Your Donations in Action: Anna Sophia McKenney, PhD

Glioblastoma (GBM) is the most common primary brain tumor and a rapidly fatal disease. The clinical and imaging diagnosis of tumor progression is often confounded by pseudoprogression, a treatment-induced inflammatory reaction with increasing and/or new enhancing lesions on MRI that mimics tumor progression. Pseudoprogression occurs in up to one-third of patients. To help identify those who could potentially avoid invasive neurosurgery and/or delay inappropriate treatment changes, an algorithm is being developed to identify lesions with pseudoprogression noninvasively using dynamic contrast enhanced MRI (DCE-MRI) perfusion with comparison to a gold standard of histopathological diagnosis at subsequent surgery.

In her 2017 RSNA Research Medical Student Grant study, “Texture Feature and 4D Texture Kinetic Analyses of Dynamic Contrast Enhanced T1 MRI Perfusion of Pseudoprogression in Glioblastoma,” Anna Sophia McKenney, PhD, is developing a model to computationally differentiate between true progression and pseudoprogression. Of 179 patients who underwent a second primary neurological tumor resection at her institution, 110 patients were confirmed to have primary GBM by previous pathology and 93 patients had appropriate perfusion imaging studies for inclusion. Of these, histopathological review classified 57 patients with progression of disease and 36 with pseudoprogression.

“Machine learning algorithm analysis is ongoing, but this study highlights the potential of texture features of perfusionderived series. The use of textures has great potential impact on radiologists everywhere as it may allow them to make more nuanced observations about a patient’s state, and machine learning algorithms like the ones we hope to create will offer additional adaptations,” Dr. McKenney said. “In time, we hope that the properties of perfusion scans like these can be used prospectively to guide treatment decisions about whether to pursue invasive therapy sooner or watchful waiting.”