Because their accuracy makes it possible to personalize treatments, predict recurrence and therefore improve the outcomes of patients with intracranial tumors, new imaging technologies represent a major step forward for neuro-oncology.

Just a few years ago, half the world’s population did not have access to medical imaging. Today, the number of images, in both 2D and 3D, is growing exponentially. Doctors specializing in neuro-oncology are confronted with a paradoxical injunction: faced with the need to detect diseases at an early stage, improve diagnoses and evaluate treatments, they must at the same time compensate for the widespread shortage of X-ray technicians and ensure that healthcare costs are kept to a minimum. AI is a way for them to meet these challenges.

2021 was a pivotal year for the fight against brain tumors, for which the WHO presented a new classification. It focused on molecular markers, particularly the IDH mutation, and made the previous 2016 classification obsolete. This new nomenclature requires more accurate detection of histopathological changes and imaging biomarkers. This is because it takes into account, for example, the fact that even non-contrast tumors can be aggressive. This highlights the importance of a multimodal approach to patient diagnosis and management. The traditional sequences of MRI, while they remain important, are no longer sufficient to address these new challenges in neuro-oncology. 

The rise of radiomics

Advanced MRI, including techniques such as perfusion, spectroscopy and saturation transfer imaging (CEST), offers much needed additional capabilities. It allows for better characterization of tumors by detecting microscopic details and subtle abnormalities that escape conventional MRI. For example, magnetic resonance spectroscopy analyzes chemical metabolites in brain tissue, helping to differentiate tumor types and determine their grade. MRI infusion measures blood flow through the tissues, providing essential information about the vascularization of tumors.

But it is on the artificial intelligence and radiomics side that the greatest advances are expected. Radiomics uses state-of-the-art image analysis technologies to extract detailed and precise information. Once obtained, the data are analyzed using statistical methods and machine learning techniques. Radiomics can thus provide detailed information about the texture, boundaries, and heterogeneity of tumors. These details can be integrated into clinical and genomic data to personalize treatments, plan operations, predict recurrences and therefore improve patient outcomes.

Managing and interpreting data

However, the integration of AI into neuro-oncology poses challenges, particularly in terms of managing and interpreting the large amount of data produced. It is essential to develop robust systems to make this information easily usable in clinical practice. Many experts also stress the need to standardize procedures and validate AI models, to avoid over-adjustments and ensure their clinical relevance.

Optimized protocols and AI therefore open up new perspectives for improving the diagnosis and treatment of intracranial tumors. But only smooth collaboration between radiologists, neurosurgeons, oncologists, and researchers will make it possible to take full advantage of this potential and restore hope to patients.