KamdiO/Diabetic_Retinopathy_Class

1import numpy as np
2import pickle
3import streamlit as st
4from PIL import Image
5import os
6from fastai.vision.all import *
7
8# set page config
9st.set_page_config(
10    page_title="Diabetic Retinopathy Model Comparison",
11    layout="wide"
12)
13
14# define function to load models
15@st.cache_resource
16def load_models():
17    models = {}
18    model_files = {
19        "Model 1": "C:\Users\Kamdi\Desktop\diabeticretin_proj\models\dr_model_resnet1.pkl",
20        "Model 2":"C:\Users\Kamdi\Desktop\diabeticretin_proj\models\dr_model_resnet1.pkl",
21        "Model 3":"C:\Users\Kamdi\Desktop\diabeticretin_proj\models\dr_model_resnet1.pkl",
22        "Model 4":"C:\Users\Kamdi\Desktop\diabeticretin_proj\models\dr_model_resnet1.pkl",
23        "Model 5":"C:\Users\Kamdi\Desktop\diabeticretin_proj\models\dr_model_resnet1.pkl",
24    }
25
26    for model_name, model_path in model_files.items():
27        try:
28            models[model_name] = load_learner(model_path)
29        except Exception as e:
30            st.error(f"Error loading {model_name}: {str(e)}")
31    return models
32
33# define function to load test images
34@st.cache_data
35def load_test_images():
36    images = {}
37    image_files = {
38        "Test Image 1": "C:\Users\Kamdi\Desktop\diabeticretin_proj\test_images\image1.jpeg",
39        "Test Image 2": "C:\Users\Kamdi\Desktop\diabeticretin_proj\test_images\image2.jpeg",
40        "Test Image 3": "C:\Users\Kamdi\Desktop\diabeticretin_proj\test_images\image3.jpeg",
41        "Test Image 4": "C:\Users\Kamdi\Desktop\diabeticretin_proj\test_images\image4.jpeg",
42        "Test Image 5": "C:\Users\Kamdi\Desktop\diabeticretin_proj\test_images\image5.jpeg",
43    }
44
45    for image_name, image_path in image_files.items():
46        try:
47            images[image_name] = Image.open(image_path)
48        except Exception as e:
49            st.error(f"Error loading {image_name}: {str(e)}")
50    return images
51
52# define a function to preprocess images
53def preprocess_image(image):
54    # Convert PIL Image to fastai format
55    img = PILImage.create(np.array(image))
56    return img
57
58def get_prediction(model, image):
59    processed_image = preprocess_image(image)
60    # Get prediction and probability
61    pred, pred_idx, probs = model.predict(processed_image)
62    return pred_idx
63
64def main():
65    st.title("Diabetic Retinopathy Model Comparison")
66    
67    # Load models and images
68    try:
69        models = load_models()
70        images = load_test_images()
71    except Exception as e:
72        st.error(f"Error loading models or images: {str(e)}")
73        return
74
75    # Create two columns
76    col1, col2 = st.columns(2)
77
78    with col1:
79        st.subheader("Select Test Image")
80        # Image selection
81        selected_image_name = st.selectbox(
82            "Choose a test image:",
83            list(images.keys())
84        )
85        
86        # Display selected image
87        if selected_image_name:
88            st.image(images[selected_image_name], caption=selected_image_name, use_column_width=True)
89
90    with col2:
91        st.subheader("Model Selection and Results")
92        # Model selection
93        selected_models = st.multiselect(
94            "Select models to compare:",
95            list(models.keys())
96        )
97
98        if st.button("Run Analysis"):
99            if selected_models and selected_image_name:
100                # Create results table
101                results = []
102                for model_name in selected_models:
103                    try:
104                        prediction = get_prediction(models[model_name], images[selected_image_name])
105                        results.append({
106                            "Model": model_name,
107                            "Severity Score (0-4)": int(prediction)
108                        })
109                    except Exception as e:
110                        st.error(f"Error with {model_name}: {str(e)}")
111                
112                # Display results
113                st.table(results)
114                
115                # Display interpretation guide
116                st.markdown("""
117                **Severity Score Interpretation:**
118                - 0: No DR
119                - 1: Mild DR
120                - 2: Moderate DR
121                - 3: Severe DR
122                - 4: Proliferative DR
123                """)
124            else:
125                st.warning("Please select at least one model and an image.")
126
127if __name__ == "__main__":
128    main()