You can deploy this model and start earning money today!

palmyz0/
brain-tumor-resnet50-finetune-with-optuna

Name: brain-tumor-resnet50-finetune-with-optuna
Availability: InStock

Views

No views yet

🧠 Brain Tumor Classification with Optimized ResNet50

This model is a fine-tuned ResNet50 architecture designed to classify brain MRI scans into 4 categories. The model has been optimized using Optuna for hyperparameter tuning to achieve high accuracy and robustness on medical imaging.

🎯 Model Performance

Accuracy: ~95.38% (on Test set)
Model Architecture: ResNet50 (Pre-trained on ImageNet)
Optimization Strategy: Bayesian Optimization via Optuna
Input Resolution: 224x224 pixels

📂 Class Labels

The model predicts one of the following 4 classes:

glioma
meningioma
notumor (Healthy)
pituitary

🛠️ How to Use (Python Code)

You can use this model directly with the Hugging Face huggingface_hub library or raw PyTorch.

Important: Images must be resized to 224x224 and normalized using ImageNet statistics before inference.

python

import torch
import torch.nn as nn
from torchvision import transforms, models
from PIL import Image
from huggingface_hub import hf_hub_download

# 1. Define Model Architecture (ResNet50)
class AdvancedResNetClassifier(nn.Module):
    def __init__(self, num_classes=4):
        super(AdvancedResNetClassifier, self).__init__()
        # Load ResNet50 architecture (weights=None because we load our own state_dict)
        self.backbone = models.resnet50(weights=None)
        
        # Modify the fully connected layer to match the training configuration
        num_ftrs = self.backbone.fc.in_features
        self.backbone.fc = nn.Sequential(
            nn.Linear(num_ftrs, 512),
            nn.ReLU(),
            nn.Dropout(0.48), 
            nn.Linear(512, num_classes)
        )
        
    def forward(self, x):
        return self.backbone(x)

# 2. Configuration
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
REPO_ID = "Suphawit/brain-tumor-resnet50-finetune-with-optuna" 
FILENAME = "model.pth"

# 3. Load Model
model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)
model = AdvancedResNetClassifier(num_classes=4)
model.load_state_dict(torch.load(model_path, map_location=DEVICE))
model.to(DEVICE)
model.eval()

# 4. Predict
# Define Transform
transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])

# Load Image
img_path = "path_to_your_mri_scan.jpg"
img = Image.open(img_path).convert('RGB')
input_tensor = transform(img).unsqueeze(0).to(DEVICE)

# Inference
with torch.no_grad():
    output = model(input_tensor)
    probs = torch.nn.functional.softmax(output, dim=1)
    confidence, prediction = torch.max(probs, 1)
    
    class_names = ['glioma', 'meningioma', 'notumor', 'pituitary']
    predicted_class = class_names[prediction.item()]
    
    print(f"Predicted Class: {predicted_class}")
    print(f"Confidence: {confidence.item()*100:.2f}%")

"Try out the live model demo here: https://huggingface.co/spaces/palmyz0/brain-tumor-detection-demo"

👤 Author

Suphawit MeeSak

Junior AI Engineer / Junior Data Scientist

🔗 LinkedIn: Suphawit MeeSak
📧 Email: Suphawit11@icloud.com

palmyz0/brain-tumor-resnet50-finetune-with-optuna