Abstract:
The brain tumor is one of the deadliest diseases in the world nowadays. Only in the United States of
America, today the number of people having brain tumor is more than 700,000 [1]. Approximately
16,000 people would die in the process of a brain tumor in the year 2020 [1]. It'll be really grateful
for monitoring and identification if the characterization of tumors in the brain can be done at a very
pre-mature stage. Numerous researchers have already taken some attempts to use various techniques, such as digital mammography, MRI, CT (Computed Tomography), etc. To detect the exact type of
brain tumor from MRI images CapsNets became an improved architecture. Since these networks
can operate with fewer training samples. We used a dataset from kaggle to monitor the tumor in the
brain at the very initial stage. AT first, in the CNN model, each of the input pictures will move
through a set of filter convolution layers (called Kernels), then pooling, then completely related
layers (FC) and applying Soft-max function to define a probabilistic meaning object. The outcome
from the proposed technique reveals that 92 percent of accuracy can be gained from this technique.