app.py

import torch
from transformers import ViTImageProcessor, ViTForImageClassification
from fastai.learner import load_learner
from fastai.vision.core import PILImage
from PIL import Image
import matplotlib.pyplot as plt
import numpy as np
import gradio as gr
import io
import base64

# --- Cargar modelo ViT preentrenado fine‑tuned HAM10000 ---
TF_MODEL_NAME = "Anwarkh1/Skin_Cancer-Image_Classification"
feature_extractor_tf = ViTImageProcessor.from_pretrained(TF_MODEL_NAME)
model_tf_vit = ViTForImageClassification.from_pretrained(TF_MODEL_NAME)
model_tf_vit.eval()

# 🔹 Cargar modelo ViT base
MODEL_NAME = "ahishamm/vit-base-HAM-10000-sharpened-patch-32"
feature_extractor = ViTImageProcessor.from_pretrained(MODEL_NAME)
model_vit = ViTForImageClassification.from_pretrained(MODEL_NAME)
model_vit.eval()

# 🔹 Cargar modelos Fast.ai locales
model_malignancy = load_learner("ada_learn_malben.pkl")
model_norm2000 = load_learner("ada_learn_skin_norm2000.pkl")

# Clases estándar de HAM10000
CLASSES = [
    "Queratosis actínica / Bowen", "Carcinoma células basales",
    "Lesión queratósica benigna", "Dermatofibroma",
    "Melanoma maligno", "Nevus melanocítico", "Lesión vascular"
]
RISK_LEVELS = {
    0: {'level': 'Moderado', 'color': '#ffaa00', 'weight': 0.6},
    1: {'level': 'Alto',     'color': '#ff4444', 'weight': 0.8},
    2: {'level': 'Bajo',     'color': '#44ff44', 'weight': 0.1},
    3: {'level': 'Bajo',     'color': '#44ff44', 'weight': 0.1},
    4: {'level': 'Crítico',  'color': '#cc0000', 'weight': 1.0},
    5: {'level': 'Bajo',     'color': '#44ff44', 'weight': 0.1},
    6: {'level': 'Bajo',     'color': '#44ff44', 'weight': 0.1}
}
MALIGNANT_INDICES = [0, 1, 4]  # akiec, bcc, melanoma

def analizar_lesion_combined(img):
    img_fastai = PILImage.create(img)

    # ViT base
    inputs = feature_extractor(img, return_tensors="pt")
    with torch.no_grad():
        outputs = model_vit(**inputs)
        probs_vit = outputs.logits.softmax(dim=-1).cpu().numpy()[0]
    idx_vit = int(np.argmax(probs_vit))
    class_vit = CLASSES[idx_vit]
    conf_vit = probs_vit[idx_vit]

    # Fast.ai modelos
    _, _, probs_mal = model_malignancy.predict(img_fastai)
    prob_malign = float(probs_mal[1])
    pred_fast_type, _, _ = model_norm2000.predict(img_fastai)

    # ViT pre-trained fine-tuned (último modelo recomendado)
    inputs_tf = feature_extractor_tf(img, return_tensors="pt")
    with torch.no_grad():
        outputs_tf = model_tf_vit(**inputs_tf)
        probs_tf = outputs_tf.logits.softmax(dim=-1).cpu().numpy()[0]
    idx_tf = int(np.argmax(probs_tf))
    class_tf_model = CLASSES[idx_tf]
    conf_tf = probs_tf[idx_tf]
    mal_tf = "Maligno" if idx_tf in MALIGNANT_INDICES else "Benigno"

    # Gráfico ViT base
    colors = [RISK_LEVELS[i]['color'] for i in range(7)]
    fig, ax = plt.subplots(figsize=(8, 3))
    ax.bar(CLASSES, probs_vit*100, color=colors)
    ax.set_title("Probabilidad ViT base por tipo de lesión")
    ax.set_ylabel("Probabilidad (%)")
    ax.set_xticks(np.arange(len(CLASSES)))
    ax.set_xticklabels(CLASSES, rotation=45, ha='right')
    ax.grid(axis='y', alpha=0.2)
    plt.tight_layout()
    buf = io.BytesIO()
    plt.savefig(buf, format="png")
    plt.close(fig)
    html_chart = f'<img src="data:image/png;base64,{base64.b64encode(buf.getvalue()).decode()}" style="max-width:100%"/>'

    informe = f"""
    <div style="font-family:sans-serif; max-width:800px; margin:auto">
    <h2>🧪 Diagnóstico por múltiples modelos de IA</h2>
    <table style="width:100%; font-size:16px; border-collapse:collapse">
      <tr><th>Modelo</th><th>Resultado</th><th>Confianza</th></tr>
      <tr><td>🧠 ViT base</td><td><b>{class_vit}</b></td><td>{conf_vit:.1%}</td></tr>
      <tr><td>🧬 Fast.ai (tipo)</td><td><b>{pred_fast_type}</b></td><td>N/A</td></tr>
      <tr><td>⚠️ Fast.ai (malignidad)</td><td><b>{'Maligno' if prob_malign > 0.5 else 'Benigno'}</b></td><td>{prob_malign:.1%}</td></tr>
      <tr><td>🌟 ViT fined‑tuned (HAM10000)</td><td><b>{mal_tf} ({class_tf_model})</b></td><td>{conf_tf:.1%}</td></tr>
    </table><br>
    <b>🩺 Recomendación automática:</b><br>
    """
    risk = sum(probs_vit[i] * RISK_LEVELS[i]['weight'] for i in range(7))
    if prob_malign > 0.7 or risk > 0.6:
        informe += "🚨 <b>CRÍTICO</b> – Derivación urgente a oncología dermatológica"
    elif prob_malign > 0.4 or risk > 0.4:
        informe += "⚠️ <b>ALTO RIESGO</b> – Consulta con dermatólogo en 7 días"
    elif risk > 0.2:
        informe += "📋 <b>RIESGO MODERADO</b> – Evaluación programada en 2-4 semanas"
    else:
        informe += "✅ <b>BAJO RIESGO</b> – Seguimiento de rutina (3-6 meses)"
    informe += "</div>"""

    return informe, html_chart

demo = gr.Interface(
    fn=analizar_lesion_combined,
    inputs=gr.Image(type="pil"),
    outputs=[gr.HTML(label="Informe"), gr.HTML(label="Gráfico ViT base")],
    title="Detector de Lesiones Cutáneas (ViT + Fast.ai)",
)
if __name__ == "__main__":
    demo.launch()