English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

PURPOSE Epidermal growth factor receptor amplification (EGFRamp), occurs in &gt;60% of Glioblastoma (GBM) patients, and can predict response to alkylating chemotherapy. However, due to high genetic instability in GBM, a significant caveat to sampling EGFRamp during biopsy, is that reliable identification of its presence is dependent on the site of the biopsy, and may vary throughout the tumor. Previous studies have employed radiomics (computerized feature extraction) from the entire tumor to characterize EGFRamp presence. Our work is based on the rationale that radiomics measurements from stereotactic biopsy locations on Gd-T1w MRI, may allow to build accurate training models for predicting EGFRamp compared to using radiomic features from the entire tumor. METHODS MRI scans from 78 GBM subjects (33 EGFRamp, 45 non-amp) were obtained, of which 54 subjects were used for training while 24 were used for testing. Stereotactic biopsy locations were mapped by co-registering CT-images with MRIs (confirmed by two radiologists). For each biopsy region (1-cm diameter sphere), gradient entropy features and gender information were used to train a discriminant analysis classifier, on a voxel-wise manner. Our training model also included per-voxel spatial probabilities that were obtained by building atlases to quantify frequency of occurrence of EGFRamp and non-amp cases. All of these features were then used within the test set to predict EGFR status. RESULTS The model achieved a training accuracy of 81.4%, correctly classifying 17/23-amp, 27/31 non-amp cases. Accuracy on test set was 79.2% (7/10 amp, 12/14 non-amp cases correctly classified). Extracting features from the entire tumor achieved a training accuracy of 74% (17/23-amp, 23/31 non-amp cases correctly classified), and a testing accuracy of 70.8% (6/10 amp, 11/14 non-amp cases correctly classified). CONCLUSION Radiomic features, trained using features from only the stereotactic biopsy locations, may be more reliable in training models to predict EGFR mutation in GBM.

NIMG-75. RADIOMIC FEATURES LOCALIZED TO STEREOTACTIC BIOPSY LOCATIONS CAN CAPTURE EGFR PRESENCE IN GLIOBLASTOMA