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/.

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.

Object: High-grade gliomas (HGGs) and metastasis (MET) are the most common intracranial lesions in neurosurgical routine. Both of them show an invasive growth pattern extending into neural tissue beyond the margins of contrast enhancement on MRI. These “undetected” areas might be the origin of early tumor recurrence. The aim of the present study was to evaluate whether 5-aminolevulinic acid (5-ALA) fluorescence provides an additional benefit in detection of invasive tumor compared with intraoperative MRI (iMRI). Methods: The authors prospectively enrolled 45 patients harboring contrast-enhancing lesions, in whom gross-total resection was intended. All patients had surgery in which iMRI and 5-ALA–guided resection were used following a specific protocol. First, a typical white light tumor resection was performed. Then, spatial location of residual fluorescence was marked. After that, an iMRI was performed and residual uptake of contrast was marked. Navigated biopsy samples were taken from all marked areas and from additional sites according to the surgeon's judgment. Cross tables and receiver operating characteristic curves were calculated, assessing performance of the imaging methods for tumor detection alone and for combined detection of infiltration zone and solid tumor (pathological tissue). Also, correlations of histopathological findings with imaging results were tested using Spearman rho. Results: Thirty-four patients with HGGs and 11 with METs were enrolled. Three patients harboring a MET showed no 5-ALA enhancement and were excluded; 127 histopathological samples were harvested in the remaining patients. In HGG, sensitivity for tumor detection was significantly higher (p < 0.001) in 5-ALA (0.85) than in iMRI (0.41). Specificity was significantly lower (p < 0.001) in 5-ALA (0.43) than in iMRI (0.70). For detection of pathological tissue, 5-ALA significantly exceeded iMRI in specificity (0.80 vs 0.60) and sensitivity (0.91 vs 0.66) (p < 0.001). Imaging results of iMRI and 5-ALA did not correlate significantly; only 5-ALA showed a significant correlation with final histopathological diagnosis of the specimen and with typical histopathological features of HGGs. In METs, sensitivity and specificity for tumor detection were equal in 5-ALA and iMRI. Both techniques showed high values for sensitivity (0.75) and specificity (0.80). The odds ratio for detection of tumor tissue was 12 for both techniques. Concerning pathological tissue, no statistically significant difference was found either. Imaging results of iMRI and 5-ALA correlated significantly (p < 0.022), as with final histopathological diagnosis in METs. Conclusions: In METs, due to the rate of nonenhancing lesions, the authors found no additional benefit of 5-ALA compared with iMRI. In HGG, imaging results of 5-ALA and iMRI are significantly different at the border zone; 5-ALA has a higher sensitivity and a lower specificity for tumor detection than Gd-DTPA–enhanced iMRI. For detection of infiltrating tumor at the border of the resection cavity, 5-ALA is superior to Gd-DTPA–enhanced iMRI concerning both sensitivity and specificity. Thus, use of 5-ALA in addition to iMRI might be beneficial to maximize extent of resection. Clinical synergistic effects will be evaluated in a prospective randomized trial.

The ideal treatment strategy for low-grade gliomas (LGGs) is a controversial topic. Additionally, only smaller single-center series dealing with the concept of intraoperative magnetic resonance imaging (iMRI) have been published. To investigate determinants for patient outcome and progression-free-survival (PFS) after iMRI-guided surgery for LGGs in a multicenter retrospective study initiated by the German Study Group for Intraoperative Magnetic Resonance Imaging. A retrospective consecutive assessment of patients treated for LGGs (World Health Organization grade II) with iMRI-guided resection at 6 neurosurgical centers was performed. Eloquent location, extent of resection, first-line adjuvant treatment, neurophysiological monitoring, awake brain surgery, intraoperative ultrasound, and field-strength of iMRI were analyzed, as well as progression-free survival (PFS), new permanent neurological deficits, and complications. Multivariate binary logistic and Cox regression models were calculated to evaluate determinants of PFS, gross total resection (GTR), and adjuvant treatment. A total of 288 patients met the inclusion criteria. On multivariate analysis, GTR significantly increased PFS (hazard ratio, 0.44; P < .01), whereas "failed" GTR did not differ significantly from intended subtotal-resection. Combined radiochemotherapy as adjuvant therapy was a negative prognostic factor (hazard ratio: 2.84, P < .01). Field strength of iMRI was not associated with PFS. In the binary logistic regression model, use of high-field iMRI (odds ratio: 0.51, P < .01) was positively and eloquent location (odds ratio: 1.99, P < .01) was negatively associated with GTR. GTR was not associated with increased rates of new permanent neurological deficits. GTR was an independent positive prognostic factor for PFS in LGG surgery. Patients with accidentally left tumor remnants showed a similar prognosis compared with patients harboring only partially resectable tumors. Use of high-field iMRI was significantly associated with GTR. However, the field strength of iMRI did not affect PFS. EoR, extent of resectionFLAIR, fluid-attenuated inversion recoveryGTR, gross total resectionIDH1, isocitrate dehydrogenase 1iMRI, intraoperative magnetic resonance imagingLGG, low-grade gliomaMGMT, methylguanine-deoxyribonucleic acid methyltransferasenPND, new permanent neurological deficitOS, overall survivalPFS, progression-free survivalSTR, subtotal resectionWHO, World Health Organization.

There is rising evidence that in glioblastoma(GBM) surgery an increase of extent of resection(EoR) leads to an increase of patient’s survival. Based on histopathological assessments tumor depiction of Gd-DTPA enhancement and 5-aminolevulinic-acid-fluorescence(5-ALA) might be synergistic for intraoperative resection control. To assess impact of additional use of 5-ALA in intraoperative MRI(iMRI) assisted surgery of GBMs on extent of resection(EoR), progression free survival(PFS) and overall survival(OS). We prospectively enrolled 33 patients with GBMs eligible for gross-total-resection(GTR) and performed a combined approach using 5-ALA and iMRI. As a control group, we performed a retrospective matched pair assessment, based on 144 patients with iMRI-assisted surgery. Matching criteria were, MGMT promotor methylation, recurrent surgery, eloquent location, tumor size and age. Only patients with an intended GTR and primary GBMs were included. We calculated Kaplan Mayer estimates to compare OS and PFS using the Log-Rank-Test. We used the T-test to compare volumetric results of EoR and the Chi-Square-Test to compare new permanent neurological deficits(nPND) and general complications between the two groups. Median follow up was 31 months. No significant differences between both groups were found concerning the matching criteria. GTR was achieved significantly more often (p <0.010) using 5-ALA&iMRI (100%) compared to iMRI alone(82%). Mean EoR was significantly(p<0.004) higher in 5-ALA&iMRI-group(99.7%) than in iMRI-alone-group(97.4%) Rate of complications did not differ significantly between groups(21% iMRI-group,27%5-ALA&iMRI-group,p<0.518). nPND were found in 6% in both groups. Median PFS (6mo resp.;p<0.309) and median OS(iMRI:17mo;5-ALA&iMRI-group:18mo;p<0.708)) were not significantly different between both groups. We found a significant increase of EoR when combining 5-ALA&iMRI compared to use of iMRI alone. Maximizing EoR did not lead to an increase of complications or neurological deficits if used with neurophysiological monitoring in eloquent lesions. No final conclusion can be drawn whether a further increase of EoR benefits patient’s progression free survival and overall survival.

PURPOSE: Prospective data suggested a superiority of intraoperative MRI (iMRI) over 5-aminolevulinic acid (5-ALA) for achieving complete resections of contrast enhancement in glioblastoma surgery. We investigated this hypothesis in a prospective clinical trial and correlated residual disease volumes with clinical outcome in newly diagnosed glioblastoma. METHODS: This is a prospective controlled multicenter parallel-group trial with two center-specific treatment arms (5-ALA and iMRI) and blinded evaluation. The primary end point was complete resection of contrast enhancement on early postoperative MRI. We assessed resectability and extent of resection by an independent blinded centralized review of preoperative and postoperative MRI with 1-mm slices. Secondary end points included progression-free survival (PFS) and overall survival (OS), patient-reported quality of life, and clinical parameters. RESULTS: We recruited 314 patients with newly diagnosed glioblastomas at 11 German centers. A total of 127 patients in the 5-ALA and 150 in the iMRI arm were analyzed in the as-treated analysis. Complete resections, defined as a residual tumor ≤0.175 cm³, were achieved in 90 patients (78%) in the 5-ALA and 115 (81%) in the iMRI arm ( P = .79). Incision-suture times ( P < .001) were significantly longer in the iMRI arm (316 v 215 [5-ALA] minutes). Median PFS and OS were comparable in both arms. The lack of any residual contrast enhancing tumor (0 cm³) was a significant favorable prognostic factor for PFS ( P < .001) and OS ( P = .048), especially in methylguanine-DNA-methyltransferase unmethylated tumors ( P = .006). CONCLUSION: We could not confirm superiority of iMRI over 5-ALA for achieving complete resections. Neurosurgical interventions in newly diagnosed glioblastoma shall aim for safe complete resections with 0 cm³ contrast-enhancing residual disease, as any other residual tumor volume is a negative predictor for PFS and OS.