NIMG-101. COMPARISON OF PRE-SURGICAL MRI FINDINGS FROM LONG-TERM (≥ 3 YRS) VERSUS SHORT-TERM (< 3 YRS) GLIOBLASTOMA SURVIVORS: A BLINDED, CASE-CONTROL REVIEW

s vii188 NEURO-ONCOLOGY • November 00 line disease was often supratentorial (420, 74%), multifocal (274, 49%), and bilateral (224, 40%). Fewer had leptomeningeal disease (94/430 CSF+, 22%; 117/559 radiographically positive, 21%). Most received methotrexate (MTX)-based induction (534, 96%), most commonly rituximab, MTX, procarbazine, vincristine (R-MVP; 257, 46%) or MVP (144, 26%). Overall, 286 (51%) patients relapsed with a median PFS of 2.1 years (95% CI 1.8-2.5). Of 275 characterizable relapses, 75 (27%) were exclusively local intraparenchymal, 91 (33%) were distant intraparenchymal, and 109 (40%) were other. Base-line unifocal disease often relapsed unifocally (66/118 baseline unifocal; 56%). Of 126 unifocal relapses, 45 (36%) were local. Of 202 patients with base-line supratentorial disease, 152 (75%) relapsed supratentorially. Unilateral relapses often remain ipsilateral (49% left-and 54% right-hemispheric lesions at baseline and relapse). CONCLUSIONS: In a large PCNSL cohort, relapses often maintained baseline characteristics. For a disease historically considered multifocal, a substantial proportion of relapses were local. Future analyses will evaluate predictors of relapse, which can further efforts in informing personal-ization of treatment in the era of cellular therapies. OBJECTIVE: Cortical mapping during tumor resection is critical to reducing the risk of neurological complications. While the somatosensory evoked potential (SSEPs) is widely employed to delineate the central sulcus, the evoked response can be misconstrued from the manual peak interpretation. This is due to the poor spatial resolution of the strip electrode as well as a result of edema or tumor infiltration of the overlying cortex. We present a real-time computer-based visualization system that uses recorded SSEPs with a subdural grid to aid in cortical mapping. Method: The neural data during electrical stimulation of the median nerve at 0.6Hz are picked up with a bio-amplifier. The stimulation artifact recorded from the bipolar electromyo-gram (EMG) is used as the stimulation onset. The ECoG data are assessed online with MATLAB Simulink to process and visualize the SSEPs waveform. The visualization system is programmed to display the SSEP's peak activation as a heat map on a 2D grid and projected onto a screen, showcasing the nature of the cortical activities over the contact surface area in real-time. RESULTS: SSEPs were recorded from 3 patients during their awake crani-otomies. The ECoG grid occupied a large cortical surface where the heatmap delineated the central sulcus. The color-coding of the heatmap revealed the anterior and posterior channels, providing a consistently high separation accuracy in each patient (accuracy = 98%). The map could be viewed at any time point along the SSEP trace without peak interpretation. CONCLUSION: We believe that this visualization system will aid in the rapid definition of the sensorimotor area during surgical planning to provide additional information during cortical mapping and facilitate interpreting ECoG grid data. OBJECTIVE: While the intraoperative assessment of the short-latency somatosensory evoked potentials (SSEPs) phase reversal in electrocortico-gram (ECoG) is routinely used to locate the central sulcus (CS), its spatial-spectral patterns in the primary sensorimotor cortex and the relation to consciousness is not well characterized. Our goal was to evaluate the functional utility of using median nerve stimulation-induced ECoG sub-band modulations in pre-and post-CS to differentiate between anesthetized and awake states. Method: SSEPs were recorded from the sensorimotor cortex with ECoG during routine intraoperative cortical mapping of 8 patients in the anesthetized and awake states. We conducted a time-frequency analysis on the SSEP trace in each state to extract the spectral modulations up to 900Hz and contrasted their intensity in M1 and S1 to discern awake from anesthetized states. RESULTS: We observed late gamma activity ranging from 60-250Hz, starting approximately 50ms after stimulation onset and extending up to 250ms. The late gamma activity was suppressed in the anes-thetized state in both M1 (p< 0.01) and S1 (p=0.0369). Whiles in the awake state, the late gamma power increased significantly with respect to baseline both in M1 (p=0.0180) and S1 (p< 0.01). The difference in late gamma power between the anesthetized and the awake state was highly significant for M1 (p< 0.01) and for S1 (p< 0.01). CONCLUSION: The results show that besides CS delineation, SSEPs long latency gamma modulations can serve as an additional biomarker to monitor the level of consciousness in neurosurgical practice. BACKGROUND: The median overall survival of patients with glio-blastoma (GBM) is 10-12 months and the five-year survival ranges from 5-10%. Favorable clinical and molecular prognostic markers have been described in the literature, although an individual patient's prognosis cannot easily be predetermined on these features as shown in the NOB-LTS project by Briceno et al. (SNO 2021). Until now, no radiologic characteristics have been correlated with better patient outcomes. We reviewed eighteen MRI features to determine if these could aid the identification of long-term survivors (LTS). METHODS: From the NOB-LTS cohort of our Natural History Study, 16 long-term survivals (LTS; ≥ 3 years survival post-diagnosis) were matched based on sex, age, and extent of resection to 32 control (STS, < 3 years survival post-diagnosis) patients with NGS-confirmed IDH-wt glioblastoma. Pre-surgical MRIs (T1, T1 post-contrast, T2, T2/FLAIR, DWI, ADC) were reviewed by three blinded reviewers and analyzed based on pre-set criteria. RESULTS: T1 hypointensity occurred commonly in the LTS (71%) vs STS (29%) while central heterogeneous signal occurred commonly in the STS (73%) vs LTS (50%) patients. Restricted diffusion was seen in 82% of STS and 70% of LTS cases. No difference was seen in irregular enhancement, peripheral and grey matter enhancement on T1 post-contrast images. Evaluation of T2/FLAIR images showed no difference in the number of lobes involved, presence of mass effect, heterogeneity of the signal or the perilesional signal abnormality or involvement of grey matter. CONCLUSION: Our results suggest that pre-surgical MRI might be a useful prognostic tool as T1 hypointensity more frequently, and T1 central het-erogeneity less commonly seen in the LTS group. These imaging findings may provide additional insights regarding the differences in tumor biology between LTS and STS. Further validation and correlation with histological/ molecular characteristics are planned to better predict patient outcomes with newly diagnosed glioblastoma. Volumetric measurements of whole tumor and its components on MRI scans, facilitated by automatic segmentation tools, are essential to reduce inter-observer variability in monitoring tumor progression and response assessment for pediatric brain tumors. Here, we present a fully automatic segmentation model based on deep learning that reliably delin-eates the tumor components recommended by the Response Assessment in Pediatric Neuro-Oncology (RAPNO) working group for evaluation