A Brief Survey on Feature Extraction Models for Brain Tumor Detection

Today, tumours are the second leading cause of cancer deaths. Cancer poses a significant threat to a large population of patients. The medical community needs a quick, automated, efficient, and trustworthy method for detecting tumours like brain tumours. Detection is crucial to effective treatment. If doctors are able to catch a tumour in its earliest stages, they have a better chance of preserving the patient's health. To do this, several distinct image processing methods are used. Through this method, doctors have been able to effectively treat tumours and save the lives of many patients. Tumors are simply abnormal growths of cells that cannot be stopped. As brain tumour cells multiply, they eventually deplete the brain's supply of nutrients. Clinicians currently use MR images (MRI) of the patient's brain to manually pinpoint the location and extent of a brain tumour. Brain tumours can develop at any age in both children and adults. However, this is not the case if detection is timely and accurate. This investigation focuses on three subtypes of brain cancer: gliomas, meningiomas, and pituitary tumours. While there have been numerous publications on the topic of brain tumour classification and prediction, very few have focused on the importance of feature extraction. Manual diagnosis and conventional feature extraction methods have their limitations, and new approaches are needed to overcome them. An automated diagnostic system is necessary for extracting features and making an accurate diagnosis of brain cancer. Although advancements are being made, automatic brain tumour diagnosis continues to struggle with issues like low accuracy and a high proportion of false-positive findings. In this research work, a brief survey is provided on feature extraction for brain tumor detection using machine learning and deep learning techniques.