Data Science Project Explainable AI: Brain Tumor Classification with EfficientNet and Gradient-Weighted Class Activation Mapping (Grad-CAM) Visualization

btd
15 min readSep 21, 2021
Canva

Introduction

A brain tumor is a growth of abnormal cells that have formed in the brain. Brain and other nervous system cancer is the 10th leading cause of death for men and women. According to National Brain Tumor Society (2021), an estimated 700,000 people in the United States are living with a primary brain tumor, and approximately 85,000 more will be diagnosed in 2021 and 18,600 of those people will die.

Early and accurate diagnosis of a brain tumor is essential for disease management. However, with so many types of brain tumors, accurately diagnosing one can be a complicated process. Among the most helpful tools for brain tumor detection is Magnetic Resonance Imaging (MRI), a medical imaging technique that provides valuable data insights regarding the patients’ health and disease information. Medical imaging is an immensely data-heavy field and human interpretation can be prone to errors due to various factors such as stress, lack of context, and lack of expertise. This is where deep neural network has proven its potential for effectively and accurately classifying medical images, aiding human professionals in disease diagnosis and treatment.

--

--

btd
btd

No responses yet