Research Project

About NeuralScan.AI

An AI-powered brain tumor classification system built with state-of-the-art deep learning — achieving 95.75% test accuracy. Trained to detect Glioma, Meningioma, Pituitary Tumors, and identify Healthy Scans from brain MRI images.

Why It Matters

Brain tumors are among the most devastating diagnoses. Early and accurate detection from MRI scans can significantly improve patient outcomes and treatment planning.

EfficientNet-V2-S Advantage

With ~21M parameters and Fused-MBConv blocks, EfficientNet-V2-S trains faster than its predecessor while achieving superior ImageNet accuracy. Our fine-tuned model reaches 97% validation accuracy.

Advanced Training Techniques

Trained with Mixup augmentation (α=0.4), AdamW optimizer with layer-wise learning rates, CosineAnnealingWarmRestarts scheduler, and label smoothing (0.1) — achieving 95.75% on the test set.

End-to-End Pipeline

From image preprocessing (224×224 resize + ImageNet-1K normalization) through inference to Softmax probability output — a complete, production-ready classification system.

Model Overview

Key Metrics at a Glance

Everything you need to know about the model powering NeuralScan.AI

Architecture
EfficientNet-V2-S
Pre-trained on ImageNet-1K
Parameters
~21M (20,967,252)
91.3% trainable
Test Accuracy
95.75%
800 test samples
Val Accuracy
97.00%
Best at epoch 49
Input Size
224 × 224
RGB normalized
Dataset
5,712 train / 800 test
4 tumor classes
JP
The Developer

Jay Patel

Creating intelligent digital solutions. Specializing in Machine Learning & AI, building dynamic Full Stack web applications (MERN, Next.js), and architecting robust .NET systems. Bridging the gap between complex data and intuitive user experiences.

Skills

Technical Arsenal

Technologies I use to build robust solutions

Python
PyTorch
TensorFlow
Scikit-learn
Next.js
React
Node.js
TypeScript
MongoDB
ASP.NET
Flutter
Docker
Git
Azure
Tailwind CSS
Framer Motion

Let's Connect & Collaborate

Whether you have a question, a project idea, or just want to say hi — I'm always open to discussing new opportunities.