With the goal to augment the standard of medical care, the exploration of clinical images offers helpful advice for practitioners to comprehend illnesses and look into diagnostic difficulties. Delineating cerebral tumor is one of the several clinical image interpretation jobs that has garnered a lot of interest from scientists across the globe. Brain tumor delineation on magnetic resonance images (MRI) entails separating brain malignancies from healthy brain tissue. Creating a binary or multi-class segmentation map that precisely depicts the tumor's position along with its dimension is the intention of segmenting brain tumors. Tumor segmentation from an MRI by hand is laborious and vulnerable to mistakes. A quick and precise process for delineating tumors is required. In computer vision, bio-inspired optimization algorithms have lately demonstrated exceptional performance for problems involving image segmentation and classification. A novel bio-inspired optimization system that has validated efficacy in resolving numerous optimization issues is the flower pollination algorithm (FPA). However, the equilibrium struck between the steps of exploration and exploitation is crucial to the efficacy of FPA. Despite the instance of many optima, simple exploitation operations encourage suboptimal solutions, whereas pure exploration procedures favor non-accurate solutions. The efficacy and resilience of the original FPA technique were enhanced in the existing investigation by introducing adaptive parameter tuning, which dynamically adjusts control variables such as step size and normalized fitness value. This enhanced the algorithm's capacity to respond to a range of brain tumor sizes, shapes, and complexity levels. BraTS images are segmented using the experiment's defined methodology. The proposed method delivers the average computation time: MSE, PSNR, TC, and DS were 16.62 s, 0.18, 55.66 dB, 89.63%, and 92.66%, correspondingly. The program can automatically distinguish brain cancers from MR images in a matter of seconds.