app.py

import streamlit as st
import requests
import os
import io
import shutil
from PIL import Image
from pathlib import Path

# Define API endpoint
API_URL = "https://nagur-shareef-shaik-retinal-health-diagnostics.hf.space/predict"

###### Sidebar ######

# Custom CSS to make the sidebar fixed and adjust width
st.markdown(
    """
    <style>
        [data-testid="stSidebar"] {
            width: 750px !important;  /* Adjust width as needed */
            background-color: #f8f9fa; /* Light grey background */
        }
        [data-testid="stSidebarNav"] {
            width: 750px;
        }
    </style>
    """,
    unsafe_allow_html=True,
)

st.sidebar.header("Intelligent Application for Retinal Diseases Diagnosis")
st.sidebar.image("images/rhd_img_1.jpeg", width=1250)

st.sidebar.markdown("""
### Get AI-Powered Diagnosis Report  

1. **Enter Patient Details**  
   - Name, Age, Gender, Contact Information, Address  
   - Select Diagnosis Task (**General, Retinopathy, DME, Cataracts**)
        - **General:** Diagnoses Diabetic Retinopathy, Cataracts, Glaucoma, & Normal conditions from retinal fundus images.
        - **Retinopathy:** Detects 7 retinopathy severity grades (No DR, Mild NPDR, Moderate NPDR, Severe NPDR, Very Severe NPDR, PDR, and Advanced PDR) from retinal scans.
        - **DME:** Identifies the presence of Diabetic Macular Edema (DME) from retinal scans.
        - **Cataracts:** Diagnoses Cataracts from digital eye images.
   - Add Medical Notes (if any)  

2. **Upload Retinal Scan**  
   - Accepts **JPG, JPEG, PNG** formats  
   - Ensure clear, high-quality images  

3. **Submit for AI Analysis**  
   - Click **"Submit Request"** to send details to AI model  
   - AI processes the image and generates a diagnosis  

4. **View Results Instantly**  
   - AI-powered diagnosis displayed with insights  
   - Download or share the report if needed  

""")

# Create two columns
col1, col2 = st.columns([1, 3])  # Adjust the width ratio as needed

with col1:
    st.image("images/rhd-logo.png", width=300)  # Adjust width as needed

with col2:
    st.title("Retinal Health Diagnostics")

# User input fields
task = st.selectbox("Select Task", ["general", "retinopathy", "dme", "cataracts"], index=0)
name = st.text_input("Name")
a, b, c = st.columns([1, 1, 1])
with a:
    age = st.text_input("Age")
with b:
    gender = st.selectbox("Gender", ["Male", "Female", "Other"], index=0)
with c:
    phone = st.text_input("Phone")
email = st.text_input("Email")
address = st.text_area("Address")
med_notes = st.text_area("Medical Notes")

# Image upload
uploaded_file = st.file_uploader("Upload Retinal Image", type=["jpg", "jpeg", "png"])

if uploaded_file:
    # Convert file to bytes
    files = {"file": (uploaded_file.name, uploaded_file, uploaded_file.type)}

# Submit button
if st.button("Submit Request"):
    if not uploaded_file:
        st.error("Please upload an image.")
    else:
        # Create request payload
        payload = {
            "task": task,
            "name": name,
            "age": age,
            "gender": gender,
            "address": address,
            "phone": phone,
            "email": email,
            "med_notes": med_notes,
        }
        
        try:
            response = requests.post(API_URL, files=files, data=payload)
            if response.status_code == 200:
                st.success("Diagnosis Complete!")

                # Extract image from response
                img_bytes = response.content  # Get image as bytes

                # Convert bytes to PIL Image
                image = Image.open(io.BytesIO(img_bytes))

                # Get response metadata
                predicted_class = response.headers.get('Predicted-Class')
                confidence = response.headers.get('Confidence')

                st.markdown(
                    f"""
                    <div style="padding: 15px; border-radius: 10px; background-color: #f0f2f6; text-align: center;">
                        <h3 style="color: #4CAF50;">Diagnosis Result</h3>
                        <p style="font-size: 18px;"><strong>Findings:</strong> <span style="color: #FF5722;">{predicted_class}</span></p>
                    </div>
                    """,
                    unsafe_allow_html=True
                )
                st.image(image)

                # st.json(response.json())
            else:
                st.error(f"Error {response.status_code}: {response.text}")
        except requests.exceptions.RequestException as e:
            st.error(f"Request failed: {e}")