Mrina Eliseus Mtenga, “Radiomic Analysis of Brainstem Infiltration in Late-Stage Glioblastoma Patients”
Mentor: Peter LaViolette, MCW (partnership with UWM)
Poster #162
About 44,000 new cases of primary brain tumors are diagnosed in the United States each year, and glioblastoma (GBM) continues to be a major cause of cancer-related death 1, 2. Brainstem extension in GBM is characterized by tumor cells diffusely infiltrating the midbrain, pons, or medulla oblongata, rather than forming a localized mass. The increased observation of late-stage brainstem tumor invasion calls for more studies to examine its roles and whether there needs to be new and improved therapeutic practices to help improve overall patient survival care. Radiomics is an emerging field that extracts high-dimensional data from medical images to identify quantitative features and patterns, or signatures, that are invisible to the human eye 3-6. For brainstem extension, radiomic analysis could play a pivotal role in non-invasively characterizing the extent of infiltration and correlating these features with molecular and clinical profiles. This research used multimodal MRI and radiomics techniques to distinguish between brainstem invasion and mass effect in glioblastoma patients. MRI scans, including T1-weighted post-contrast, T2-weighted/FLAIR, diffusion tensor imaging (DTI), and magnetic resonance spectroscopy (MRS), were used to capture detailed anatomical, functional, and metabolic information. A histological component of this study will integrate machine learning-based annotation using k-nearest neighbors, naïve Bayes, decision trees, and random-under-sampling-boosted random forest (RUS Tree) to predict tumor annotations related to brainstem infiltration using segmented pathology data. This study is looking to identify unique radiomic features obtained from MR imaging that strongly correspond with brainstem invasion in glioblastoma. The longitudinal analysis of radiomics could reveal temporal patterns of tumor progression into the brainstem, providing a better understanding of the dynamics of GBM cell migration. Using radiomics to create advanced detection methods and molecular characterization of brainstem infiltrating cells potentially improves targeted therapies and advanced patient care where standard treatments fail.