1. Introduction
The brain is the most complex and complicated part of the human body. Its structure is not easy to understand, as more than one hundred nerves communicate with each other to enable brain function [
1]. Brain-related diseases are spreading day by day [
2]. Some disorders can be detected at early stages, but some need proper diagnosis, like brain tumours, a type of brain-related disease that may lead to death. It has two major types: malignant or high-grade tumours and benign or low-level tumours [
3]. High-level tumours are cancerous and may be metastatic and may affect other organs, while low-grade tumours are non-cancerous and non-metastatic. High-level-grade tumours are one of the most life-threatening tumour types. They may lead to death instantly due to their metastatic property. Furthermore, malignant and benign tumours are categorized as glioma, meningioma, and pituitary tumours [
4]. According to statistics and research, in 2023, almost 1.4 million people struggled with brain tumours. Around 18,990 people in the world have died due to brain tumours [
5]. This number is expected to increase to 24 million by 2035. According to Cancer Net, there is a 36% five-year survival rate and over 30% ten-year survival rate of brain tumours [
5].
Figure 1 shows the difference between the average normal brain and tumour brain.
The Internet of Medical Things (IoMT) is the interconnected system of software applications, medical devices, and health products and services [
6]. IoMT refers to the system of medical devices and applications connected to healthcare IT systems through computer networks [
7]. These devices collect, analyse, and transmit health data to healthcare providers [
8]. IoMT enables real-time monitoring and improved healthcare outcomes to a large extent. IoMT includes smart medical equipment and mobile health applications that help in effective patient management [
9]. Brain tumour requires special treatment and patient care. Therefore, IoMT offers efficient services in the healthcare department to effectively manage patients suffering from brain tumours [
8]. IoMT has combined several automated disease identification models for appropriate disease diagnosis [
9].
Image processing has become an essential part of the field of computer vision. It is used in several medical fields to diagnose disease and abnormality of any body part. This terminology is used in different fields like medicine, geography, and multimedia. Medical experts like radiologists and doctors interpret medical images in the medical field. Still, due to the chances of human error and fatigue, they use computer-aided systems to obtain better results [
10]. A biopsy is one of the techniques used for diagnosis with the help of a needle, which may lead to infections. The second technique that is used widely is X-ray. The excess use of this technique may lead to skin cancer or brain cancer. The third technique used for detection is positron emission tomography (PET). In this technique, images of the brain are captured using a radioactive tracer. These images help doctors detect the disease, but it may lead to unpleasant pain and conditions. A CT scan uses the different views and angles of the brain, or any area for which images are captured, and the sections are known as slices. However, the images received using this technique could be more transparent for detecting a high-level medical disease.
Magnetic resonance imaging (MRI) is used to scan the brain to detect brain tumours. MRI images provide better and more precise results than CT (computed tomography) scans. X-ray radiation is not used in MRI scanning, and there is no pain throughout the procedure. Several scholars and researchers used this scanning technique in their research as it is easy to scan an image and then transfer it to the computer system for calculating the results. MRI scans are of several types, including T1- and T2-weighted, and Flair images [
11]. These images are quite different from each other. T1-weighted images show cerebrospinal fluid (CFS) as dark, while T2-weighted images show cerebrospinal fluid as bright. Flair images are like T2-weighted images, while abnormalities are brightly visible in these images. Three-dimensional views of the brain are used in scanning. Coronal, horizontal, and sagittal views are included for scanning the brain from all sides to detect the abnormality properly. Experts are needed to make a good decision. In recent studies, many techniques have been used for detecting brain tumours using MRI as the manual detection of cancer may involve human error, which can be life-threatening.
Machine learning (ML) techniques have shown a significant impact on medical image processing. ML can help doctors in different medical fields. It assists doctors in the radiology, cardiology, and neurology fields. In the past few years, the ML field has shown state-of-the-art results in image recognition with a high accuracy rate. ML can extract prominent features not adequately visible to the human eye. ML techniques often depend upon voxel intensities and textural features. Individual vectors are classified based on feature vectors. The error rate and computational cost are less in the image recognition field using ML techniques. MLT was chosen due to its high examination accuracy with less complex datasets [
12]. In the past few years, two techniques have been used in the ML approach: (1) the supervised method and (2) the unsupervised method. The supervised method includes input classes. This method comprises an artificial neural network (ANN) and a support vector machine (SVM).
Meanwhile, the unsupervised method consists of no input classes. This method includes K-mean clustering and ISODATA techniques. Along with ML, the Internet of Things (IoT) also has shown benefits in the healthcare sector. It has exhibited state-of-the-art results by connecting different medical devices to provide better medical services. The network created by these devices is also known as the IoMT. IoMT has shown beneficial outcomes in real-time settings [
13].
In this area, ML has been demonstrated as one of the most appropriate computational models with the embedded intelligence of IoMT for the prediction and contemporary diagnosis of diseases, especially brain tumours. Many efforts have been made to accurately identify brain tumours from different types of brain images. The existing solutions are limited to a few classes that are unable to classify the stage of brain tumours. Multi-class brain tumour identification solutions are limited and suffer from low accuracy. Existing techniques and methods have limitations in using small datasets, poor texture and resolution of MRI scans, and low accuracy in prediction. The multi-class identification of brain tumours is still challenging in the fields of radiology and neurology as compared to a binary classification of brain tumours. The multi-class identification of brain tumours in the IoMT environment is even more important. It has a significant advantage in the efficient diagnosis of patients suffering from brain tumours, given the limitations of other detection methods in the existing literature.
To address the limitations of existing methods, a new solution is proposed, and the major contributions of the proposed solution are as follows:
To design and implement a neural network model based on the EfficientNet architecture for the accurate classification of brain tumours;
To optimize and fine-tune the EfficientNet model to achieve the highest possible accuracy and efficiency in brain tumour identification;
To explore and apply various data augmentation and preprocessing techniques to increase the generalization capability of the EfficientNet model for accurate brain tumour classification;
To evaluate the performance of the EfficientNet model using various metrics such as accuracy, precision, recall, and F1 score in classifying four brain tumour classes.
This study is organized into different sections: Literature Review, Materials and Methods, Results, and Discussion. Conclusions are drawn in the final section.
Author Contributions
Conceptualization, A.I. and R.J.; methodology, A.I. and R.J.; software, A.I.; validation, A.I., M.A.J. and R.J.; formal analysis, A.I.; investigation, R.J.; resources, M.A.J.; data curation, A.I.; writing—original draft preparation, A.I.; writing—review and editing, R.J. and M.A.J.; visualization, R.J.; supervision, M.A.J. and R.J.; project administration, M.A.J.; funding acquisition, A.I. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Data are available on request from the authors.
Conflicts of Interest
The authors declare no conflict of interest.
References
- Li, Y.Q.; Chiu, K.S.; Liu, X.R.; Hsiao, T.Y.; Zhao, G.; Li, S.J.; Lin, C.P.; Sun, C.W. Polarization-sensitive optical coherence tomography for brain tumor characterization. IEEE J. Sel. Top. Quantum Electron. 2019, 25, 1–7. [Google Scholar] [CrossRef]
- Graber, J.J.; Cobbs, C.S.; Olson, J.J. Congress of neurological surgeons systematic review and evidence-based guidelines on the use of stereotactic radiosurgery in the treatment of adults with metastatic brain tumors. Neurosurgery 2019, 84, E168–E170. [Google Scholar] [CrossRef] [PubMed]
- Swati, Z.N.K.; Zhao, Q.; Kabir, M.; Ali, F.; Ali, Z.; Ahmed, S.; Lu, J. Brain tumor classification for MR images using transfer learning and fine-tuning. Comput. Med Imaging Graph. 2019, 75, 34–46. [Google Scholar] [CrossRef] [PubMed]
- Bodapati, J.D.; Shaik, N.S.; Naralasetti, V.; Mundukur, N.B. Joint training of two-channel deep neural network for brain tumor classification. Signal Image Video Process. 2021, 15, 753–760. [Google Scholar] [CrossRef]
- Society, A.C. 2023. Available online: http://www.cancer.net/cancer-types/brain-tumor/statistics/ (accessed on 23 May 2024).
- Elhoseny, M.; Bian, G.B.; Lakshmanaprabu, S.; Shankar, K.; Singh, A.K.; Wu, W. Effective features to classify ovarian cancer data in internet of medical things. Comput. Networks 2019, 159, 147–156. [Google Scholar] [CrossRef]
- Din, I.U.; Guizani, M.; Kim, B.S.; Hassan, S.; Khan, M.K. Trust management techniques for the Internet of Things: A survey. IEEE Access 2018, 7, 29763–29787. [Google Scholar] [CrossRef]
- Huang, C.; Wang, J.; Wang, S.H.; Zhang, Y.D. Applicable artificial intelligence for brain disease: A survey. Neurocomputing 2022, 504, 223–239. [Google Scholar] [CrossRef]
- Hammad, M.; ElAffendi, M.; Ateya, A.A.; Abd El-Latif, A.A. Efficient brain tumor detection with lightweight end-to-end deep learning model. Cancers 2023, 15, 2837. [Google Scholar] [CrossRef] [PubMed]
- Meera, R.; Anandhan, P. A review on automatic detection of brain tumor using computer aided diagnosis system through MRI. EAI Endorsed Trans. Energy Web 2018, 5, e17. [Google Scholar]
- Zulfiqar, F.; Bajwa, U.I.; Mehmood, Y. Multi-class classification of brain tumor types from MR images using EfficientNets. Biomed. Signal Process. Control 2023, 84, 104777. [Google Scholar] [CrossRef]
- Asiri, A.A.; Iqbal, A.; Ferzund, J.; Ali, T.; Aamir, M.; Alshamrani, K.A.; Alshamrani, H.A.; Alqahtani, F.F.; Irfan, M.; Alshehri, A.H. A Novel Hybrid Machine Learning Approach for Classification of Brain Tumor Images. Comput. Mater. Contin. 2022, 73. [Google Scholar] [CrossRef]
- Gao, Z.; Yang, Y.; Khosravi, M.R.; Wan, S. Class consistent and joint group sparse representation model for image classification in internet of medical things. Comput. Commun. 2021, 166, 57–65. [Google Scholar] [CrossRef]
- Díaz-Pernas, F.J.; Martínez-Zarzuela, M.; Antón-Rodríguez, M.; González-Ortega, D. A deep learning approach for brain tumor classification and segmentation using a multiscale convolutional neural network. Healthcare 2021, 9, 153. [Google Scholar] [CrossRef] [PubMed]
- Brindha, P.G.; Kavinraj, M.; Manivasakam, P.; Prasanth, P. Brain tumor detection from MRI images using deep learning techniques. In Proceedings of the IOP Conference Series: Materials Science and Engineering, Sanya, China, 12–14 November 2021; IOP Publishing: Bristol, UK, 2021; Volume 1055, p. 012115. [Google Scholar]
- Kakarla, J.; Isunuri, B.V.; Doppalapudi, K.S.; Bylapudi, K.S.R. Three-class classification of brain magnetic resonance images using average-pooling convolutional neural network. Int. J. Imaging Syst. Technol. 2021, 31, 1731–1740. [Google Scholar] [CrossRef]
- Ganesan, M.; Sivakumar, N.; Thirumaran, M. Internet of medical things with cloud-based e-health services for brain tumour detection model using deep convolution neural network. Electron. Gov. Int. J. 2020, 16, 69–83. [Google Scholar] [CrossRef]
- Kibriya, H.; Masood, M.; Nawaz, M.; Rafique, R.; Rehman, S. Multiclass brain tumor classification using convolutional neural network and support vector machine. In Proceedings of the 2021 Mohammad Ali Jinnah University International Conference on Computing (MAJICC), Karachi, Pakistan, 15–17 July 2021; IEEE: New York, NY, USA, 2021; pp. 1–4. [Google Scholar]
- Kibriya, H.; Masood, M.; Nawaz, M.; Nazir, T. Multiclass classification of brain tumors using a novel CNN architecture. Multimed. Tools Appl. 2022, 81, 29847–29863. [Google Scholar] [CrossRef]
- Tripathi, P.C.; Bag, S. A computer-aided grading of glioma tumor using deep residual networks fusion. Comput. Methods Programs Biomed. 2022, 215, 106597. [Google Scholar] [CrossRef]
- Isunuri, B.V.; Kakarla, J. Three-class brain tumor classification from magnetic resonance images using separable convolution based neural network. Concurr. Comput. Pract. Exp. 2022, 34, e6541. [Google Scholar] [CrossRef]
- Deepak, S.; Ameer, P. Automated categorization of brain tumor from mri using cnn features and svm. J. Ambient Intell. Humaniz. Comput. 2021, 12, 8357–8369. [Google Scholar] [CrossRef]
- Vankdothu, R.; Hameed, M.A.; Ameen, A.; Unnisa, R. Brain image identification and classification on Internet of Medical Things in healthcare system using support value based deep neural network. Comput. Electr. Eng. 2022, 102, 108196. [Google Scholar] [CrossRef]
- Kamil, O.A.; Al-Shammari, S.W. IoT framework for brain tumor classification using optimized CNN-MRFO model. Am. J. Bioinform. Res. 2021, 11, 32–37. [Google Scholar]
- Rajathi, G.I.; Kumar, R.R.; Ravikumar, D.; Joel, T.; Kadry, S.; Jeong, C.W.; Nam, Y. Brain Tumor Diagnosis Using Sparrow Search Algorithm Based Deep Learning Model. Comput. Syst. Sci. Eng. 2023, 44, 1793–1806. [Google Scholar] [CrossRef]
- Kurdi, S.Z.; Ali, M.H.; Jaber, M.M.; Saba, T.; Rehman, A.; Damaševičius, R. Brain tumor classification using meta-heuristic optimized convolutional neural networks. J. Pers. Med. 2023, 13, 181. [Google Scholar] [CrossRef] [PubMed]
- Biswas, A.; Islam, M.S. A Hybrid Deep CNN-SVM Approach for Brain Tumor Classification. J. Inf. Syst. Eng. Bus. Intell. 2023, 9. [Google Scholar] [CrossRef]
- Hamada, A. Br35H :: Brain Tumor Detection 2020. 2020. Available online: https://www.kaggle.com/datasets/ahmedhamada0/brain-tumor-detection (accessed on 12 July 2024).
- Filatov, D.; Ahmad Hassan Yar, G.N. Brain tumor diagnosis and classification via pre-trained convolutional neural networks. medRxiv 2022, arXiv:2208.00768. [Google Scholar]
- Gómez-Guzmán, M.A.; Jiménez-Beristaín, L.; García-Guerrero, E.E.; López-Bonilla, O.R.; Tamayo-Perez, U.J.; Esqueda-Elizondo, J.J.; Palomino-Vizcaino, K.; Inzunza-González, E. Classifying brain tumors on magnetic resonance imaging by using convolutional neural networks. Electronics 2023, 12, 955. [Google Scholar] [CrossRef]
- Rajkumar, S.; Karthick, K.; Selvanathan, N.; Saravanan, U.B.; Murali, M.; Dhiyanesh, B. Brain Tumor Detection Using Deep Learning Neural Network for Medical Internet of Things Applications. In Proceedings of the 2021 6th International Conference on Communication and Electronics Systems (ICCES), Coimbatore, India, 8–10 July 2021; pp. 1–8. [Google Scholar]
| Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).