Smartbrush tool

For the PICTURE data, 275 GBM and 205 LGG cases were manually segmented into 3 classes consistent with the BraTS challenges – whole tumor (WT), tumor core (TC), and enhancing tumor (ET), see Fig. 1. The WT defines the full extent of the tumor, including the tumor core and oedema, indicated by hyperintensity on FLAIR and T2w. The TC is the main body of the tumor and most likely area of resection. The TC includes the enhancing tumor (ET) and necrosis.
Manual segmentations were carried out according to the VASARI Research Project (https://wiki.cancerimagingarchive.net/display/Public/VASARI+Research+Project). One rater (HP) with 9 years of brain MRI manual segmentation experience performed segmentations under the supervision and approval of an expert neuroradiologist (FB), using the semiautomatic SmartBrush tool (BrainLab, Feldkirchen, Germany). The rater’s performance was in line with experts^{29 (link)}. All segmentations were exported on the T1c image. The segmentation was resampled to SRI24 atlas space using the same transform from the T1c to SRI24 registration.

Clinical notes of eligible patients were reviewed for demographic data (gender, age, Karnofsky Performance Status (KPS)), comorbidities, diagnosis, previous treatment, location of the tumor-associated cyst, and preoperative clinical status. Functional assessment included a range of focal deficits, such as motor deficits, sensory deficits, aphasia, apraxia, cerebellar symptoms, and neurocognitive changes, which encompassed alterations in personality and memory. We also considered the presence or frequency of seizures and signs indicative of elevated intracranial pressure, such as headache, nausea, vomiting, and a decreased level of alertness. “Immediate” changes in symptoms or cyst volumes were defined as changes that were registered before discharge from our department. Semiautomatic quantitative volumetry of singular or multiple intracranial cysts was performed on T2-weighted MRI images preoperatively, before discharge, and during follow-up using the SmartBrush tool (Brainlab, Munich, Germany).
The primary outcome parameters were the postoperative course of symptoms, cyst volumes, and cyst progression-free survival (cyst PFS), defined as the time after treatment with permanently reduced cyst volume. Patients were categorized as achieving “cyst progression-free survival” when their cyst volume remained consistently reduced within a follow-up of 12 months after the intervention.

Statistical analysis and figure editing were performed using JMP 14.0 software (SAS Institute, Cary, NC, USA), GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA, USA) and the open-source GIMP2 program. Evaluation of the immunohistochemical preparations was performed using a BX50 light microscope (Olympus, Tokyo, Japan). Tumor volumetriation was performed using the SmartBrush tool of the Brainlab Elements software (Brainlab AG, Munich, Germany).
Descriptive statistical methods mean (±SD) or median and interquartile range (IQR) (25–75%) were used to analyze data. Shapiro-Wilk test was used to assess normality of continuous variables. Normally distributed continuous variables (Hb value at hospital discharge) were compared with the two-sided T-test. Non-normally distributed continuous variables were compared with Mood’s median test. Categorical variables were compared with Pearson’s Chi Square test. For multivariate analysis, logistic regression was conducted. Survival analyses were performed using Kaplan-Meier analyses. To compare the survival curves, we used Wilcoxon and log-rank tests for censored data. A p-value < 0.05 was considered statistically significant.