app.py

import gradio as gr
import torch
import numpy as np
import nibabel as nib
from model import My3DModel

model = My3DModel()
model.load_state_dict(torch.load("model.pt", map_location="cpu"))
model.eval()

def predict_from_mri(file_obj):
    img = nib.load(file_obj.name)
    data = img.get_fdata().astype(np.float32)
    data = np.expand_dims(data, axis=0)  # (1, D, H, W)
    tensor = torch.tensor(data).unsqueeze(0)  # (1, 1, D, H, W)

    with torch.no_grad():
        output = model(tensor)
        pred_class = torch.argmax(output, dim=1).item()
    
    return ["CN", "MCI", "AD"][pred_class]

iface = gr.Interface(fn=predict_from_mri,
                     inputs=gr.File(label="MRI .nii dosyası"),
                     outputs="text",
                     title="3D MRI Alzheimer Teşhis Modeli")
iface.launch()