SABViT: A Pilot Feasibility Study of a Self-Attention-Based Vision Transformer for Binary Brain Tumor Detection in MRI

Brain tumor segmentation with deep convolutional symmetric neural network

Detection of brain tumors in MRI images is the first step in brain cancer diagnosis. The accuracy of the diagnosis depends highly on the expertise of radiologists. Therefore, automated diagnosis of brain cancer from MRI is receiving a large amount of attention. Also, MRI tumor detection is usually followed by a biopsy (an invasive procedure), which is a medical procedure for brain tumor classification. It is of high importance to devise automated methods to aid radiologists in brain cancer tumor diagnosis without resorting to invasive procedures. Convolutional neural network (CNN) is deemed to be one of the best machine learning algorithms to achieve high-accuracy results in tumor identification and classification. In this paper, a CNN-based technique for brain tumor classification has been developed. The proposed CNN can distinguish between normal (no-cancer), astrocytoma tumors, gliomatosis cerebri tumors, and glioblastoma tumors. The implemented CNN was tested on MRI images that underwent a motion-correction procedure. The CNN was evaluated using two performance measurement procedures. The first one is a k-fold cross-validation testing method, in which we tested the dataset using k = 8, 10, 12, and 14. The best accuracy for this procedure was 96.26% when k = 10. To overcome the over-fitting problem that could be occurred in the k-fold testing method, we used a hold-out testing method as a second evaluation procedure. The results of this procedure succeeded in attaining 97.8% accuracy, with a specificity of 99.2% and a sensitivity of 97.32%. With this high accuracy, the developed CNN architecture could be considered an effective automated diagnosis method for the classification of brain tumors from MRI images.

In the human visual system, visible objects are recognized by features, which can be classified into local features that are based on their simple components (i.e., line segment, angle, color, etc.) and global features that are based on the whole objects (i.e., connectivity, number of holes, etc.). Over the past half century, anatomical, physiological, behavioral and computational studies of the visual systems have led to a generally accepted model of vision, which starts at processing local features in the early stages of the visual pathways, followed by integrating them to global features in the later stages of the visual pathways. However, this popular local-to-global model has been challenged by a set of experiments showing that the visual systems in humans, non-human primates and honey bees are more sensitive to global features than local features. These “global-first” studies further motivated developing new paradigms and approaches to understand human vision and build new vision models. In this study, we started a new series of experiments that examine how two representative pre-trained Convolutional Neural Networks (CNN) (AlexNet and VGG-19) process local and global features. The CNNs were trained to classify geometric shapes into two categories based on local features (e.g., triangle, square and circle) or a global feature (e.g., having a hole). In contrast to the biological visual systems, the CNNs were more effective at classifying images based on local features than the global feature. We further showed that adding distractors greatly lowered the performance of the CNNs, again different from the biological visual systems. Ongoing studies will extend these analyses to other geometrical invariants and internal representations of the CNNs. The overarching goal is to use the powerful CNNs as a tool to gain insights into the biological visual systems, including that of humans and non-human primates.

Focus issue: Artificial intelligence in medical physics.

Abstract— Abstract — Brain tumors are among the most life-threatening and complex conditions, requiring early and accurate detection to ensure effective treatment and improved survival rates. Manual interpretation of MRI scans for diagnosis is a labor-intensive process, demanding significant expertise and carrying the risk of human inaccuracies. This research proposes a deep learning-based framework using a Deep Convolutional Neural Network (DCNN) specifically instantiated with the pretrained ResNet50 model for the automated detection and classification of brain tumors from MRI images. MRI scans are processed into a classification system that identifies four conditions: glioma, meningioma, pituitary tumor, or a healthy (no tumor) state. We prepare the dataset for this task using normalization, resizing, and augmentation to improve model robustness and reduce the risk of overfitting. The DCNN specifically with pretrained ResNet50 architecture is designed with multiple convolutional, pooling, and dense layers to extract complex features and learn spatial hierarchies within the data. The model is trained using categorical cross-entropy loss and optimized via the Adam optimizer to achieve high classification accuracy. Extensive validation and testing show that the model achieves reliable performance with high precision and recall across all tumor types. The trained model is further integrated into a user-friendly web interface using the Flask framework, enabling real-time prediction from uploaded MRI scans. This system provides an accessible and effective diagnostic tool, especially beneficial in resource-constrained settings, and contributes significantly to the field of medical imaging and intelligent healthcare solutions. Keywords—Brain tumor detection, Deep Convolutional Neural Network, DCNN, ResNet50, medical imaging, MRI classification, Flask deployment.

Brain is the regulatory unit in human body. It controls the functions such as memory, vision, hearing, knowledge, personality, problem solving, etc. The main reason for brain tumor is the abandoned progress of brain cells. Many health organizations have recognized brain tumor as the second foremost dispute that causes many human deaths all around the world. Identification of brain tumor at a premature stage offers opportunity of effective medical treatment. Use of Magnetic Resonance Imaging images have been recognized as more detailed and more consistent images when compared to Computed Tomography images. In this project we are predicting the brain tumor using machine learning algorithms. there are various techniques to detect brain tumor or neoplasms. The most competent and effective algorithm that we will be using in this project is CNN (Convolutional Neural Network).It is a Deep Learning Algorithm that is Used for Image Processing Starting with Pre-processing brain images then, segmenting them, feature extraction of unwanted noisy and blurry part, clustering and detection of the tumor are the methodologies of our Project. Key Word: Brain Tumor Prediction, MRI Images, CNN, Image Processing and Deep Learning.

Brain tumors pose significant threats within neurological disorders, demanding accurate classification for effective diagnosis and treatment. This study explores brain tumor classification employing Classical Local Binary Patterns (CLBP) and Convolutional Neural Networks (CNN), alongside texture feature extraction from MRI images using classical LBP and HOG (Histogram of Oriented Gradients). These methods adeptly capture both local and global texture patterns crucial for tumor identification. Our proposed framework encompasses three pivotal steps: image pre-processing, feature extraction via CLBP, and classification utilizing CNN. Evaluation on a publicly available brain tumor dataset showcased an impressive 95.6% accuracy in tumor classification, affirming the efficacy of the CLBP+CNN approach. This method bears promising implications for enhancing clinical diagnosis and treatment planning. Furthermore, we propose future extensions including CLBPs such as DLBP and LBP. DLBP introduces a parameter, ’D’, dictating pixel distance, while LBP varies pixel values across specified ranges. Additionally, tumor classification was explored employing ANN, AlDE, and LDA classification methods, with future prospects of incorporating DLBP, LBP, and CLBP extractions from MRI images within the dataset.

Non-automated segmentation of brain tumors from the MRI images is a tedious task, which may sometimes produce inaccurate results. Therefore, the accuracy and reliability of brain tumor segmentation is also key to decision-making, treatment design, and analysis of treatment results. The automated brain tumor segmentation strategy uses manually developed parameters. As with convolutional neural networks, training requires excessive amount of annotated data that is generally challenging to retrieve in the medical field. The prior demand is the design of a task adapted along with robust feature selection and representation. This seems mandatory for effective and efficient segmentation of complicated brain tumors, which in turn is possible with two-channel Convolutional Neural Networks (CNN). Hence, two-channel CNN is adapted for the segmentation as it provided promising results for similar research. The two-channel CNN model contributes much in lessening jitter and over fitting parameters. Conclusively, a two-way cascade design to CNN is introduced, where the base CNN output is processed as an auxiliary source and combined at the final level. When testing the model in the BRATS2013 and BRATS2020 knowledge sets, the CNN integrated package in the 2-channel design was not hidden. Compared to the progressive method described above, the overall performance has been improved and the process quality remains attractive.

Brain tumors are problematic, leading to the highest degree of very low life span. Recovery planning therefore represents a critical step toward improving patients’ quality of life. Different imaging methods Computer tomography, magnetic resonance imaging (CT) (MRI) and ultrasound scanning are commonly recommended for brain, lung, liver, breast, and prostate tumor control. MRI images are used in this research particularly to recognize tumors inside the brain. However, in a given time the enormous amount of MRI scanned data prevent the manual classification into tumor and non-tumor. But for a limited number of images, there is some constraint, i.e., to giving precise quantitative measurements. Accordingly, to reduce mortality rates in humans, it is important to have a trustworthy and automatic classification scheme. An extremely challenging task is the automatic classification of brain tumors into the enormous spatial and auxiliary alteration of the area containing brain tumors. In proposed research, automatic brain tumor identification is done using Convolution Neural Networks (CNN). The architecture’s deeper nature is accomplished using small kernels. Neuron weight is given as low as this. Results from experiments show CNN achieve a low complexity rate of 97.5% accuracy relative to all other approaches to state of the art.KeywordsNeural networksBrain tumorCNNClassification

Brain tumors affect 7 of the 100,000 population and are the 10th leading cause of death in Indonesia. The process of diagnosing brain tumors is done using MRI images which are manually analyzed by radiologists. However, the small number of radiologists with uneven distribution means that the determination of the location and size of the tumor is severely hampered. The solution to this problem is to create technology that can accurately determine the location and size of tumors, one of which is the Deep Convolutional Neural Network. This paper has simulated the location and size of brain tumor segmentation based on MRI images using Deep CNN U-Net architecture and modified it with ResNet (ResU-Net) to make it easier for radiologists to examine the brain accurately. The author has conducted trials with U-Net and ResU-Net architectures to receive the location and size of brain tumors with accurate results from training, validation, and segmentation as measured by the Tanimoto Index, Dice Coefficient, and Tversky Index as evaluation metrics. The author also analyzes the performance of each architecture based on the number of layers, parameters, and training time. Based on these simulation scenarios, the maximum accuracy of the segmentation for U-Net is 88.35% and ResU-Net is 90.04%.

Brain tumor is a serious life-threatening disease. The early diagnosis of brain tumor is very important for the proper treatment. The removal of brain tumor requires implementation of an invasive procedure relying on biopsy and spinal tap method. There are more than 120 types of brain tumor. Therefore, the identification of a particular brain tumor is a very difficult task because of the small size and complex structure of the brain. A lot of research has been conducted to classify brain tumors into benign and malignant tumors but there is very limited research work available to classify brain tumor by its types. Among the various types of brain tumors is Glioblastoma which is very aggressive and life threatening. The main purpose of the proposed approach is accurate classification of brain tumor contained images followed by localization on those images. The proposed approach uses Convolutional Neural Networks (CNNs) for classification and Region-based Convolutional Neural Networks (R-CNN) and Faster Region-based Convolutional Neural Networks (Faster R-CNN) to localize tumor regions in MRI images after classification into three types, i.e. Glioblastoma, Dendroglioma and Astrocytoma. Moreover, the proposed approach involves the use of transfer learning techniques, namely feature extraction and fine tuning and achieves classification rate of 95.40% and detection rates of 89.10%.

This research highlights that great efforts are being made to improve the accuracy and validity of brain tumor classification using MRI images. Working at multiple resolutions helps to preserve the important features of MRI while reducing irrelevant noise. Below, we briefly discuss the method and its benefits. In this study a non-local means pre-processing technique combining deep learning models ResNet50 and residual connections with CNN allows for better understanding and classification of complex tumor images by utilizing hierarchical and spatial feature extraction. The pre-training and fine-tuning stages allow for better generalization to new unseen material. Fuzzy Clustering Segmentation Method What is interesting about this method is that it combines deep learning methods (CNN and ResNet50) with a classical image processing method, Non-Local Means (NLM). Diagnosing brain tumor and differentiating between tumor types. Fuzzy clustering segmentation method is an effective imaging technique for delineating tumor boundaries. This hybrid approach leverages the best of both worlds to improve tumor detection and classification, aiding in early diagnosis and treatment planning. This is key to identifying the region of interest (tumor), which is often challenging in medical image processing due to the different shapes, sizes, and appearances of tumor. ResNet50-CNN Classification This work aims to address the challenges in brain tumor detection and classification using advanced techniques such as non-local means pre-processing, fuzzy clustering segmentation method, and ResNet50 Convolutional Neural Network (CNN) classification. The reported results show that the proposed method outperforms state-of-the-art methods in various performance metrics such as accuracy, sensitivity, F1 score, specificity, and error rate, indicating that the proposed method is highly effective in terms of accuracy.

This study aims to classify brain MRI images into several types of brain tumors using the Convolutional Neural Network (CNN) approach with transfer learning. This method has the advantage of processing complex images in a shorter time than conventional CNN approaches. In this study, the data used was a public database from Kaggle, which consisted of four categories: glioma, meningioma, no tumor, and pituitary. Before entering the transfer learning process, data augmentation is carried out on the training data. Four pre-trained CNN models were used: VGG19, ResNet50, InceptionV3, and DenseNet121. The four models compared their ability to classify MRI images with several evaluation metrics: accuracy, precision, recall, and F1 score. The results of the performance comparison of the four pre-trained models show that the ResNet50 is the best model, with an accuracy of 98%. Meanwhile, VGG19, DenseNet121, and InceptionV3 produce 97%, 96%, and 95% accuracy, respectively. The ResNet50 architecture demonstrated superior performance in brain tumor classification, achieving 98% accuracy. It can be attributed to its residual learning structure, which efficiently manages complex MRI features. Further research should concentrate on larger, more diverse datasets and advanced preprocessing techniques to enhance model generalizability.

Rapid and uncontrolled cellular proliferation is what distinguishes a brain tumor. Unfortunately, brain tumors cannot be prevented other than via surgery. As predicted, deep learning may help diagnose and cure brain cancers. The segmentation approach has been widely studied for brain tumor removal. This uses the segmentation approach, one of the most advanced methods for object detection and categorization. To efficiently assess the tumor's size, an accurate and automated brain tumor segmentation approach is needed. We present a fully automated brain tumor separation method for imaging investigations. The approach has been developed with convolutional neural networks. The Multimodal Brain Tumor Image Segmentation (BRATS) datasets tested our strategy. This result suggests that DL should investigate heterogeneous MRI image segmentation to improve brain tumor segmentation accuracy and efficacy. This study may lead to more accurate medical diagnoses and treatments. Researchers in this study also found a way to automatically find cancerous tumours by using the Grey Level Co-Occurrence Matrix (GLCM) and discrete wavelet transform (DWT) to find features in MRI images. They then used a CNN to guess the final prognosis. The preceding section details this technique. When compared to the other algorithm, the CNN method uses computer resources better.

Diagnosis, detection and classification of tumors, in the brain MRI images, are important because misdiagnosis can lead to death. This paper proposes a method that can diagnose brain tumors in the MRI images and classify them into 5 categories using a Convolutional Neural Network (CNN). The proposed network uses a Convolutional Auto-Encoder Neural Network (CANN) to extract and learn deep features of input images. Extracted deep features from each level are combined to make desirable features and improve results. To classify brain tumor into three categories (Meningioma, Glioma, and Pituitary) the proposed method was applied on Cheng dataset and has reached a considerable performance accuracy of 99.3%. To diagnosis and grading Glioma tumors, the proposed method was applied on IXI and BraTS 2017 datasets, and to classify brain images into six classes including Meningioma, Pituitary, Astrocytoma, High-Grade Glioma, Low-Grade Glioma and Normal images (No tumor), the all datasets including IXI, BraTS2017, Cheng and Hazrat-e-Rassol, was used by the proposed network, and it has reached desirable performance accuracy of 99.1% and 98.5%, respectively.