Advances in Brain Tumor Segmentation Using Transformer Models

Brain tumors, defined by the uncontrolled proliferation of abnormal cells within the brain, represent a formidable global health challenge, affecting individuals across all age groups. These neoplasms exhibit remarkable diversity, ranging from benign, slow-growing low-grade tumors, such as meningiomas, which may allow for a life expectancy of several years, to aggressive, malignant high-grade tumors, like glioblastomas, which are associated with a median survival of less than two years. Gliomas, the most prevalent primary brain tumors, are classified by the World Health Organization into four grades, with low-grade gliomas (grades I–II) being less infiltrative and high-grade gliomas (grades III–IV) exhibiting rapid growth and poor prognosis ( Komori, 2022 ). Factors such as exposure to ionizing radiation, genetic predispositions, and family history of brain cancer are known to elevate risk, yet the precise etiology of most brain tumors remains elusive. The accurate segmentation of brain tumors and their associated subregions, active tumorous tissue, necrotic core, and peritumoral edema is pivotal for clinical applications, including precise diagnosis, surgical planning, radiation therapy targeting, chemotherapy optimization, and longitudinal monitoring of disease progression ( Alqhtani et al., 2024 ).