MURA-EfficientNetV2-Fracture-Detection

MURA Bone Fracture Detection Model

Model Description

This is a custom-trained EfficientNetV2 model designed to detect bone fractures in musculoskeletal radiographs. It was trained using transfer learning on the MURA (Musculoskeletal Radiographs) dataset.

• Architecture: EfficientNetV2 (Base) + Custom GlobalAveragePooling & Dense Head
• Task: Binary Classification (fractured vs. not_fractured)
• Framework: TensorFlow / Keras
• Input Resolution: 224x224 RGB images

Usage

You can load this model directly using TensorFlow/Keras to run inference on new X-ray images:

import tensorflow as tf
from tensorflow.keras.preprocessing import image
import numpy as np

# Load the model
model = tf.keras.models.load_model('MURA_EfficientNetV2L.h5')

# Preprocess image
img = image.load_img('path_to_xray.jpg', target_size=(224, 224))
img_array = image.img_to_array(img)
img_array = np.expand_dims(img_array, axis=0)
img_array = tf.keras.applications.efficientnet_v2.preprocess_input(img_array)

# Predict
prediction = model.predict(img_array)
print(f"Fracture Probability: {prediction[0][0]:.2%}")