|
|
--- |
|
|
language: |
|
|
- en |
|
|
pipeline_tag: image-classification |
|
|
tags: |
|
|
- stroke |
|
|
- medical |
|
|
- classification |
|
|
- ischemic |
|
|
- hemorrhagic |
|
|
--- |
|
|
|
|
|
# Stroke Classification ResNet Model |
|
|
|
|
|
This is a Keras model for classifying MRI images into: |
|
|
- Hemorrhagic Stroke |
|
|
- Ischemic Stroke |
|
|
- No Stroke |
|
|
|
|
|
The model is a fine-tuned ResNet50. |
|
|
|
|
|
## How to use this model (example in Python) |
|
|
|
|
|
```python |
|
|
import tensorflow as tf |
|
|
from PIL import Image |
|
|
import numpy as np |
|
|
import os # Added for os.path.join |
|
|
|
|
|
# Load the model |
|
|
# Ensure the model file 'stroke_classification_model.h5' is in the same directory |
|
|
# or provide the full path. |
|
|
model = tf.keras.models.load_model('stroke_classification_model.h5') |
|
|
|
|
|
# Define your class names (must match how your model was trained) |
|
|
CLASS_NAMES = ['hemorrhagic_stroke', 'ischemic_stroke', 'no_stroke'] # Automatically populated from your Colab session |
|
|
|
|
|
def preprocess_image_for_prediction(image_path, target_size=(224, 224), pixel_threshold=40): |
|
|
img = Image.open(image_path).convert("L") |
|
|
original_width, original_height = img.size |
|
|
data = np.array(img) |
|
|
rows_with_content = np.any(data > pixel_threshold, axis=1) |
|
|
cols_with_content = np.any(data > pixel_threshold, axis=0) |
|
|
try: |
|
|
min_row = np.where(rows_with_content)[0][0] |
|
|
max_row = np.where(rows_with_content)[0][-1] |
|
|
min_col = np.where(cols_with_content)[0][0] |
|
|
max_col = np.where(cols_with_content)[0][-1] |
|
|
except IndexError: |
|
|
cropped_img = img |
|
|
else: |
|
|
buffer = 5 |
|
|
min_row = max(0, min_row - buffer) |
|
|
max_row = min(original_height - 1, max_row + buffer) |
|
|
min_col = max(0, min_col - buffer) |
|
|
max_col = min(original_width - 1, max_col + buffer) |
|
|
cropped_img = img.crop((min_col, min_row, max_col + 1, max_row + 1)) |
|
|
processed_img = cropped_img.resize(target_size, Image.LANCZOS) |
|
|
if processed_img.mode == 'L': |
|
|
processed_img = processed_img.convert('RGB') |
|
|
img_array = tf.keras.utils.img_to_array(processed_img) |
|
|
img_array = tf.expand_dims(img_array, 0) |
|
|
return img_array |
|
|
|
|
|
# Example usage: |
|
|
# image_path = "path/to/your/new_mri_image.jpg" |
|
|
# preprocessed_img = preprocess_image_for_prediction(image_path) |
|
|
# if preprocessed_img is not None: |
|
|
# predictions = model.predict(preprocessed_img) |
|
|
# predicted_class_index = np.argmax(predictions[0]) |
|
|
# predicted_class_name = CLASS_NAMES[predicted_class_index] |
|
|
# confidence = np.max(predictions[0]) * 100 |
|
|
# print(f"Predicted: {predicted_class_name} with {confidence:.2f}% confidence") |
|
|
``` |
