis in several applications. Yet the lack of explainability of deep neural models is considered the principal restriction before applying these methods in clinical practice. METHODS: In this study, we propose a NeuroXAI framework for explainable AI of deep learning networks to increase the trust of medical experts. NeuroXAI implements seven state-of-the-art explanation methods providing visualization maps to help make deep learning models transparent. RESULTS: NeuroXAI has been applied to two applications of the most widely investigated problems in brain imaging analysis, i.e., image classification and segmentation using magnetic resonance (MR) modality. Visual attention maps of multiple XAI methods have been generated and compared for both applications. Another experiment demonstrated that NeuroXAI can provide information flow visualization on internal layers of a segmentation CNN. CONCLUSION: Due to its open architecture, ease of implementation, and scalability to new XAI methods, NeuroXAI could be utilized to assist radiologists and medical professionals in the detection and diagnosis of brain tumors in the clinical routine of cancer patients. The code of NeuroXAI is publicly accessible at https://github.com/razeineldin/NeuroXAI .

DeepSeg: deep neural network framework for automatic brain tumor segmentation using magnetic resonance FLAIR images.

PURPOSE: Gliomas are the most common and aggressive type of brain tumors due to their infiltrative nature and rapid progression. The process of distinguishing tumor boundaries from healthy cells is still a challenging task in the clinical routine. Fluid-attenuated inversion recovery (FLAIR) MRI modality can provide the physician with information about tumor infiltration. Therefore, this paper proposes a new generic deep learning architecture, namely DeepSeg, for fully automated detection and segmentation of the brain lesion using FLAIR MRI data. METHODS: The developed DeepSeg is a modular decoupling framework. It consists of two connected core parts based on an encoding and decoding relationship. The encoder part is a convolutional neural network (CNN) responsible for spatial information extraction. The resulting semantic map is inserted into the decoder part to get the full-resolution probability map. Based on modified U-Net architecture, different CNN models such as residual neural network (ResNet), dense convolutional network (DenseNet), and NASNet have been utilized in this study. RESULTS: The proposed deep learning architectures have been successfully tested and evaluated on-line based on MRI datasets of brain tumor segmentation (BraTS 2019) challenge, including s336 cases as training data and 125 cases for validation data. The dice and Hausdorff distance scores of obtained segmentation results are about 0.81 to 0.84 and 9.8 to 19.7 correspondingly. CONCLUSION: This study showed successful feasibility and comparative performance of applying different deep learning models in a new DeepSeg framework for automated brain tumor segmentation in FLAIR MR images. The proposed DeepSeg is open source and freely available at https://github.com/razeineldin/DeepSeg/.

Sodium Fluorescein as Intraoperative Visualization Tool During Peripheral Nerve Biopsies.

OBJECTIVE: Owing to technical development of specific fluorophore filters, the neurosurgical application of sodium fluorescein (SF) has regained value in brain tumor surgery. The aim of this study was to determine the usefulness of SF during nerve biopsies. METHODS: This single-center study included 5 cases of nerve biopsies performed under microscope-based fluorescence with SF performed between March 2016 and February 2017. SF was applied intravenously (1 mg/kg body weight). After microsurgical dissection of the involved nerve segment, fluorescence-guided fascicular biopsy was performed. Selection of target fascicles was at the surgeon's discretion and took into account nerve stimulation for preservation of motor function and fluorescence intensity. Correlation to histopathologic examination was examined. Video analysis of intraoperative images comparing target fascicles with intense fluorescent response to adjacent fascicles of the same nerve segment was performed using ImageJ. RESULTS: All patients had motor or sensory deficits. Magnetic resonance imaging findings were similar, depicting long segments of gadolinium enhancement (minimum 11.7 cm). Each biopsy sample was positive resulting in diverse histopathologic results. Digital image analysis revealed a statistically significant difference of the complementary color green (P = 0.0473). CONCLUSIONS: Magnetic resonance imaging is the gold standard in diagnostic work-up of peripheral nerve disorders. Longitudinal nerve thickening with positive contrast enhancement is an unspecific magnetic resonance imaging finding. Various pathologies, such as tumors and inflammatory lesions, may cause this morphologic phenomenon. Nerve biopsies may be needed for diagnostic work-up. Intraoperative SF may help to depict the most affected fascicles and identify target fascicles for biopsy and increase diagnostic certainty of nerve biopsies.

3D Cellular Architecture Modulates Tyrosine Kinase Activity, Thereby Switching CD95-Mediated Apoptosis to Survival.

The death receptor CD95 is expressed in every cancer cell, thus providing a promising tool to target cancer. Activation of CD95 can, however, lead to apoptosis or proliferation. Yet the molecular determinants of CD95's mode of action remain unclear. Here, we identify an optimal distance between CD95Ligand molecules that enables specific clustering of receptor-ligand pairs, leading to efficient CD95 activation. Surprisingly, efficient CD95 activation leads to apoptosis in cancer cells in vitro and increased tumor growth in vivo. We show that allowing a 3D aggregation of cancer cells in vitro switches the apoptotic response to proliferation. Indeed, we demonstrate that the absence or presence of cell-cell contacts dictates the cell response to CD95. Cell contacts increase global levels of phosphorylated tyrosines, including CD95's tyrosine. A tyrosine-to-alanine CD95 mutant blocks proliferation in cells in contact. Our study sheds light into the regulatory mechanism of CD95 activation that can be further explored for anti-cancer therapies.

Diagnostic accuracy of intraoperative perfusion-weighted MRI and 5-aminolevulinic acid in relation to contrast-enhanced intraoperative MRI and 11C-methionine positron emission tomography in resection of glioblastoma: a prospective study.

The aim of our study was to compare depicted pre-, intra-, and postoperative tumor volume of met-PET, perfusion-weighed MRI (PWI), and Gd-DTPA MRI. Further, to assess their sensitivity and specificity in correlation with histopathological specimen. Inclusion criteria of the prospective study were histological confirmed glioblastoma (GB), age > 18, and eligible for gross total resection (GTR). Met-PET was performed before and after surgery. Gd-DTPA MRI and PWI were performed before, during, and after surgery. A combined 5-aminolevulinic acid (5-ALA) and iMRI-guided surgery was performed. Volumetric analysis was evaluated for all imaging modalities except for 5-ALA. A total of 59 navigated biopsies were taken. Sensitivity and specificity were calculated for Gd-DTPA MRI, PWI, met-PET, and 5-ALA according to the histology of specimen. Met-PET depicted significantly larger tumor volume before surgery (p = 0.01) compared to PWI and Gd-DTPI MRI. We found no significant difference in tumor volume between met-PET and PWI after surgery (p = 0.059). Both PWI and met-PET showed significantly larger tumor volume after surgery when compared to Gd-DTPA (p = 0.018 and p = 0.003, respectively). Intraoperative PWI reading was impaired in 33.3% due to artifacts. Met-PET showed the highest sensitivity for detection of GB with 95%. The lowest sensitivity was found with Gd-DTPA MRI (50%), while 5-ALA and intraoperative PWI showed similar results (69 and 67%). Met-Pet is the imaging modality with the highest sensitivity to detect a residual tumor in GB. Intraoperative PWI seems to have a synergistic effect to Gd-DTPA and 5-ALA. However, its value may be limited by artifacts. Both pre- and intraoperative PWI cannot substitute met-PET in tumor detection.

Comparison of hemispheric dominance and correlation of evoked speech responses between functional magnetic resonance imaging and navigated transcranial magnetic stimulation in language mapping.

BACKGROUND: Non-invasive language mapping is commonly facilitated via two different approaches, functional magnetic resonance imaging (fMRI) and navigated transcranial magnetic stimulation (nTMS). The aim of our examination was to compare both methods regarding hemispheric dominance and detection of involved cortex in speech-processing. METHODS: Since 2015, patients with language eloquent tumors received both fMRI and nTMS language mapping in our facility. Hemispheric dominance for fMRI (word pairing task) and nTMS (picture naming) was defined according to activation/response predominance between hemispheres. For a detailed comparison of methods, we used a cortical parcellation system and correlated activations of each gyrus. Statistical analysis was performed using Sign test, Spearman's rank correlation coefficient and Mann-Whitney-U test. RESULTS: Twenty-three patients with predominantly frontal and temporal tumors were enrolled. fMRI and nTMS predicted the same hemispheric dominance in 67% of examinations. No significant difference was found between the tests (P=0.63). When examining the correlation of anatomical areas with speech relevance between the two methods, we found significant differences in activation patterns of language relevant areas when comparing patients inter-individually. Using 2/3 rule for evoked language errors in nTMS improved congruence in perisylvian regions. CONCLUSIONS: fMRI and nTMS lead to comparable results in determining hemispheric dominance, however they show significantly different results in predicting language relevant areas. We recommend using 2/3 rule for nTMS to improve congruence. Both tools seem to be valuable methods for preoperative planning. However, surgical decision making concerning resectability of language eloquent lesions should be based on awake craniotomy, furthermore.

HOXA7, HOXA9, and HOXA10 are differentially expressed in clival and sacral chordomas.

Chordomas are rare tumours of the bone arising along the spine from clivus to sacrum. We compared three chordoma cell lines of the clivus region including the newly established clivus chordoma cell line, U-CH14, with nine chordoma cell lines originating from sacral primaries by morphology, on genomic and expression levels and with patient samples from our chordoma tissue bank. Clinically, chordomas of the clivus were generally smaller in size at presentation and patients with sacral chordomas had more metastases and more often recurrent disease. All chordoma cell lines had a typical physaliphorous morphology and expressed brachyury, S100-protein and cytokeratin. By expression analyses we detected differentially expressed genes in the clivus derived cell lines as compared to the sacral cell lines. Among these were HOXA7, HOXA9, and HOXA10 known to be important for the development of the anterior-posterior body axis. These results were confirmed by qPCR. Immunohistologically, clivus chordomas had no or very low levels of HOXA10 protein while sacral chordomas showed a strong nuclear positivity in all samples analysed. This differential expression of HOX genes in chordomas of the clivus and sacrum suggests an oncofetal mechanism in gene regulation linked to the anatomic site.

Activation of Cytotoxic Natural Killer Cells After Aneurysmal Subarachnoid Hemorrhage.

OBJECTIVE: Cell-mediated inflammation is critical in the development of cerebrovascular complications after aneurysmal subarachnoid hemorrhage. We analyzed the course for activated CD16brightCD56dim cytotoxic natural killer (NK) cells in cerebrospinal fluid of 15 patients. METHODS: Patients were classified by occurrence of cerebral vasospasm (CV) and delayed cerebral ischemia. NK were monitored by flow cytometry between day 1 and 14 after hemorrhage. RESULTS: Twelve patients (80%) developed CV with a mean day of detection at 3.9 ± 1.6. In those patients, cell count for NK increased from 1.40 ± 1.42 cells/µL on day 1 to a peak of 11.66 ± 11.56 cells/µL on day 6.1 ± 2.9 (P = 0.001). An increase of mean cerebral blood flow velocity in transcranial Doppler from 71.33 ± 12.93 cm/second to 166.20 ± 20.19 cm/second (P < 0.01) and an increase in number of vascular axes affected by CV was detected (P < 0.01). In patients with grade 3 CV (n = 4, 33.3%), activated NK counts were significantly higher than in patients who did not have CV (23.18 ± 13.92 cells/µL vs. 0.02 ± 0.01 cells/µL; P = 0.029). NK counts were significantly different between patients with grade 1 and grade 3 CV (P = 0.04). Patients who did not have CV who showed low NK counts achieved better functional outcome (Glasgow Outcome Scale [GOS] score, 4.6 ± 0.6) at discharge than did patients with CV grade 2 (GOS score, 3.3 ± 0.5) and CV grade 3 (GOS score, 2.3 ± 0.5) who showed increased NK cell counts (CV grade 0 vs. CV grade 2, P = 0.048; CV grade 0 vs. CV grade 3, P = 0.001). Activated CD16brightCD56dim cytotoxic NKCSF cell counts showed a mean maximum (14.15 ± 12.21 cells/µL) when delayed cerebral ischemia occurred. CONCLUSIONS: The increase of activated CD16brightCD56dim cytotoxic NK cells in cerebrospinal fluid after aneurysmal subarachnoid hemorrhage suggests an increased risk of CV and delayed cerebral ischemia.

Impact of extent of resection and recurrent surgery on clinical outcome and overall survival in a consecutive series of 170 patients for glioblastoma in intraoperative high field magnetic resonance imaging.

BACKGROUND: In patients with a glioblastoma (GBM), few unselected data exists using actual standard adjuvant treatment and contemporary surgical techniques like iMRI. Aim of study is to assess impact of EoR and recurrent surgery on survival and outcome. METHODS: We assessed a consecutive unselected series of 170 surgeries for GBM (2008-2014) applying intraoperative MRI (iMRI). All patients received adjuvant radio-chemo-therapy. Overall-survival (OS), progression free survival (PFS), complications and new permanent neurological deficits (nPND) were assessed. Uni- and multivariate-cox-regression-models were calculated. RESULTS: Mean follow-up was 40mo. GTR was intended in 82% and achieved in 77% of these cases. A nPND was found in 7% of patients. In multivariate cox-regression, GTR (HR:0.6, P<0.024) and absence of MGMT methylation (HR:1.6, P<0.042) was significantly associated with PFS. We found no difference in PFS after primary surgery and recurrent surgery. Concerning OS, in multivariate assessment an un-methylated MGMT-promotor (HR2.0, P<0.01) and presence of a complication (HR1.7, P<0.06) were negative prognosticators. Only GTR was significantly beneficial for OS (HR0.4, P<0.028) compared to a failed GTR and a STR. Repeated surgery for recurrent disease was positively associated with OS (HR0.6, P<0.06). CONCLUSIONS: Surgery in a contemporary setup using iMRI, brain mapping and modern adjuvant treatment, has a higher OS and lower complication rates as previously published. A maximum but safe resection should be the goal of surgery since a perioperative complication significantly decreases OS. Recurrent surgery has a beneficial effect on OS without an increase of complications.

Sensitivity and specificity of linear array intraoperative ultrasound in glioblastoma surgery: a comparative study with high field intraoperative MRI and conventional sector array ultrasound.

INTRODUCTION: Linear array intraoperative ultrasound (lioUS) is an emerging technology for intracranial use. We evaluated sensitivity and specificity of lioUS to detect residual tumor in patients harboring a glioblastoma. METHODS: After near total resection in 20 patients, residual tumor detection using lioUS, conventional intraoperative ultrasound (cioUS), and gadopentetic-diethylenetriamine penta-acetic acid (Gd-DTPA)-enhanced intraoperative MRI (iMRI) were compared. Sensitivity and specificity were calculated based on 68 navigated biopsies. Receiver operator characteristic (ROC) curves and correlation with histopathological findings of each imaging modality were calculated. Additionally, results were evaluated in the subgroup of recurrent disease (23 biopsies in 8 patients). RESULTS: Sensitivity of lioUS (76 %) was significantly higher compared with iMRI (55 %) and cioUS (24 %). Specificity of lioUS (58 %) was significantly lower than in cioUS (96 %), while there was no significant difference to iMRI (74 %). All imaging modalities correlated significantly with histopathological findings. In the subgroup of recurrent disease, sensitivity and specificity decreased in all modalities. However, cioUS showed significant lower values than iMRI and lioUS. In ROC curves, lioUS showed a higher area und the curve (AUC) in comparison with iMRI and cioUS. We found similar results in the subgroup of recurrent disease. CONCLUSION: Tumor detection using a lioUS is significantly superior to cioUS. Overall test performance in lioUS is comparable with results of iMRI. While, the latter has a higher specificity and a significantly lower sensitivity in comparison with lioUS.

Linear array ultrasound in low-grade glioma surgery: histology-based assessment of accuracy in comparison to conventional intraoperative ultrasound and intraoperative MRI.

INTRODUCTION: In low-grade glioma (LGG) surgery, intraoperative differentiation between tumor and most likely tumor-free brain tissue can be challenging. Intraoperative ultrasound can facilitate tumor resection. The aim of this study is to evaluate the accuracy of linear array ultrasound in comparison to conventional intraoperative ultrasound (cioUS) and intraoperative high-field MRI (iMRI). METHODS: We prospectively enrolled 13 patients harboring a LGG of WHO Grade II. After assumed near total removal, a resection control was performed using navigated cioUS, navigated lioUS, and iMRI. We harvested 30 navigated biopsies from the resection cavity and compared the histopathological findings with the respective imaging results. Spearman's rho was calculated to test for significant correlations. Sensitivity and specificity as well as receiver operating characteristics (ROC) were calculated to assess test performance of each imaging modality. RESULTS: Imaging results of lioUS correlated significantly (p < 0.009) with iMRI. Both iMRI and lioUS correlated significantly with final histopathological diagnosis (p < 0.006, p < 0.014). cioUS did not correlate with other imaging findings or with final diagnosis. The highest sensitivity for residual tumor detection was found in iMRI (83 %), followed by lioUS (79 %). The sensitivity of cioUS was only 21 %. Specificity was highest in cioUS (100 %), whereas iMRI and lioUS both achieved 67 %. ROC curves showed fair results for iMRI and lioUS and a poor result for cioUS. CONCLUSIONS: Intraoperative resection control in LGGs using lioUS reaches a degree of accuracy close to iMRI. Test results of lioUS are superior to cioUS. cioUS often fails to discriminate solid tumors from "normal" brain tissue during resection control. Only in lesions <10 cc cioUS does show good accuracy.

Inhibition of NF-κB signaling ablates the invasive phenotype of glioblastoma.

UNLABELLED: Glioblastoma multiforme, the most common primary brain tumor, is highly refractory to therapy, mainly due to its ability to form micrometastases, which are small clusters or individual cells that rapidly transverse the brain and make full surgical resection impossible. Here, it is demonstrated that the invasive phenotype of glioblastoma multiforme is orchestrated by the transcription factor NF-κB which, via metalloproteinases (MMP), regulates fibronectin processing. Both, cell lines and tumor stem cells from primary glioblastoma multiforme, secrete high levels of fibronectin which when cleaved by MMPs forms an extracellular substrate. Subsequently, forming and interacting with their own microenvironment, glioblastoma multiforme cells are licensed to invade their surroundings. Mechanistic study revealed that NF-κB inhibition, either genetically or pharmacologically, by treatment with Disulfiram, significantly abolished the invasive phenotype in the chick chorioallantoic membrane assay. Furthermore, having delineated the underlying molecular mechanism of glioblastoma multiforme invasion, the potential of a disulfiram-based therapy was revealed in a highly invasive orthotrophic glioblastoma multiforme mouse model. IMPLICATIONS: This study defines a novel therapeutic approach that inhibits micrometastases invasion and reverts lethal glioblastoma into a less aggressive disease.

Combined inhibition of HER1/EGFR and RAC1 results in a synergistic antiproliferative effect on established and primary cultured human glioblastoma cells.

Glioblastoma is the most frequent brain tumor of glial origin in adults. With the best available standard-of-care, patients with this disease have a life expectancy of only approximately 15 months after diagnosis. Because the EGF receptor (HER1/EGFR) is one of the most commonly dysregulated oncogenes in glioblastoma, HER1/EGFR-targeted agents, such as erlotinib, were expected to provide a therapeutic benefit. However, their application in the clinical setting failed. Seeking an explanation for this finding, we previously identified several candidate genes for resistance of human glioblastoma cell lines toward erlotinib. On the basis of this panel of genes, we aimed at identifying drugs that synergistically enhance the antiproliferative effect of erlotinib on established and primary glioblastoma cell lines. We found that NSC23766, an inhibitor of RAC1, enhanced the antineoplastic effects of erlotinib in U87MG, T98MG, and A172MG glioblastoma cell lines for the most part in a synergistic or at least in an additive manner. In addition, the synergistic antiproliferative effect of erlotinib and NSC23766 was confirmed in primary cultured cells, indicating a common underlying cellular and molecular mechanism in glioblastoma. Therefore, agents that suppress RAC1 activation may be useful therapeutic partners for erlotinib in a combined targeted treatment of glioblastoma.

A conceptually new treatment approach for relapsed glioblastoma: coordinated undermining of survival paths with nine repurposed drugs (CUSP9) by the International Initiative for Accelerated Improvement of Glioblastoma Care.

To improve prognosis in recurrent glioblastoma we developed a treatment protocol based on a combination of drugs not traditionally thought of as cytotoxic chemotherapy agents but that have a robust history of being well-tolerated and are already marketed and used for other non-cancer indications. Focus was on adding drugs which met these criteria: a) were pharmacologically well characterized, b) had low likelihood of adding to patient side effect burden, c) had evidence for interfering with a recognized, well-characterized growth promoting element of glioblastoma, and d) were coordinated, as an ensemble had reasonable likelihood of concerted activity against key biological features of glioblastoma growth. We found nine drugs meeting these criteria and propose adding them to continuous low dose temozolomide, a currently accepted treatment for relapsed glioblastoma, in patients with recurrent disease after primary treatment with the Stupp Protocol. The nine adjuvant drug regimen, Coordinated Undermining of Survival Paths, CUSP9, then are aprepitant, artesunate, auranofin, captopril, copper gluconate, disulfiram, ketoconazole, nelfinavir, sertraline, to be added to continuous low dose temozolomide. We discuss each drug in turn and the specific rationale for use- how each drug is expected to retard glioblastoma growth and undermine glioblastoma's compensatory mechanisms engaged during temozolomide treatment. The risks of pharmacological interactions and why we believe this drug mix will increase both quality of life and overall survival are reviewed.

The rationale of targeting neutrophils with dapsone during glioblastoma treatment.

Data from past research is presented showing that neutrophils are active participants in new vessel formation in normal physiology, in proliferating human endometrium, in non-cancer pathologies as in the pannus of rheumatoid arthritis, and in various cancers, among them glioblastoma. These data show that interleukin-8 (IL-8) is a major chemokine attracting neutrophil infiltrates in these states. Since the old anti-Hansen's disease drug dapsone inhibits neutrophil migration along an IL-8 gradient towards increasing concentrations, and is used therapeutically for this attribute to good effect in dermatitis herpetiformis, bullous pemphigoid and rheumatoid arthritis, we suggest dapsone may deprive glioblastoma of neutrophil-mediated growth promoting effects. We review past research showing that vascular endothelial growth factor, VEGF, is carried predominantly intracellularly within neutrophils--only 2% of circulating VEGF is found free in serum. Based on the available evidence summarized by the authors, dapsone has a strong theoretical potential to become a useful anti-VEGF, anti-angiogenic agent in glioblastoma treatment.

Drug combinations enhancing the antineoplastic effects of erlotinib in high-grade glioma.

Glioblastoma multiforme (GBM) is the most common malignant brain tumor in adults. Especially in this disease, qualitative and quantitative aspects render the dysregulated epidermal growth factor receptor (HER1/EGFR) an outstanding therapeutic target. A variety of therapeutic compounds was developed to target HER1/EGFR among which the clinically most advanced agents are small molecule tyrosine kinase (TK) inhibitors. Unfortunately, clinical studies examining their therapeutic efficacy have so far failed to document a major therapeutic break-through in the setting of GBM. Thus, the targeted approach against HER1/EGFR likely requires a synergistic drug combination strategy to ultimately become successful in this disease. This patents review focuses on innovative therapeutic strategies combining HER1/EGFR-targeted TK inhibitors with novel agents which for the most part have not been evaluated for the treatment of GBM yet but which constitute interesting candidates for further evaluation in this setting.

Intraoperative high-resolution ultrasound: a new technique in the management of peripheral nerve disorders.

OBJECT: Surgical treatment of nerve lesions in continuity remains difficult, even in the most experienced hands. The regenerative potential of those injuries can be evaluated by intraoperative electrophysiological studies and/or intraneural dissection. The present study examines the value of intraoperative high-frequency ultrasound as an imaging tool for decision making in the management of traumatic nerve lesions in continuity. METHODS: Intraoperative high-frequency ultrasound was applied to 19 traumatic or iatrogenic nerve lesions of differing extents. The information obtained was correlated with intraoperative electrophysiological, microsurgical intraneural dissection, and histopathological findings in resected nerve segments. RESULTS: The intraoperative application of high-resolution, high-frequency ultrasound enabled morphological examination of nerve lesions in continuity, with good image quality. The assessment of the severity of the underlying nerve injury matched perfectly with the judgment obtained from intraoperative electrophysiological studies. Both intraneural nerve dissection and neuropathological examination of the resected nerve segments confirmed the sonographic findings. In addition, intraoperative ultrasound proved to be very time efficient. CONCLUSIONS: With intraoperative ultrasound, the extent of traumatic peripheral nerve lesions can be examined morphologically for the first time. It is a promising, noninvasive method that seems capable of assessing the type (intraneural/perineural) and grade of nerve fibrosis. Therefore, in combination with intraoperative neurophysiological studies, intraoperative high-resolution ultrasound may represent a major tool for noninvasive assessment of the regenerative potential of a nerve lesion.

Pre- and intraoperative processing and integration of various anatomical and functional data in neurosurgery.

A software system is presented, capable of integrating various information sources for neurosurgical procedures. These include anatomical data such as a standard 3D DICOM image stacks, atlas data, as well as functional information (e.g. fmri, MEG, EEG). The system is programmed in C++ using Open GL for visualisation, and was developed in a close cooperation with a neurosurgical department to match existing needs. Preoperatively the data may be combined, registered by a rigid or elastic matching process, enriched by user specified planning information such as annotations or trajectories, and visualised in a standard fashion using different segmentation schemes, interactive rotation, zooming etc. Selected portions of the gathered and generated information may then be exported for neuronavigation input in DICOM or a vendor specific format. Intraoperatively, this information may, on the one hand, be simply used as an integrational part of the routinely used navigational data. On the other hand, the system is also capable of interacting with the navigational system to integrate the actual spatial information of the ongoing procedure into the preoperative data, thus allowing further planning and visualisation beyond the scope of the navigational unit. Furthermore, intraoperatively updated information such as intraoperative MR images or electrophysiological data may be integrated and correlated to the existing information. Ongoing developments comprise redistribution of the relevant data for injection onto the screen of the navigational system or into the optical pathway of the (3D capable) microscope display/ocular. This also includes information about the actual automatically optimized accuracy of the navigation process in relation to the head markers in use.

A deformable digital brain atlas system according to Talairach and Tournoux.

Brain atlases are valuable tools which assist neurosurgeons during the planning of an intervention. Since a printed atlas book has several disadvantages-among them the difficulty to map the information onto a patient's individual anatomy-we have developed a digital version of the well-established stereotaxic brain atlas of Talairach and Tournoux. Our atlas system is mainly dedicated to assist neurosurgical planning, and its benefits are: (i) a three-dimensional (3D) representation of most brain structures contained in the Talairach atlas; (ii) a nonrigid matching capability which warps the standard atlas anatomy to an individual brain magnetic resonance imaging (MRI) dataset in a few minutes and which is able to take deformations due to tumors into account; (iii) the integration of several sources of neuroanatomical knowledge; (iv) an interface to a navigation system which allows utilization of atlas information intraoperatively. In this paper we outline the algorithm we have developed to achieve 3D surface models of the brain structures. Moreover, we describe the nonrigid matching method which consists of two tasks: firstly, point correspondences between the atlas and the patient are established in an automatic fashion, and secondly these displacement vectors are interpolated using a radial basis function approach to form a continuous transformation function. To generate appropriate target structures for the first of these tasks, we implemented a quick segmentation tool which is capable to segment the cortex and ventricles in less than 5 min. An evaluation shows that our nonrigid approach is more precise than the conventional piecewise linear matching, though it should be further improved for the region around the deep grey nuclei. Summarizing, we developed a Win32 program which permits the convenient and fast application of standardized anatomy to individual brains which potentially contain tumors.