<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>Brain tumor is a very serious disease from which many people die every day. Appropriate early diagnosis is extremely important in treatment. In recent years, machine learning methods have come to the aid of doctors, allowing them to automate the process of brain tumor detection. It is a useful tool that can support doctors in their daily work. We consider here the use of machine learning methods to detect brain tumors based on magnetic resonance images. We use artificial neural networks to classify the images into those containing and those without a brain tumor. More specifically, we apply convolutional neural networks on appropriately transformed input data. The three proposed convolutional neural network models were created based on the pre-trained VGG19, DenseNet-121, and InceptionV3 networks, which achieved an accuracy of 92.59%, with areas under the ROC curve ranging from 0.95 to 0.96. The precision, sensitivity, and F1-score are also satisfactory and promising. These results are better than those for the models presented on the Kaggle platform.</p>
</abstract>

Brain tumor is a very serious disease from which many people die every day. Appropriate early diagnosis is extremely important in treatment. In recent years, machine learning methods have come to the aid of doctors, allowing them to automate the process of brain tumor detection. It is a useful tool that can support doctors in their daily work. We consider here the use of machine learning methods to detect brain tumors based on magnetic resonance images. We use artificial neural networks to classify the images into those containing and those without a brain tumor. More specifically, we apply convolutional neural networks on appropriately transformed input data. The three proposed convolutional neural network models were created based on the pre-trained VGG19, DenseNet-121, and InceptionV3 networks, which achieved an accuracy of 92.59%, with areas under the ROC curve ranging from 0.95 to 0.96. The precision, sensitivity, and F1-score are also satisfactory and promising. These results are better than those for the models presented on the Kaggle platform.

Machine learning methods in the detection of brain tumors

Biometrical Letters

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

{"article-title":"Machine learning methods in the detection of brain tumors"}

Brain tumor is a very serious disease from which many people die every day. Appropriate early diagnosis is extremely important in treatment. In recent years,...

Machine learning methods in the detection of brain tumors

Data publikacji: 29 gru 2023

Zakres stron: 125 - 148

DOI: https://doi.org/10.2478/bile-2023-0009

Słowa kluczowe
brain tumor, classification, computer vision, convolutional neural network, machine learning, transfer learning

© 2023 Maria Reszke et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Machine learning methods in the detection of brain tumors

Data publikacji: 29 gru 2023

Zakres stron: 125 - 148

DOI: https://doi.org/10.2478/bile-2023-0009

Słowa kluczowebrain tumor, classification, computer vision, convolutional neural network, machine learning, transfer learning

© 2023 Maria Reszke et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Słowa kluczowe
brain tumor, classification, computer vision, convolutional neural network, machine learning, transfer learning