Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification

Brain tumour classification using Magnetic Resonance Imaging (MRI) is crucial for medical decision-making. The variability in tumour shape, size, and position poses challenges to classification methods. Convolutional Neural Networks (CNNs) are commonly used due to their proven performance, but their...

Full description

Saved in:
Bibliographic Details
Main Authors: Irwan Budi Santoso, Shoffin Nahwa Utama, Supriyono
Format: Article
Language:English
Published: Elsevier 2025-07-01
Series:Array
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2590005625000256
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849422216466268160
author Irwan Budi Santoso
Shoffin Nahwa Utama
Supriyono
author_facet Irwan Budi Santoso
Shoffin Nahwa Utama
Supriyono
author_sort Irwan Budi Santoso
collection DOAJ
description Brain tumour classification using Magnetic Resonance Imaging (MRI) is crucial for medical decision-making. The variability in tumour shape, size, and position poses challenges to classification methods. Convolutional Neural Networks (CNNs) are commonly used due to their proven performance, but their effectiveness diminishes with the high variability of tumour characteristics. This study proposes a meta-learning approach, leveraging the softmax average of multiple CNN models with a Multi-Layer Perceptron (MLP) as the meta-learner. The base-learner models include MobileNetV2, InceptionV3, Xception, DenseNet201, and ResNet50. This approach combines the softmax outputs of these CNN models, capturing their strengths to handle diverse tumour characteristics. The averaged outputs are fed into the MLP for increased classification performance. To evaluate the proposed method, we used several brain MRI image datasets, including Dataset 1 (Thomas Dubail Dataset), Dataset 2 (Mesoud Nickparcar Dataset), and Dataset 3 (Fernando Feltrin Dataset). The test results showed the proposed method's effectiveness in improving classification performance. For Dataset 1, the MLP with one hidden layer (128 neurons) achieved 97.47 % accuracy, improving the base learners' performance by 1.94 %–7.42 %. On Dataset 2, the MLP with 64 neurons reached 99.54 % accuracy, with a 0 %–2.44 % improvement. For Dataset 3, an MLP with two hidden layers (256 and 125 neurons) achieved 98.87 % accuracy, enhancing performance by 0.46 %–5.67 %.
format Article
id doaj-art-539d63477ce7472ca079efc4f4842637
institution Kabale University
issn 2590-0056
language English
publishDate 2025-07-01
publisher Elsevier
record_format Article
series Array
spelling doaj-art-539d63477ce7472ca079efc4f48426372025-08-20T03:31:11ZengElsevierArray2590-00562025-07-012610039810.1016/j.array.2025.100398Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classificationIrwan Budi Santoso0Shoffin Nahwa Utama1 Supriyono2Corresponding author.; Informatics Engineering, Faculty of Science and Technology, Universitas Islam Negeri Maulana Malik Ibrahim, Malang, IndonesiaInformatics Engineering, Faculty of Science and Technology, Universitas Islam Negeri Maulana Malik Ibrahim, Malang, IndonesiaInformatics Engineering, Faculty of Science and Technology, Universitas Islam Negeri Maulana Malik Ibrahim, Malang, IndonesiaBrain tumour classification using Magnetic Resonance Imaging (MRI) is crucial for medical decision-making. The variability in tumour shape, size, and position poses challenges to classification methods. Convolutional Neural Networks (CNNs) are commonly used due to their proven performance, but their effectiveness diminishes with the high variability of tumour characteristics. This study proposes a meta-learning approach, leveraging the softmax average of multiple CNN models with a Multi-Layer Perceptron (MLP) as the meta-learner. The base-learner models include MobileNetV2, InceptionV3, Xception, DenseNet201, and ResNet50. This approach combines the softmax outputs of these CNN models, capturing their strengths to handle diverse tumour characteristics. The averaged outputs are fed into the MLP for increased classification performance. To evaluate the proposed method, we used several brain MRI image datasets, including Dataset 1 (Thomas Dubail Dataset), Dataset 2 (Mesoud Nickparcar Dataset), and Dataset 3 (Fernando Feltrin Dataset). The test results showed the proposed method's effectiveness in improving classification performance. For Dataset 1, the MLP with one hidden layer (128 neurons) achieved 97.47 % accuracy, improving the base learners' performance by 1.94 %–7.42 %. On Dataset 2, the MLP with 64 neurons reached 99.54 % accuracy, with a 0 %–2.44 % improvement. For Dataset 3, an MLP with two hidden layers (256 and 125 neurons) achieved 98.87 % accuracy, enhancing performance by 0.46 %–5.67 %.http://www.sciencedirect.com/science/article/pii/S2590005625000256Brain tumourMagnetic resonance imagingConvolutional neural networkMulti-layer perceptronSoftmaxBase-learner
spellingShingle Irwan Budi Santoso
Shoffin Nahwa Utama
Supriyono
Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
Array
Brain tumour
Magnetic resonance imaging
Convolutional neural network
Multi-layer perceptron
Softmax
Base-learner
title Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
title_full Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
title_fullStr Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
title_full_unstemmed Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
title_short Meta-learning based softmax average of convolutional neural networks using multi-layer perceptron for brain tumour classification
title_sort meta learning based softmax average of convolutional neural networks using multi layer perceptron for brain tumour classification
topic Brain tumour
Magnetic resonance imaging
Convolutional neural network
Multi-layer perceptron
Softmax
Base-learner
url http://www.sciencedirect.com/science/article/pii/S2590005625000256
work_keys_str_mv AT irwanbudisantoso metalearningbasedsoftmaxaverageofconvolutionalneuralnetworksusingmultilayerperceptronforbraintumourclassification
AT shoffinnahwautama metalearningbasedsoftmaxaverageofconvolutionalneuralnetworksusingmultilayerperceptronforbraintumourclassification
AT supriyono metalearningbasedsoftmaxaverageofconvolutionalneuralnetworksusingmultilayerperceptronforbraintumourclassification