Introducing !!!

NeuroKnightMRI

where cutting-edge machine learning prowess meets the battlefield against brain tumors, revolutionizing detection and classification with unparalleled precision.

5 Machine Learning Model Architectures

Pretty accurate

100% free & open-source

"Workflow of our Machine Learning Web App"

We have these 4 classes.

Glioma Tumor

Meningioma Tumor

No Tumor

Pituitary Tumor

The model provides reliable classification of these 3 broad tumor categories, it does not differentiate between subclasses or types within each category. This streamlined approach allows for efficient and effective screening and triaging of patients, enabling prompt medical intervention and treatment planning.

Model Architectures

We successfully trained five distinct model architectures using a meticulously partitioned dataset, comprising Training, Testing, and Validation subsets.

AlexNet
AlexNet is a deep convolutional neural network (CNN) architecture that gained significant attention after winning the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) in 2012. It consists of five convolutional layers followed by max-pooling layers, with three fully connected layers at the end.
GoogleNet (Inception)
GoogleNet, also known as Inception, is a CNN architecture developed by Google researchers. It introduced the concept of inception modules which are convolutional modules with multiple parallel convolutional operations at different spatial scales. GoogleNet won the ILSVRC in 2014 and demonstrated state-of-the-art performance on various computer vision tasks.
Vision Transformers
The Vision Transformer (ViT) is a recent approach to computer vision tasks that applies the transformer architecture, originally developed for natural language processing, to image classification tasks. ViT divides the input image into patches, which are flattened and processed by transformer blocks. By leveraging self-attention mechanisms, ViT achieves strong performance on image classification tasks without relying on CNNs.
ResNet
ResNet is a deep CNN architecture. ResNet architectures consist of residual blocks, which enable the training of extremely deep networks (hundreds of layers) while mitigating the vanishing gradient problem.
VGG-19
VGG19 is a deep convolutional neural network (CNN) architecture, an extension of the VGG16 model, with 19 layers. Its architecture comprises 16 convolutional layers and 3 fully connected layers, hence the name.

Disclaimer

Please note that while our models have been rigorously trained and tested on specific datasets for the task of brain tumor classification, they may not perform optimally on images outside of their intended domain. For instance, if a user uploads an image of a flower instead of a brain tumor MRI scan, the model's output may not be accurate or relevant. We advise users to ensure that uploaded images align with the intended use case of the selected model for optimal results. Additionally, our models have not been specifically trained to handle special cases or edge scenarios. We encourage users to exercise discretion and interpret results accordingly.

Had a lot of fun doing my 8th semester Major Project :)