Authors :
Alphonse MUGABIRE Masters; Dr. Wilson Musoni
Volume/Issue :
Volume 8 - 2023, Issue 9 - September
Google Scholar :
https://bit.ly/3TmGbDi
Scribd :
https://tinyurl.com/3hkpxna4
DOI :
https://doi.org/10.5281/zenodo.8378865
Abstract :
The incidence of brain tumors, a highly
malignant form of cancer, is widespread around the
world, affecting millions of individuals. Early detection
plays a crucial role in saving lives, but the process of
identifying and classifying tumor types accurately
requires reviewing numerous MRI images. Deep learning
models have the capability to handle such large datasets
and provide precise results. However, it is important to
note that the outcomes produced by deep learning models
can vary depending on the dataset used.
This comparative study focused on evaluating the
effectiveness of deep learning models on two distinct
Magnetic Resonance Imaging (MRI) brain tumor
datasets. The goal of this research was to identify the best
deep learning model that can achieve the highest
accuracy in detecting brain tumors compared to others in
the dataset. The models were individually applied to pre-
processed datasets to extract features from the MRI
images. Segmentation of tumor regions can be
challenging due to the visual similarity between normal
tissue and brain tumor cells. Therefore, an automatic
tumor detection approach with high accuracy is
necessary.
To train our algorithm effectively, a diverse range of
MRI images with different tumor sizes, locations, shapes,
and intensities was utilized. We employed "TensorFlow"
and "Keras" frameworks within the programming
language "Python" to develop our optimal solution, as
this language provides efficient functionality for rapid
implementation. As part of the research process, a
comprehensive literature review was conducted, and
secondary data was collected. Performance metrics were
employed for data analysis, leading to conclusions and
recommendations for the most suitable deep learning
approach model.
The incidence of brain tumors, a highly
malignant form of cancer, is widespread around the
world, affecting millions of individuals. Early detection
plays a crucial role in saving lives, but the process of
identifying and classifying tumor types accurately
requires reviewing numerous MRI images. Deep learning
models have the capability to handle such large datasets
and provide precise results. However, it is important to
note that the outcomes produced by deep learning models
can vary depending on the dataset used.
This comparative study focused on evaluating the
effectiveness of deep learning models on two distinct
Magnetic Resonance Imaging (MRI) brain tumor
datasets. The goal of this research was to identify the best
deep learning model that can achieve the highest
accuracy in detecting brain tumors compared to others in
the dataset. The models were individually applied to pre-
processed datasets to extract features from the MRI
images. Segmentation of tumor regions can be
challenging due to the visual similarity between normal
tissue and brain tumor cells. Therefore, an automatic
tumor detection approach with high accuracy is
necessary.
To train our algorithm effectively, a diverse range of
MRI images with different tumor sizes, locations, shapes,
and intensities was utilized. We employed "TensorFlow"
and "Keras" frameworks within the programming
language "Python" to develop our optimal solution, as
this language provides efficient functionality for rapid
implementation. As part of the research process, a
comprehensive literature review was conducted, and
secondary data was collected. Performance metrics were
employed for data analysis, leading to conclusions and
recommendations for the most suitable deep learning
approach model.
