Spaces:
Paused
Paused
Commit
路
ded489b
1
Parent(s):
d6b54da
Added new features
Browse files
app.py
CHANGED
|
@@ -1,43 +1,63 @@
|
|
| 1 |
import os
|
| 2 |
import utils
|
|
|
|
| 3 |
import numpy as np
|
| 4 |
import gradio as gr
|
| 5 |
import tensorflow as tf
|
| 6 |
import matplotlib.pyplot as plt
|
| 7 |
from ttictoc import tic,toc
|
|
|
|
| 8 |
from urllib.request import urlretrieve
|
| 9 |
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
# 3D U-Net
|
| 13 |
if not os.path.exists("unet.h5"):
|
| 14 |
urlretrieve("https://dl.dropboxusercontent.com/s/ay5q8caqzlad7h5/unet.h5?dl=0", "unet.h5")
|
| 15 |
-
|
|
|
|
| 16 |
if not os.path.exists("resnet_50_23dataset.pth"):
|
| 17 |
urlretrieve("https://dl.dropboxusercontent.com/s/otxsgx3e31d5h9i/resnet_50_23dataset.pth?dl=0", "resnet_50_23dataset.pth")
|
| 18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
path_3d_unet = 'unet.h5'
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
|
|
|
|
|
|
|
|
|
|
| 21 |
with tf.device("cpu:0"):
|
| 22 |
model_unet = utils.import_3d_unet(path_3d_unet)
|
| 23 |
|
| 24 |
-
#
|
| 25 |
-
|
|
|
|
| 26 |
|
| 27 |
-
#
|
| 28 |
-
|
| 29 |
-
# brain = utils.brain_stripping(img, model_unet)
|
| 30 |
-
# print(toc())
|
| 31 |
|
| 32 |
-
#
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
# device_ids = [0]
|
| 36 |
-
# mednet = utils.create_mednet(weight_path, device_ids)
|
| 37 |
|
| 38 |
-
#
|
| 39 |
-
|
| 40 |
|
|
|
|
| 41 |
def load_img(file):
|
| 42 |
sitk, array = utils.load_img(file.name)
|
| 43 |
|
|
@@ -66,28 +86,52 @@ def show_img(img, mri_slice, update):
|
|
| 66 |
|
| 67 |
# return fig, gr.update(visible=True)
|
| 68 |
|
| 69 |
-
def process_img(img, brain_slice
|
| 70 |
-
progress(
|
| 71 |
|
| 72 |
with tf.device("cpu:0"):
|
| 73 |
brain = utils.brain_stripping(img, model_unet)
|
| 74 |
|
| 75 |
fig, update_slider, _ = show_img(brain, brain_slice, update=True)
|
| 76 |
|
| 77 |
-
return brain, fig, update_slider
|
| 78 |
|
| 79 |
-
def
|
| 80 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
|
| 82 |
-
|
|
|
|
| 83 |
|
| 84 |
-
# demo = gr.Interface(fn=load_img,
|
| 85 |
-
# inputs=gr.File(file_count="single", file_type=[".nii"]),
|
| 86 |
-
# outputs=gr.Plot()
|
| 87 |
-
# # outputs='text'
|
| 88 |
-
# )
|
| 89 |
|
| 90 |
-
|
| 91 |
|
| 92 |
with gr.Blocks(theme=gr.themes.Base()) as demo:
|
| 93 |
with gr.Row():
|
|
@@ -112,11 +156,36 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
|
|
| 112 |
input_sex = gr.Dropdown(["Male", "Female"], label="Sex")
|
| 113 |
|
| 114 |
with gr.Tab("Clinical data"):
|
| 115 |
-
input_MMSE = gr.
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
|
| 121 |
with gr.Tab("Vital Signs"):
|
| 122 |
input_Diastolic_blood_pressure = gr.Number(label='Diastolic Blood Pressure(mm Hg)')
|
|
@@ -142,33 +211,47 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
|
|
| 142 |
clear_button = gr.Button(value="Clear")
|
| 143 |
|
| 144 |
# Bot贸n para procesar imagen
|
| 145 |
-
process_button = gr.Button(value="
|
|
|
|
|
|
|
|
|
|
| 146 |
|
| 147 |
# Outputs
|
| 148 |
with gr.Column(variant="panel", scale=1):
|
| 149 |
gr.Markdown('<h2 style="text-align: center; color:#235784;">MRI visualization</h2>')
|
| 150 |
|
| 151 |
-
|
| 152 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
|
| 154 |
-
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
|
| 158 |
-
|
| 159 |
-
label="MRI Slice",
|
| 160 |
-
visible=False)
|
| 161 |
-
|
| 162 |
-
# Plot para im谩gen procesada
|
| 163 |
-
plot_brain = gr.Plot(label="Imagen MRI procesada", visible=True)
|
| 164 |
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 172 |
|
| 173 |
# componentes =
|
| 174 |
|
|
@@ -177,6 +260,7 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
|
|
| 177 |
original_input_img = gr.State()
|
| 178 |
brain_img = gr.State()
|
| 179 |
|
|
|
|
| 180 |
update_true = gr.State(True)
|
| 181 |
update_false = gr.State(False)
|
| 182 |
|
|
@@ -191,10 +275,6 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
|
|
| 191 |
[original_input_img, mri_slider, update_true],
|
| 192 |
[plot_img_original, mri_slider, process_button])
|
| 193 |
|
| 194 |
-
# Limpiar campos
|
| 195 |
-
clear_button.click(fn=clear,
|
| 196 |
-
outputs=[input_file, plot_img_original, mri_slider])
|
| 197 |
-
|
| 198 |
# Actualizar imagen original
|
| 199 |
mri_slider.change(show_img,
|
| 200 |
[original_input_img, mri_slider, update_false],
|
|
@@ -203,42 +283,24 @@ with gr.Blocks(theme=gr.themes.Base()) as demo:
|
|
| 203 |
# Procesar imagen
|
| 204 |
process_button.click(fn=process_img,
|
| 205 |
inputs=[original_input_sitk, brain_slider],
|
| 206 |
-
outputs=[brain_img,plot_brain,brain_slider])
|
| 207 |
|
| 208 |
# Actualizar imagen procesada
|
| 209 |
brain_slider.change(show_img,
|
| 210 |
[brain_img, brain_slider, update_false],
|
| 211 |
[plot_brain])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 212 |
|
| 213 |
|
| 214 |
if __name__ == "__main__":
|
| 215 |
-
demo.queue(concurrency_count=20)
|
| 216 |
demo.launch()
|
| 217 |
-
|
| 218 |
-
# # Visualizaci贸n resultados
|
| 219 |
-
# mri_slice = 100
|
| 220 |
-
|
| 221 |
-
# # Plot Comparaci贸n m谩scaras
|
| 222 |
-
# fig, axs = plt.subplots(1,2)
|
| 223 |
-
# fig.subplots_adjust(bottom=0.15)
|
| 224 |
-
# fig.suptitle('Comparaci贸n M谩scaras Obtenidas')
|
| 225 |
-
|
| 226 |
-
# axs[0].set_title('MRI original')
|
| 227 |
-
# axs[0].imshow(img[mri_slice,:,:],cmap='gray')
|
| 228 |
-
|
| 229 |
-
# axs[1].set_title('Cerebro extraido con 3D U-Net')
|
| 230 |
-
# axs[1].imshow(brain[mri_slice,:,:],cmap='gray')
|
| 231 |
-
|
| 232 |
-
|
| 233 |
-
# # Slider para cambiar slice
|
| 234 |
-
# ax_slider = plt.axes([0.15, 0.05, 0.75, 0.03])
|
| 235 |
-
# mri_slice_slider = Slider(ax_slider, 'Slice', 0, 192, 100, valstep=1)
|
| 236 |
-
|
| 237 |
-
# def update(val):
|
| 238 |
-
# mri_slice = mri_slice_slider.val
|
| 239 |
-
|
| 240 |
-
# axs[0].imshow(img[:,:,mri_slice],cmap='gray')
|
| 241 |
-
# axs[1].imshow(brain[mri_slice,:,:],cmap='gray')
|
| 242 |
-
|
| 243 |
-
# # Actualizar plot comparaci贸n m谩scaras
|
| 244 |
-
# mri_slice_slider.on_changed(update)
|
|
|
|
| 1 |
import os
|
| 2 |
import utils
|
| 3 |
+
import pickle
|
| 4 |
import numpy as np
|
| 5 |
import gradio as gr
|
| 6 |
import tensorflow as tf
|
| 7 |
import matplotlib.pyplot as plt
|
| 8 |
from ttictoc import tic,toc
|
| 9 |
+
from keras.models import load_model
|
| 10 |
from urllib.request import urlretrieve
|
| 11 |
|
| 12 |
+
'''--------------------------- Descarga de modelos ----------------------------'''
|
| 13 |
+
|
| 14 |
+
# 3D U-Net
|
| 15 |
if not os.path.exists("unet.h5"):
|
| 16 |
urlretrieve("https://dl.dropboxusercontent.com/s/ay5q8caqzlad7h5/unet.h5?dl=0", "unet.h5")
|
| 17 |
+
|
| 18 |
+
# Med3D
|
| 19 |
if not os.path.exists("resnet_50_23dataset.pth"):
|
| 20 |
urlretrieve("https://dl.dropboxusercontent.com/s/otxsgx3e31d5h9i/resnet_50_23dataset.pth?dl=0", "resnet_50_23dataset.pth")
|
| 21 |
|
| 22 |
+
# Clasificador de im谩gen SVM
|
| 23 |
+
if not os.path.exists("svm_model.pickle"):
|
| 24 |
+
urlretrieve("https://dl.dropboxusercontent.com/s/n3tb3r6oyf06xfx/svm_model.pickle?dl=0", "svm_model.pickle")
|
| 25 |
+
|
| 26 |
+
# Nivel de riesgo
|
| 27 |
+
if not os.path.exists("mlp_probabilidad.h5"):
|
| 28 |
+
urlretrieve("https://dl.dropboxusercontent.com/s/78fjlg374mvjygd/mlp_probabilidad.h5?dl=0", "mlp_probabilidad.h5")
|
| 29 |
+
|
| 30 |
+
# Scaler para scores
|
| 31 |
+
if not os.path.exists("scaler.pickle"):
|
| 32 |
+
urlretrieve("https://dl.dropboxusercontent.com/s/ow6pe4k45r3xkbl/scaler.pickle?dl=0", "scaler.pickle")
|
| 33 |
+
|
| 34 |
path_3d_unet = 'unet.h5'
|
| 35 |
+
weight_path = 'resnet_50_23dataset.pth'
|
| 36 |
+
svm_path = "svm_model.pickle"
|
| 37 |
+
prob_model_path = "mlp_probabilidad.h5"
|
| 38 |
+
scaler_path = "scaler.pickle"
|
| 39 |
|
| 40 |
+
|
| 41 |
+
'''---------------------------- Carga de modelos ------------------------------'''
|
| 42 |
+
# 3D U-Net
|
| 43 |
with tf.device("cpu:0"):
|
| 44 |
model_unet = utils.import_3d_unet(path_3d_unet)
|
| 45 |
|
| 46 |
+
# MedNet
|
| 47 |
+
device_ids = [7]
|
| 48 |
+
mednet_model = utils.create_mednet(weight_path, device_ids)
|
| 49 |
|
| 50 |
+
# SVM model
|
| 51 |
+
svm_model = pickle.load(open(svm_path, 'rb'))
|
|
|
|
|
|
|
| 52 |
|
| 53 |
+
# Nivel de riesgo
|
| 54 |
+
with tf.device("cpu:0"):
|
| 55 |
+
prob_model = load_model(prob_model_path)
|
|
|
|
|
|
|
| 56 |
|
| 57 |
+
# Scaler
|
| 58 |
+
scaler = pickle.load(open(scaler_path, 'rb'))
|
| 59 |
|
| 60 |
+
'''-------------------------------- Funciones ---------------------------------'''
|
| 61 |
def load_img(file):
|
| 62 |
sitk, array = utils.load_img(file.name)
|
| 63 |
|
|
|
|
| 86 |
|
| 87 |
# return fig, gr.update(visible=True)
|
| 88 |
|
| 89 |
+
def process_img(img, brain_slice):
|
| 90 |
+
# progress(None,desc="Processing...")
|
| 91 |
|
| 92 |
with tf.device("cpu:0"):
|
| 93 |
brain = utils.brain_stripping(img, model_unet)
|
| 94 |
|
| 95 |
fig, update_slider, _ = show_img(brain, brain_slice, update=True)
|
| 96 |
|
| 97 |
+
return brain, fig, update_slider, gr.update(visible=True)
|
| 98 |
|
| 99 |
+
def get_diagnosis(brain_img, age, MMSE, GDSCALE, CDR, FAQ, NPI, sex):
|
| 100 |
+
# Extracci贸n de caracter铆sticas de imagen
|
| 101 |
+
features = utils.get_features(brain_img, mednet_model)
|
| 102 |
+
|
| 103 |
+
# Clasificaci贸n de imagen
|
| 104 |
+
label_img = np.array([svm_model.predict(features)])
|
| 105 |
+
|
| 106 |
+
if sex == "Male":
|
| 107 |
+
sex_dum = 1
|
| 108 |
+
else:
|
| 109 |
+
sex_dum = 0
|
| 110 |
+
|
| 111 |
+
scores = np.array([age, MMSE, GDSCALE, CDR, FAQ, NPI, sex_dum, label_img])
|
| 112 |
+
|
| 113 |
+
print(scores)
|
| 114 |
+
|
| 115 |
+
# Normalizaci贸n de scores
|
| 116 |
+
scores_norm = scaler.transform(scores.reshape(1,-1))
|
| 117 |
+
|
| 118 |
+
print(scores_norm)
|
| 119 |
+
|
| 120 |
+
with tf.device("cpu:0"):
|
| 121 |
+
# Probabilidad de tener MCI
|
| 122 |
+
prob = prob_model.predict(scores_norm)[0,0]
|
| 123 |
+
|
| 124 |
+
# Probabilidad de tener MCI
|
| 125 |
+
print(prob)
|
| 126 |
+
diagnosis = f"The patient has a probability of {(100*prob):.2f}% of having MCI"
|
| 127 |
+
|
| 128 |
+
return gr.update(value=diagnosis)
|
| 129 |
|
| 130 |
+
def clear():
|
| 131 |
+
return gr.File.update(value=None), gr.Plot.update(value=None), gr.update(visible=False), gr.Plot.update(value=None), gr.update(visible=False), gr.update(value="The diagnosis will show here..."), gr.update(visible=False), gr.update(visible=False)
|
| 132 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
|
| 134 |
+
'''--------------------------------- Interfaz ---------------------------------'''
|
| 135 |
|
| 136 |
with gr.Blocks(theme=gr.themes.Base()) as demo:
|
| 137 |
with gr.Row():
|
|
|
|
| 156 |
input_sex = gr.Dropdown(["Male", "Female"], label="Sex")
|
| 157 |
|
| 158 |
with gr.Tab("Clinical data"):
|
| 159 |
+
input_MMSE = gr.Slider(minimum=0,
|
| 160 |
+
maximum=30,
|
| 161 |
+
value=0,
|
| 162 |
+
step=1,
|
| 163 |
+
label="MMSE total score")
|
| 164 |
+
|
| 165 |
+
input_GDSCALE = gr.Slider(minimum=0,
|
| 166 |
+
maximum=12,
|
| 167 |
+
value=0,
|
| 168 |
+
step=1,
|
| 169 |
+
label="GDSCALE total score")
|
| 170 |
+
|
| 171 |
+
input_CDR = gr.Slider(minimum=0,
|
| 172 |
+
maximum=3,
|
| 173 |
+
value=0,
|
| 174 |
+
step=0.5,
|
| 175 |
+
label="Global CDR")
|
| 176 |
+
|
| 177 |
+
input_FAQ = gr.Slider(minimum=0,
|
| 178 |
+
maximum=30,
|
| 179 |
+
value=0,
|
| 180 |
+
step=1,
|
| 181 |
+
label="FAQ total score")
|
| 182 |
+
|
| 183 |
+
input_NPI_Q = gr.Slider(minimum=0,
|
| 184 |
+
maximum=30,
|
| 185 |
+
value=0,
|
| 186 |
+
step=1,
|
| 187 |
+
label="NPI-Q total score")
|
| 188 |
+
|
| 189 |
|
| 190 |
with gr.Tab("Vital Signs"):
|
| 191 |
input_Diastolic_blood_pressure = gr.Number(label='Diastolic Blood Pressure(mm Hg)')
|
|
|
|
| 211 |
clear_button = gr.Button(value="Clear")
|
| 212 |
|
| 213 |
# Bot贸n para procesar imagen
|
| 214 |
+
process_button = gr.Button(value="Process MRI", visible=False, variant="primary")
|
| 215 |
+
|
| 216 |
+
# Bot贸n para obtener diagnostico
|
| 217 |
+
diagnostic_button = gr.Button(value="Get diagnosis", visible=False, variant="primary")
|
| 218 |
|
| 219 |
# Outputs
|
| 220 |
with gr.Column(variant="panel", scale=1):
|
| 221 |
gr.Markdown('<h2 style="text-align: center; color:#235784;">MRI visualization</h2>')
|
| 222 |
|
| 223 |
+
with gr.Box():
|
| 224 |
+
gr.Markdown('<h4 style="color:#235784;">Loaded MRI</h4>')
|
| 225 |
+
# Plot para im谩gen original
|
| 226 |
+
plot_img_original = gr.Plot(show_label=False)
|
| 227 |
+
|
| 228 |
+
# Slider para im谩gen original
|
| 229 |
+
mri_slider = gr.Slider(minimum=0,
|
| 230 |
+
maximum=192,
|
| 231 |
+
value=100,
|
| 232 |
+
step=1,
|
| 233 |
+
label="MRI Slice",
|
| 234 |
+
visible=False)
|
| 235 |
|
| 236 |
+
with gr.Box():
|
| 237 |
+
gr.Markdown('<h4 style="color:#235784;">Proccessed MRI</h4>')
|
| 238 |
+
|
| 239 |
+
# Plot para im谩gen procesada
|
| 240 |
+
plot_brain = gr.Plot(show_label=False, visible=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 241 |
|
| 242 |
+
# Slider para im谩gen procesada
|
| 243 |
+
brain_slider = gr.Slider(minimum=0,
|
| 244 |
+
maximum=192,
|
| 245 |
+
value=100,
|
| 246 |
+
step=1,
|
| 247 |
+
label="MRI Slice",
|
| 248 |
+
visible=False)
|
| 249 |
+
|
| 250 |
+
with gr.Box():
|
| 251 |
+
gr.Markdown('<h2 style="text-align: center; color:#235784;">Diagnosis</h2>')
|
| 252 |
+
|
| 253 |
+
# Texto del diagnostico
|
| 254 |
+
diagnosis_text = gr.Textbox(label="Diagnosis",interactive=False, placeholder="The diagnosis will show here...")
|
| 255 |
|
| 256 |
# componentes =
|
| 257 |
|
|
|
|
| 260 |
original_input_img = gr.State()
|
| 261 |
brain_img = gr.State()
|
| 262 |
|
| 263 |
+
|
| 264 |
update_true = gr.State(True)
|
| 265 |
update_false = gr.State(False)
|
| 266 |
|
|
|
|
| 275 |
[original_input_img, mri_slider, update_true],
|
| 276 |
[plot_img_original, mri_slider, process_button])
|
| 277 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 278 |
# Actualizar imagen original
|
| 279 |
mri_slider.change(show_img,
|
| 280 |
[original_input_img, mri_slider, update_false],
|
|
|
|
| 283 |
# Procesar imagen
|
| 284 |
process_button.click(fn=process_img,
|
| 285 |
inputs=[original_input_sitk, brain_slider],
|
| 286 |
+
outputs=[brain_img,plot_brain,brain_slider, diagnostic_button])
|
| 287 |
|
| 288 |
# Actualizar imagen procesada
|
| 289 |
brain_slider.change(show_img,
|
| 290 |
[brain_img, brain_slider, update_false],
|
| 291 |
[plot_brain])
|
| 292 |
+
|
| 293 |
+
# Actualizar diagnostico
|
| 294 |
+
diagnostic_button.click(fn=get_diagnosis,
|
| 295 |
+
inputs=[brain_img, input_age, input_MMSE, input_GDSCALE, input_CDR, input_FAQ, input_NPI_Q, input_sex],
|
| 296 |
+
outputs=[diagnosis_text])
|
| 297 |
+
|
| 298 |
+
# Limpiar campos
|
| 299 |
+
clear_button.click(fn=clear,
|
| 300 |
+
outputs=[input_file, plot_img_original, mri_slider, plot_brain, brain_slider, diagnosis_text, process_button, diagnostic_button])
|
| 301 |
+
|
| 302 |
|
| 303 |
|
| 304 |
if __name__ == "__main__":
|
| 305 |
+
# demo.queue(concurrency_count=20)
|
| 306 |
demo.launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
utils.py
CHANGED
|
@@ -27,7 +27,6 @@ def import_3d_unet(path_3d_unet):
|
|
| 27 |
return (2. * K.sum(flat_y_true * flat_y_pred) + smoothing_factor) / (K.sum(flat_y_true) + K.sum(flat_y_pred) + smoothing_factor)
|
| 28 |
|
| 29 |
# Cargar modelo preentrenado
|
| 30 |
-
# with tf.device('/cpu:0'):
|
| 31 |
model = load_model(path_3d_unet, custom_objects={'dice_coefficient':dice_coefficient, 'iou_score':sm.metrics.IOUScore(threshold=0.5)})
|
| 32 |
return model
|
| 33 |
|
|
@@ -179,4 +178,14 @@ def get_features(brain, mednet_model):
|
|
| 179 |
features = features.cpu().numpy()
|
| 180 |
|
| 181 |
torch.cuda.empty_cache()
|
| 182 |
-
return features
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
return (2. * K.sum(flat_y_true * flat_y_pred) + smoothing_factor) / (K.sum(flat_y_true) + K.sum(flat_y_pred) + smoothing_factor)
|
| 28 |
|
| 29 |
# Cargar modelo preentrenado
|
|
|
|
| 30 |
model = load_model(path_3d_unet, custom_objects={'dice_coefficient':dice_coefficient, 'iou_score':sm.metrics.IOUScore(threshold=0.5)})
|
| 31 |
return model
|
| 32 |
|
|
|
|
| 178 |
features = features.cpu().numpy()
|
| 179 |
|
| 180 |
torch.cuda.empty_cache()
|
| 181 |
+
return features
|
| 182 |
+
|
| 183 |
+
# Classify image
|
| 184 |
+
def get_prediction(features, scores, svm_model, dl_model):
|
| 185 |
+
prediction = svm_model.predict(features)
|
| 186 |
+
|
| 187 |
+
# x = np.concatenate((scores, prediction))
|
| 188 |
+
|
| 189 |
+
# prob = dl_model.predict(x)
|
| 190 |
+
|
| 191 |
+
return prediction
|