Background Pretreatment assessments for glioblastoma (GBM) patients, especially elderly or frail patients, are critical for treatment planning. However, genetic profiling with intracranial biopsy carries a significant risk of permanent morbidity. We previously demonstrated that the CUL2 gene, encoding the scaffold cullin2 protein in the cullin2-RING E3 ligase (CRL2), can predict GBM radiosensitivity and prognosis. CUL2 expression levels are closely regulated with its copy number variations (CNVs). This study aims to develop artificial neural networks (ANNs) for pretreatment evaluation of GBM patients with inputs obtainable without intracranial surgical biopsies. Methods Public datasets including Ivy-GAP, The Cancer Genome Atlas Glioblastoma (TCGA-GBM), the Chinese Glioma Genome Atlas (CGGA) were used for training and testing of the ANNs. T1 images from corresponding cases were studied using automated segmentation for features of heterogeneity and tumor edge contouring. A ratio comparing the surface area of tumor borders vs. the total volume (SvV) was derived from the DICOM-SEG conversions of segmented tumors. The edges of these borders were detected using the canny edge detector. Packages including Keras, Pytorch, and TensorFlow were tested to build the ANNs. A 4-layered ANN (8-8-8-2) with a binary output was built with optimal performance after extensive testing. Results The 4-layered deep learning ANN can identify a GBM patient’s overall survival (OS) cohort with 80-85% accuracy. The ANN requires 4 inputs, including CUL2 copy number, patients’ age at GBM diagnosis, Karnofsky Performance Scale (KPS), and SvV ratio. Conclusion Quantifiable image features can significantly improve the ability of ANNs to identify a GBM patients’ survival cohort. Features such as clinical measures, genetic data, and image data, can be integrated into a single ANN for GBM pretreatment evaluation.

Background and Purpose: Genetic profiling for glioblastoma multiforme (GBM) patients with intracranial biopsy carries a significant risk of permanent morbidity. We previously demonstrated that the CUL2 gene, encoding the scaffold cullin2 protein in the cullin2-RING E3 ligase (CRL2), can predict GBM radiosensitivity and prognosis mainly due to the functional involvement of CRL2 in mediating hypoxia-inducible factor 1 (HIF-1) alpha; and epidermal growth factor receptor (EGFR) degradation. Because CUL2 expression levels are closely regulated with its copy number variations (CNVs), this study aims to develop an artificial neural network (ANN) that can predict GBM prognosis and help optimize personalized GBM treatment planning. Materials and Methods: Datasets including Ivy-GAP, The Cancer Genome Atlas Glioblastoma Multiforme (TCGA-GBM), the Chinese Glioma Genome Atlas (CGGA) were analyzed. T1 images from corresponding cases were studied using automated segmentation for features of heterogeneity and tumor edge contouring. Results: We developed a 4-layer neural network that can consistently predict GBM prognosis with 80-85% accuracy with 3 inputs including CUL2 copy number, patient age at GBM diagnosis, and surface vs. volume (SvV) ratio. Conclusion: A functional 4-layer neural network was constructed that can predict GBM prognosis and potential radiosensitivity.

Molecular marker-based glioblastoma (GBM) subclassification is emerging as a key factor in personalized GBM treatment planning. Multiple genetic alterations, including methylation status and mutations, have been proposed in GBM subclassification. RNA-Sequence (RNA-Seq)-based molecular profiling of GBM is widely implemented and readily quantifiable. Machine learning (ML) algorithms have been reported as an applicable method that can consistently subgroup GBM. In this study, we systematically studied the applicability of the commonly used ML algorithms based on The Cancer Genome Atlas Glioblastoma Multiforme (TCGA-GBM) dataset and cross-validated in the Chinese Glioma Genome Atlas (CGGA) dataset.  ML algorithms studied include Binomial and multinomial Logistic Regression, Linear discriminant analysis, Decision trees, K-Nearest Neighbors, Gaussian Naive Bayes, Support Vector Machines, Gradient Boosting, Voting Ensemble, Multi-Layer Perceptron.  RNA-Seq data of 44 biomarkers were passed through the algorithms for performance evaluation. We found ML algorithms Support Vector Machines, Multi-Layer Perceptron s, and Voting Ensemble are best equipped in assigning GBM to correct molecular subgroups of GBM without histological studies.