The role of machine learning in detecting primary brain tumors in Saudi pediatric patients through MRI images

Abstract

Brain tumors are defined as the uncontrollable growth of cells. They are known as the leading cause of death among pediatric patients. Artificial Intelligence is considered as one of the techniques used to improve radiology departments. Magnetic Resonance Imaging (MRI) is a key tool in detecting brain tumors. Medical images can be challenging and time-consuming for radiologists, particularly in the case of pediatric patients where the tumors may be small and difficult to detect. This study aims to explore the role of AI and measure the accuracy of AI methods in detecting primary brain tumors in pediatric patients by using MRI images. A retrospective and analytical study was conducted, MRI images of pediatric patients with primary brain tumors were acquired from the Picture Archives and Communication System (PACS) of the radiology department at King Abdullah Specialist Children Hospital (KASCH). This study continues a total of (6435) MR images, this dataset was divided and expressed in different subsets, 70% of the images were trained (4493 in total) for model accuracy evaluation, and the rest which is 30% (1942) were tested images contains three types of primary brain tumors, glioma, Primitive Neuro-Ectodermal Tumors and pituitary tumors. Moreover, this dataset also contains normal images. Image classification and detection have been done by using Machine Learning algorithms which are coded in Python programming language. After applying Machine Learning algorithms on MRI images, Artificial Neural Network method resulted in an accurate detection of pediatric primary brain tumors and matched the radiologist's report. New Artificial Intelligence techniques applied in the imaging department have increased the information obtained from images to improve the accuracy of diagnosis along with radiologist's reports which will aid in better management of the patient's condition.

Ähnliche Arbeiten