import os
os.environ['TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD'] = '1'
import numpy as np
import torch
import gradio as gr
from PIL import Image
import fitz # PyMuPDF
import torchxrayvision as xrv
from torchvision import transforms
import re
# --- Device & Model ---
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
MODEL = xrv.models.get_model("densenet121-res224-all").to(DEVICE).eval()
LABELS = MODEL.pathologies
# --- Extended Medical Information ---
DISEASE_INFO = {
"Atelectasis": {
"description":"Collapse of part or all of a lung, reducing oxygen exchange.",
"cause":"Blocked airway, lung compression, post-surgery.",
"recommendation":"Deep breathing exercises, possibly bronchoscopy or physiotherapy.",
},
"Cardiomegaly": {
"description":"Enlargement of the heart, seen as a broad silhouette.",
"cause":"High blood pressure, valve disease, cardiomyopathy.",
"recommendation":"Echocardiogram, consult a cardiologist.",
},
"Consolidation": {
"description":"Lung region filled with liquid instead of air.",
"cause":"Often from pneumonia (bacterial or viral).",
"recommendation":"Consult a physician; likely antibiotics and follow-up chest X-ray.",
},
"Edema": {
"description":"Fluid accumulation in the lungs.",
"cause":"Heart failure, kidney issues.",
"recommendation":"Treat underlying cause, may need diuretics, consult cardiology.",
},
"Effusion": {
"description":"Fluid buildup between lung and chest wall.",
"cause":"Infection, heart failure, cancer.",
"recommendation":"May need drainage (thoracentesis), see a pulmonologist.",
},
"Emphysema": {
"description":"Damage and enlargement of lung air sacs (alveoli).",
"cause":"Mainly smoking.",
"recommendation":"Quit smoking, pulmonary rehab, inhalers.",
},
"Fibrosis": {
"description":"Scarring of lung tissue, making breathing difficult.",
"cause":"Longstanding inflammation, auto-immune disease, occupational exposure.",
"recommendation":"Pulmonologist consult, immunosuppression/antifibrotic therapy.",
},
"Fracture": {
"description":"Break/crack in a bone (commonly ribs).",
"cause":"Trauma, accident, fall.",
"recommendation":"Pain management, monitor for organ injury, orthopedic consult if severe.",
},
"Infiltration": {
"description":"Something abnormal (cells/fluid) in the lungs.",
"cause":"Most often infection, sometimes inflammation or cancer.",
"recommendation":"See physician, further tests to clarify cause.",
},
"Mass": {
"description":"Lump or growth seen on X-ray.",
"cause":"Could be benign or malignant tumor.",
"recommendation":"Consult pulmonologist or oncologist, consider CT/biopsy.",
},
"Nodule": {
"description":"Small round or oval spot in the lung.",
"cause":"Old infection, benign, or early cancer.",
"recommendation":"May need CT and follow-up scans, discuss with doctor.",
},
"Pleural_Thickening": {
"description":"Thickening of chest lining.",
"cause":"Old infection, asbestos exposure.",
"recommendation":"Pulmonology follow-up; rarely needs intervention.",
},
"Pneumonia": {
"description":"Infection causing inflammation in the lungs.",
"cause":"Bacteria, viruses, or fungus.",
"recommendation":"Antibiotics/antivirals if needed. Seek prompt medical attention.",
},
"Pneumothorax": {
"description":"Collapsed lung (air leaks into chest cavity).",
"cause":"Trauma, rupture, sometimes spontaneous.",
"recommendation":"May need emergency care to remove air; consult ER.",
},
# Add more if you want to extend—see LABELS for all possible findings
}
# --- Recommendations for top-line advice ---
ADVICE = {
"Pneumonia": "Possible infection. Recommend antibiotics and pulmonology consult.",
"Cardiomegaly": "Enlarged heart. Recommend echocardiography and cardiologist review.",
"Effusion": "Fluid in lung space. May need thoracentesis.",
"Fracture": "Possible bone break. Requires orthopedic consultation.",
"Edema": "Pulmonary fluid overload. Evaluate for heart failure.",
}
def get_advice(label):
return ADVICE.get(label, "Please consult a radiologist for further evaluation.")
def get_disease_info(label):
d = DISEASE_INFO.get(label)
if d:
return (
f"{label}: {d['description']}
"
f"Possible Causes: {d['cause']}
"
f"Recommendation: {d['recommendation']}"
)
return f"{label}: No extra info available. Please consult a radiologist."
# --- Image Preprocessing ---
def preprocess_image(pil_img: Image.Image) -> torch.Tensor:
"""Convert to grayscale, normalize, and resize for model."""
if pil_img.mode != "L":
pil_img = pil_img.convert("L")
img_array = np.array(pil_img).astype(np.float32)
img_array = xrv.datasets.normalize(img_array, 255)
img_array = img_array[None, ...] # [1, H, W]
img_array = xrv.datasets.XRayCenterCrop()(img_array)
img_array = xrv.datasets.XRayResizer(224)(img_array)
tensor = torch.from_numpy(img_array).unsqueeze(0).to(DEVICE)
return tensor
# --- X-ray Analysis (No CAM) ---
def analyse_xray(img: Image.Image):
try:
if img is None:
return "Please upload an X-ray image.", None
x = preprocess_image(img)
with torch.no_grad():
output = MODEL(x)
probs = torch.sigmoid(output)[0] * 100
topk = torch.topk(probs, 5)
html = "🩺 Top 5 Predictions
| Condition | Confidence | Details |
|---|
"
for idx in topk.indices:
label = LABELS[idx]
html += (
f"| {label} | "
f"{probs[idx]:.1f}% | "
f"{get_disease_info(label)} |
"
)
html += "
"
top_label = LABELS[topk.indices[0].item()]
html += f"
Recommended Action for '{top_label}': {get_advice(top_label)}"
return html, img.resize((224, 224))
except Exception as e:
return f"Error processing image: {str(e)}", None
# --- Report PDF Analysis ---
def analyse_report(file):
try:
if file is None:
return "Please upload a PDF report."
doc = fitz.open(file.name)
text = "\n".join(page.get_text() for page in doc)
doc.close()
found = []
for label in LABELS:
if re.search(rf"\b{label.lower()}\b", text.lower()):
found.append(label)
if found:
html = "📃 Findings Detected in Report:
"
for label in found:
html += f"- {get_disease_info(label)}
"
html += "
"
else:
html = "No known conditions detected from report text.
"
return html
except Exception as e:
return f"Error processing PDF: {str(e)}"
# --- Gradio UI ---
with gr.Blocks(title="🩻 RadiologyScan AI") as demo:
gr.Markdown(
"## 🩻 RadiologyScan AI\n"
"Perform fast AI-based analysis of Chest X-rays and medical reports\n"
"This tool provides informative summaries for common radiological findings. Not a substitute for a professional medical opinion."
)
with gr.Tabs():
with gr.Tab("🔍 X-ray Analysis"):
x_input = gr.Image(label="Upload Chest X-ray", type="pil")
x_out_html = gr.HTML()
x_out_image = gr.Image(label="Resized X-ray (224x224)")
analyze_btn = gr.Button("Analyze X-ray")
clear_btn = gr.Button("Clear")
analyze_btn.click(analyse_xray, inputs=x_input, outputs=[x_out_html, x_out_image])
clear_btn.click(lambda: (None, "", None), None, [x_input, x_out_html, x_out_image])
with gr.Tab("📄 PDF Report Analysis"):
pdf_input = gr.File(file_types=[".pdf"], label="Upload PDF Medical Report")
pdf_output = gr.HTML()
analyze_pdf_btn = gr.Button("Analyze Report")
clear_pdf_btn = gr.Button("Clear")
analyze_pdf_btn.click(analyse_report, inputs=pdf_input, outputs=pdf_output)
clear_pdf_btn.click(lambda: (None, ""), None, [pdf_input, pdf_output])
if __name__ == "__main__":
demo.launch(server_port=int(os.getenv("PORT", 7860)), show_error=True)