An abnormal cell growth in the brain is called a brain tumor. In case not treated at an initial stage, it can cause death. Despite many great efforts and encouraging outcomes in this area, accurate division and classification remain a difficult task. Tumors may be either malignant (cancerous) or benign (non-cancerous). Medical imaging techniques that are used regularly include computed tomography (CT), positron outflow tomography (PET), CT/PET, attractive reverberation imaging (MRI), and other comparative techniques. Restorative imaging has a crucial role in brain tumor localization by providing valuable information for conclusion, surgical planning. A major challenge for brain tumor discovery emerges from the varieties in tumor area, shape, and measure. The objective of this paper is to provide machine learning techniques, challenges, and future directions in enhancing healthcare access. Brain imaging modalities of tumor detection, challenges, and future directions in enhancing healthcare access. Finally, this survey provides types of brain tumors and classifications of brain tumors according to the WHO (World Health Organization). These developments allow early diagnosis, improved treatment planning, and personalized care for patients with glioma, meningioma, and pituitary tumors. Addressing these issues through federated learning, explainable AI frameworks, and generative models for data augmentation can further improve the reliability and accessibility of ML tools in clinical settings.