|
|
--- |
|
|
license: mit |
|
|
datasets: |
|
|
- brain-mri-dataset |
|
|
metrics: |
|
|
- accuracy |
|
|
- auc |
|
|
model-index: |
|
|
- name: CNN Brain Tumor Classifier |
|
|
results: |
|
|
- task: |
|
|
type: image-classification |
|
|
dataset: |
|
|
name: Brain MRI Dataset |
|
|
type: brain-mri-dataset |
|
|
metrics: |
|
|
- name: Accuracy |
|
|
type: accuracy |
|
|
value: 0.90 |
|
|
- name: AUC |
|
|
type: auc |
|
|
value: 0.90 |
|
|
--- |
|
|
|
|
|
# 🧠 CNN Brain Tumor Classifier |
|
|
|
|
|
## Model Description |
|
|
This repository contains a Convolutional Neural Network (CNN) built with **TensorFlow/Keras** for classifying brain MRI scans. |
|
|
The model can distinguish between three types of brain tumors and healthy scans. |
|
|
|
|
|
⚠️ **Disclaimer**: This model is provided for **educational and research purposes only**. |
|
|
It is **not a medical diagnostic tool** and should not be used in clinical practice. |
|
|
|
|
|
--- |
|
|
|
|
|
## Classes |
|
|
The model predicts one of the following four categories: |
|
|
- **Glioma** |
|
|
- **Meningioma** |
|
|
- **Pituitary tumor** |
|
|
- **No tumor** (healthy) |
|
|
|
|
|
--- |
|
|
|
|
|
## Training Details |
|
|
- **Framework**: TensorFlow / Keras |
|
|
- **Architecture**: Custom CNN |
|
|
- **Input size**: 224 × 224 RGB MRI images |
|
|
- **Loss function**: `categorical_crossentropy` |
|
|
- **Optimizer**: `Adam` |
|
|
- **Epochs**: 10 |
|
|
- **Metrics**: Accuracy, AUC |
|
|
|
|
|
--- |
|
|
|
|
|
## Dataset |
|
|
- **Source**: Public brain MRI dataset (glioma, meningioma, pituitary, no tumor) |
|
|
- **Preprocessing**: |
|
|
- Images resized to 224 × 224 |
|
|
- Normalized to [0, 1] range |
|
|
- Augmentation (rotation, flipping, zoom) applied during training |
|
|
|
|
|
--- |
|
|
|
|
|
## Evaluation Results |
|
|
| Metric | Value | |
|
|
|---------------------|-----------| |
|
|
| Training Accuracy | ~96% | |
|
|
| Validation Accuracy | ~85–90% | |
|
|
| AUC (training) | ~0.90 | |
|
|
|
|
|
*(values may vary depending on train/validation split)* |
|
|
|
|
|
--- |
|
|
|
|
|
## Usage |
|
|
|
|
|
### Installation |
|
|
```bash |
|
|
pip install tensorflow huggingface_hub |
|
|
``` |
|
|
|
|
|
|
|
|
```python |
|
|
from tensorflow.keras.models import load_model |
|
|
from huggingface_hub import hf_hub_download |
|
|
import numpy as np |
|
|
from tensorflow.keras.preprocessing import image |
|
|
|
|
|
# Download model file from Hugging Face Hub |
|
|
model_path = hf_hub_download( |
|
|
repo_id="larrikin-coder/brain-tumor-cnn", # replace with your repo |
|
|
filename="cnn_model.h5" |
|
|
) |
|
|
|
|
|
# Load model |
|
|
model = load_model(model_path) |
|
|
|
|
|
# Preprocess an image |
|
|
img = image.load_img("test_mri.jpg", target_size=(224, 224)) |
|
|
img_array = image.img_to_array(img) / 255.0 |
|
|
img_array = np.expand_dims(img_array, axis=0) |
|
|
|
|
|
# Predict |
|
|
pred = model.predict(img_array) |
|
|
class_names = ["glioma", "meningioma", "pituitary", "no_tumor"] |
|
|
print("Prediction:", class_names[np.argmax(pred)]) |
|
|
``` |
|
|
|
|
|
@misc{larrikin-coder2025, |
|
|
title={CNN Brain Tumor Classifier}, |
|
|
author={Larrikin Coder}, |
|
|
year={2025}, |
|
|
howpublished={\url{https://huggingface.co/larrikin-coder/brain-tumor-cnn}} |
|
|
} |
|
|
|
