Learning objectives
1. Explore the conventional and emerging applications of MRI radiomics in post-treatment GBM.
2. Describe the current limitations and future prospects of MRI radiomics in neuro-oncology.
3. Examine the clinical and methodological factors that influence the diagnostic performance of machine learning models.
Background
Early assessment of therapeutic response is crucial in patients with glioblastoma (GBM). However, differentiating tumour progression (TP) from radiation necrosis (RN) remains difficult on conventional imaging alone [1]. Radiotherapy-induced changes can mimic malignant processes with substantial overlap of imaging features on standard magnetic resonance imaging (MRI) [1,2]. Combining data from multiple imaging parameters yields higher diagnostic accuracy, as it captures voxel-based heterogeneity in relation to the tumour’s biophysical properties [3,4]. Machine learning (ML) and radiomics have emerged as valuable tools to evaluate high-dimensional features that...
Imaging findings OR Procedure details
A standard workflow of radiomics analysis can be seen in Figure 1 [7]. [Fig 1] Substantial effort has been made to improve the diagnostic performance of radiomics-based ML models in neuro-oncology. Novel ML and deep-learning (DL) methods used to distinguish RN from TP are detailed below.
Diffusion MR Radiomics
The GBM tumour microenvironment is highly heterogeneous at a molecular, cellular, and structural level [8]. Thus, accurate tissue characterisation remains difficult on conventional imaging alone. Structural imaging has limited ability to capture dynamic changes in tumour...
Conclusion
Multiparametric MRI radiomics may provide a non-invasive means of differentiating RN from recurrent GBM. Over time, these features will have to be aligned with underlying tissue properties to accurately reflect the tumour biology of GBMs. Novel machine learning methods holds considerable potential to individualise therapeutic monitoring in neuro-oncology
References
1. Chukwueke UN, Wen PY. Use of the response assessment in neuro-oncology (RANO) criteria in clinical trials and clinical practice. CNS Oncology. 2019;8(1).
2. Ellingson BM, Chung C, Pope WB, Boxerman JL, Kaufmann TJ. Pseudoprogression, radionecrosis, inflammation or true tumor progression? challenges associated with glioblastoma response assessment in an evolving Therapeutic Landscape. Journal of Neuro-Oncology. 2017;134(3):495–504.
3. Zikou A, Sioka C, Alexiou GA, Fotopoulos A, Voulgaris S, Argyropoulou MI. Radiation Necrosis, pseudoprogression, Pseudoresponse, and tumor recurrence: Imaging challenges for the evaluation of treated gliomas. Contrast...