利用切片实现AD诊断

原创已于 2025-12-14 00:47:50 修改 · 554 阅读

9 ·

CC 4.0 BY-SA版权

文章标签：

#人工智能 #医学图像 #医学影像

于 2025-12-14 00:45:35 首次发布

AgenticCoding·十二月创作之星挑战赛 10w+人浏览 329人参与

目标：利用切片实现AD诊断
数据划分：143例用于训练，62例用于验证，51例用于测试
结果：
训练30次后，验证集准确度为80.65％
在这里插入图片描述
利用测试集数据进行测试，51例数据可以正确识别出42例。
验证方法：使用五折交叉验证
ROC曲线下面积为0.92

ipynb文件：

```python
#使用自己搭建的网络来分析切片
#目的：与融合之后的label实现例数上的对应

#载入模块
#matplotlib:用于画图
import matplotlib
#从matplotlib中载入子模块
from matplotlib import pyplot as plt
#nibabel:读取和编写NIFTI文件
import nibabel as nib
#os:处理文件和目录
import os
#numpy:矩阵操作
import numpy as np
#cv2:用于图像处理
import cv2
#keras:把很多深度学习要用到的函数封装起来,使用keras可以快速调用函数搭建网络,精简代码
import keras
#从keras中载入3D卷积层(Conv),全连接层(Dense),池化层(Pool)
#Flatten:Flatten是指将多维的矩阵拉开,变成一维向量来表示
from keras.layers import Conv3D,Conv2D,Dense,Input,MaxPool2D,MaxPool3D,Flatten
#载入模型
from keras.models import Model
#载入模块
#KFold:K折交叉模块
from sklearn.model_selection import KFold
#shuffle:将数据打乱
from sklearn.utils import shuffle
#metrics:用于定义评估模型的指标
from sklearn import metrics
import matplotlib
from matplotlib import pylab as plt
import nibabel as nib
import os
import numpy as np
import cv2
import keras
from keras.layers import Conv3D,Conv2D,Dense,Input,MaxPool2D,MaxPool3D,Flatten
from keras.models import Model
import os
import numpy as np
from nilearn import datasets
from nilearn import input_data
import cv2
from sklearn.utils import shuffle
from keras import models
from keras.models import load_model
import nilearn
from nilearn.image import new_img_like
from nilearn import datasets
from nilearn import input_data
from nilearn.connectome import ConnectivityMeasure
import matplotlib.pyplot as plt
from nilearn import image
import nibabel as nib
import skimage.io as io
import numpy as np
from nilearn import datasets,plotting,image
import random
#nilearn:一个用于神经影像的机器学习工具包
#nibabel:用于读取nii文件,提取图像数据的工具包
#numpy:用于进行矩阵操作
#skimage:用于进行图像处理的模块
import keras
from keras.layers import Conv3D,Conv2D,Dense,Input,MaxPool2D,MaxPool3D,Flatten
from keras.models import Model
from keras.layers import Dense,Input,Add,LSTM,Flatten,GRU
from keras.optimizers import Adam,SGD
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.layers import concatenate
from keras.layers import Conv1D, GlobalAveragePooling1D, MaxPooling1D
from keras.layers import Conv2D, GlobalAveragePooling2D, MaxPooling2D
from keras.layers import TimeDistributed
from keras import regularizers
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.optimizers import Adam,RMSprop
from keras.layers import Dense,Input,Add,LSTM,Flatten,GRU

/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
Using TensorFlow backend.
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/local/lib/python3.5/dist-packages/skimage/__init__.py:71: ResourceWarning: unclosed file <_io.TextIOWrapper name='/usr/local/lib/python3.5/dist-packages/pytest.py' mode='r' encoding='utf-8'>
  imp.find_module('pytest')
/usr/local/lib/python3.5/dist-packages/nilearn/plotting/__init__.py:20: UserWarning: 
This call to matplotlib.use() has no effect because the backend has already
been chosen; matplotlib.use() must be called *before* pylab, matplotlib.pyplot,
or matplotlib.backends is imported for the first time.

The backend was *originally* set to 'module://ipykernel.pylab.backend_inline' by the following code:
  File "/usr/lib/python3.5/runpy.py", line 184, in _run_module_as_main
    "__main__", mod_spec)
  File "/usr/lib/python3.5/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py", line 16, in <module>
    app.launch_new_instance()
  File "/usr/local/lib/python3.5/dist-packages/traitlets/config/application.py", line 658, in launch_instance
    app.start()
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelapp.py", line 486, in start
    self.io_loop.start()
  File "/usr/local/lib/python3.5/dist-packages/tornado/platform/asyncio.py", line 132, in start
    self.asyncio_loop.run_forever()
  File "/usr/lib/python3.5/asyncio/base_events.py", line 345, in run_forever
    self._run_once()
  File "/usr/lib/python3.5/asyncio/base_events.py", line 1312, in _run_once
    handle._run()
  File "/usr/lib/python3.5/asyncio/events.py", line 125, in _run
    self._callback(*self._args)
  File "/usr/local/lib/python3.5/dist-packages/tornado/ioloop.py", line 758, in _run_callback
    ret = callback()
  File "/usr/local/lib/python3.5/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
    return fn(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 536, in <lambda>
    self.io_loop.add_callback(lambda : self._handle_events(self.socket, 0))
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 450, in _handle_events
    self._handle_recv()
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 480, in _handle_recv
    self._run_callback(callback, msg)
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 432, in _run_callback
    callback(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
    return fn(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 283, in dispatcher
    return self.dispatch_shell(stream, msg)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 233, in dispatch_shell
    handler(stream, idents, msg)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 399, in execute_request
    user_expressions, allow_stdin)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/ipkernel.py", line 208, in do_execute
    res = shell.run_cell(code, store_history=store_history, silent=silent)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/zmqshell.py", line 537, in run_cell
    return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2662, in run_cell
    raw_cell, store_history, silent, shell_futures)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2785, in _run_cell
    interactivity=interactivity, compiler=compiler, result=result)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2901, in run_ast_nodes
    if self.run_code(code, result):
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2961, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-2-9541018185df>", line 5, in <module>
    from matplotlib import pyplot as plt
  File "/usr/local/lib/python3.5/dist-packages/matplotlib/pyplot.py", line 71, in <module>
    from matplotlib.backends import pylab_setup
  File "/usr/local/lib/python3.5/dist-packages/matplotlib/backends/__init__.py", line 16, in <module>
    line for line in traceback.format_stack()


  matplotlib.use('Agg')
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)

#定义数据路径
ad_path = '/home/ADNI/outcome/AD/T1ImgNewSegment/'
cn_path = '/home/ADNI/outcome/CN/T1ImgNewSegment/'
#读取数据路径下的所有文件
#os.listdir表示遍历文件夹下的文件(只有文件名,不包含路径)
#lambda x:定义一个变量x,并遍历目录
#将文件路径和文件名拼接起来作为结果返回
#结果保存于一个list(列表文件)中
ad_files = list(map(lambda x: ad_path+x,os.listdir(ad_path)))
cn_files = list(map(lambda x: cn_path+x,os.listdir(cn_path)))
#移除文件夹中的无关文件
ad_files.remove(ad_path + '.ipynb_checkpoints')
#cn_files.remove(cn_path + '.ipynb_checkpoints')

ad_files = ad_files
cn_files = cn_files

data_1 = []
label_2 = []
for file in ad_files:
    data = nib.load(file).get_data()
    data_1.append(cv2.resize(data,(121,121)))
    label_2.append([0,1])
for file in cn_files:
    data = nib.load(file).get_data()
    data_1.append(cv2.resize(data,(121,121)))
    label_2.append([1,0])
data_1 = np.array(data_1)
label_2 = np.array(label_2)

data_1.shape

(256, 121, 121, 121)

data_2 = []
for i in range(data_1.shape[0]):
    img1 = data_1[i,:,:,60]
    img2 = data_1[i,:,60,:]
    img3 = data_1[i,60,:,:]
    img = np.stack((img1,img2,img3))
    data_2.append(img)

data_2 = np.array(data_2)

data_2.shape

(256, 3, 121, 121)

data_2 = data_2.transpose(0,3,2,1)

data_2.shape

(256, 121, 121, 3)

#数据归一化(除以最大值,将最大值转化为1)
data_2 = np.array(data_2)
max_data = np.max(data_2)
data_2 = data_2/max_data

data_2,label_2 = shuffle(data_2,label_2)

data_2.shape

(256, 121, 121, 3)

size1 = 121
size2 = 121
channels = 3
def build_model():
    inp = Input(shape=(size1,size2,channels))
    x1 = Conv2D(filters=1,kernel_size=(3,3),strides=(1,1),activation='relu', input_shape=(size1,size2,channels))(inp)
    x1 = Flatten()(x1)
    x1 = Dense(2, activation='softmax')(x1)
    model=Model(input=inp,outputs=x1)
    return model

input_shape = data_2.shape[1:]
output_shape = 2

data = data_2
label = label_2

from sklearn.metrics import roc_curve,auc
from scipy import interp
#定义n折交叉验证
KF = KFold(n_splits = 5)
tprs=[]
aucs=[]
mean_fpr=np.linspace(0,1,100)
i=0
#data为数据集,利用KF.split划分训练集和测试集
for train_index,test_index in KF.split(data):
    #建立模型，并对训练集进行测试，求出预测得分
    #划分训练集和测试集
    X_train,X_test = data[train_index],data[test_index]
    Y_train,Y_test = label[train_index],label[test_index]
    #建立模型(模型已经定义)
    model = build_model()
    #编译模型
    #定义优化器
    adam = keras.optimizers.Adam(lr=0.00005, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)
    model.compile(loss='categorical_crossentropy', optimizer= adam, metrics=['acc'])
    #训练模型
    model.fit(X_train,Y_train,batch_size = 32,validation_split = 0.3,epochs = 30)
    #利用model.predict获取测试集的预测值
    y_pred = model.predict(X_test,batch_size = 1)
    #计算fpr(假阳性率),tpr(真阳性率),thresholds(阈值)[绘制ROC曲线要用到这几个值]
    fpr,tpr,thresholds=roc_curve(Y_test[:,1],y_pred[:,1])
    #interp:插值 把结果添加到tprs列表中 
    tprs.append(interp(mean_fpr,fpr,tpr))
    tprs[-1][0]=0.0
    #计算auc
    roc_auc=auc(fpr,tpr)
    aucs.append(roc_auc)
    #画图，只需要plt.plot(fpr,tpr),变量roc_auc只是记录auc的值，通过auc()函数计算出来
    plt.plot(fpr,tpr,lw=1,alpha=0.3,label='ROC fold %d(area=%0.2f)'% (i,roc_auc))
    i +=1
#画对角线
plt.plot([0,1],[0,1],linestyle='--',lw=2,color='r',label='Luck',alpha=.8)
mean_tpr=np.mean(tprs,axis=0)
mean_tpr[-1]=1.0
mean_auc=auc(mean_fpr,mean_tpr)#计算平均AUC值
std_auc=np.std(tprs,axis=0)
plt.plot(mean_fpr,mean_tpr,color='b',label=r'Mean ROC (area=%0.2f)'%mean_auc,lw=2,alpha=.8)
std_tpr=np.std(tprs,axis=0)
tprs_upper=np.minimum(mean_tpr+std_tpr,1)
tprs_lower=np.maximum(mean_tpr-std_tpr,0)
plt.fill_between(mean_tpr,tprs_lower,tprs_upper,color='gray',alpha=.2)
plt.xlim([-0.05,1.05])
plt.ylim([-0.05,1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('ROC')
plt.legend(loc='lower right')
plt.show()

/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:9: UserWarning: Update your `Model` call to the Keras 2 API: `Model(outputs=Tensor("de..., inputs=Tensor("in...)`
  if __name__ == '__main__':


Train on 142 samples, validate on 62 samples
Epoch 1/30
142/142 [==============================] - 3s 22ms/step - loss: 0.7119 - acc: 0.4225 - val_loss: 0.7087 - val_acc: 0.5161
Epoch 2/30
142/142 [==============================] - 0s 949us/step - loss: 0.6651 - acc: 0.6127 - val_loss: 0.7267 - val_acc: 0.5161
Epoch 3/30
142/142 [==============================] - 0s 842us/step - loss: 0.6601 - acc: 0.6127 - val_loss: 0.7481 - val_acc: 0.5161
Epoch 4/30
142/142 [==============================] - 0s 824us/step - loss: 0.6549 - acc: 0.6127 - val_loss: 0.7354 - val_acc: 0.5161
Epoch 5/30
142/142 [==============================] - 0s 817us/step - loss: 0.6435 - acc: 0.6127 - val_loss: 0.7097 - val_acc: 0.5161
Epoch 6/30
142/142 [==============================] - 0s 792us/step - loss: 0.6344 - acc: 0.6127 - val_loss: 0.6866 - val_acc: 0.5161
Epoch 7/30
142/142 [==============================] - 0s 936us/step - loss: 0.6242 - acc: 0.6127 - val_loss: 0.6746 - val_acc: 0.5161
Epoch 8/30
142/142 [==============================] - 0s 946us/step - loss: 0.6148 - acc: 0.6127 - val_loss: 0.6711 - val_acc: 0.5161
Epoch 9/30
142/142 [==============================] - 0s 837us/step - loss: 0.6069 - acc: 0.6197 - val_loss: 0.6615 - val_acc: 0.5161
Epoch 10/30
142/142 [==============================] - 0s 845us/step - loss: 0.5974 - acc: 0.6127 - val_loss: 0.6600 - val_acc: 0.5161
Epoch 11/30
142/142 [==============================] - 0s 886us/step - loss: 0.5899 - acc: 0.6127 - val_loss: 0.6582 - val_acc: 0.5161
Epoch 12/30
142/142 [==============================] - 0s 812us/step - loss: 0.5822 - acc: 0.6127 - val_loss: 0.6386 - val_acc: 0.5323
Epoch 13/30
142/142 [==============================] - 0s 784us/step - loss: 0.5721 - acc: 0.6338 - val_loss: 0.6297 - val_acc: 0.5323
Epoch 14/30
142/142 [==============================] - 0s 838us/step - loss: 0.5633 - acc: 0.6479 - val_loss: 0.6222 - val_acc: 0.5323
Epoch 15/30
142/142 [==============================] - 0s 925us/step - loss: 0.5559 - acc: 0.6761 - val_loss: 0.6097 - val_acc: 0.5484
Epoch 16/30
142/142 [==============================] - 0s 923us/step - loss: 0.5475 - acc: 0.6831 - val_loss: 0.6046 - val_acc: 0.5484
Epoch 17/30
142/142 [==============================] - 0s 889us/step - loss: 0.5393 - acc: 0.6901 - val_loss: 0.5993 - val_acc: 0.5645
Epoch 18/30
142/142 [==============================] - 0s 882us/step - loss: 0.5313 - acc: 0.7042 - val_loss: 0.5878 - val_acc: 0.6290
Epoch 19/30
142/142 [==============================] - 0s 880us/step - loss: 0.5227 - acc: 0.7465 - val_loss: 0.5785 - val_acc: 0.6613
Epoch 20/30
142/142 [==============================] - 0s 883us/step - loss: 0.5154 - acc: 0.8028 - val_loss: 0.5693 - val_acc: 0.7097
Epoch 21/30
142/142 [==============================] - 0s 828us/step - loss: 0.5068 - acc: 0.8310 - val_loss: 0.5670 - val_acc: 0.6613
Epoch 22/30
142/142 [==============================] - 0s 855us/step - loss: 0.5019 - acc: 0.7465 - val_loss: 0.5679 - val_acc: 0.6129
Epoch 23/30
142/142 [==============================] - 0s 827us/step - loss: 0.4923 - acc: 0.7535 - val_loss: 0.5543 - val_acc: 0.6935
Epoch 24/30
142/142 [==============================] - 0s 844us/step - loss: 0.4839 - acc: 0.8310 - val_loss: 0.5448 - val_acc: 0.7581
Epoch 25/30
142/142 [==============================] - 0s 841us/step - loss: 0.4762 - acc: 0.8732 - val_loss: 0.5366 - val_acc: 0.7581
Epoch 26/30
142/142 [==============================] - 0s 816us/step - loss: 0.4688 - acc: 0.8803 - val_loss: 0.5284 - val_acc: 0.7581
Epoch 27/30
142/142 [==============================] - 0s 847us/step - loss: 0.4624 - acc: 0.9155 - val_loss: 0.5157 - val_acc: 0.8065
Epoch 28/30
142/142 [==============================] - 0s 1ms/step - loss: 0.4538 - acc: 0.9366 - val_loss: 0.5127 - val_acc: 0.7903
Epoch 29/30
142/142 [==============================] - 0s 815us/step - loss: 0.4465 - acc: 0.9155 - val_loss: 0.5098 - val_acc: 0.7581
Epoch 30/30
142/142 [==============================] - 0s 777us/step - loss: 0.4399 - acc: 0.9155 - val_loss: 0.5018 - val_acc: 0.7903
Train on 143 samples, validate on 62 samples
Epoch 1/30
143/143 [==============================] - 1s 4ms/step - loss: 0.7019 - acc: 0.4476 - val_loss: 0.6939 - val_acc: 0.4355
Epoch 2/30
143/143 [==============================] - 0s 863us/step - loss: 0.6875 - acc: 0.5315 - val_loss: 0.6905 - val_acc: 0.5484
Epoch 3/30
143/143 [==============================] - 0s 872us/step - loss: 0.6782 - acc: 0.6294 - val_loss: 0.6886 - val_acc: 0.5161
Epoch 4/30
143/143 [==============================] - 0s 835us/step - loss: 0.6674 - acc: 0.6154 - val_loss: 0.6870 - val_acc: 0.5161
Epoch 5/30
143/143 [==============================] - 0s 770us/step - loss: 0.6592 - acc: 0.6154 - val_loss: 0.6863 - val_acc: 0.5161
Epoch 6/30
143/143 [==============================] - 0s 839us/step - loss: 0.6504 - acc: 0.6154 - val_loss: 0.6852 - val_acc: 0.5161
Epoch 7/30
143/143 [==============================] - 0s 810us/step - loss: 0.6438 - acc: 0.6154 - val_loss: 0.6847 - val_acc: 0.5161
Epoch 8/30
143/143 [==============================] - 0s 840us/step - loss: 0.6373 - acc: 0.6154 - val_loss: 0.6827 - val_acc: 0.5161
Epoch 9/30
143/143 [==============================] - 0s 847us/step - loss: 0.6320 - acc: 0.6154 - val_loss: 0.6813 - val_acc: 0.5161
Epoch 10/30
143/143 [==============================] - 0s 810us/step - loss: 0.6260 - acc: 0.6154 - val_loss: 0.6773 - val_acc: 0.5161
Epoch 11/30
143/143 [==============================] - 0s 858us/step - loss: 0.6202 - acc: 0.6154 - val_loss: 0.6734 - val_acc: 0.5161
Epoch 12/30
143/143 [==============================] - 0s 830us/step - loss: 0.6142 - acc: 0.6154 - val_loss: 0.6686 - val_acc: 0.5161
Epoch 13/30
143/143 [==============================] - 0s 901us/step - loss: 0.6080 - acc: 0.6154 - val_loss: 0.6622 - val_acc: 0.5161
Epoch 14/30
143/143 [==============================] - 0s 907us/step - loss: 0.6022 - acc: 0.6154 - val_loss: 0.6550 - val_acc: 0.5161
Epoch 15/30
143/143 [==============================] - 0s 1ms/step - loss: 0.5956 - acc: 0.6154 - val_loss: 0.6507 - val_acc: 0.5161
Epoch 16/30
143/143 [==============================] - 0s 781us/step - loss: 0.5886 - acc: 0.6224 - val_loss: 0.6430 - val_acc: 0.5161
Epoch 17/30
143/143 [==============================] - 0s 874us/step - loss: 0.5820 - acc: 0.6224 - val_loss: 0.6359 - val_acc: 0.5323
Epoch 18/30
143/143 [==============================] - 0s 974us/step - loss: 0.5756 - acc: 0.6364 - val_loss: 0.6267 - val_acc: 0.5323
Epoch 19/30
143/143 [==============================] - 0s 949us/step - loss: 0.5682 - acc: 0.6503 - val_loss: 0.6200 - val_acc: 0.5323
Epoch 20/30
143/143 [==============================] - 0s 856us/step - loss: 0.5612 - acc: 0.6503 - val_loss: 0.6119 - val_acc: 0.5323
Epoch 21/30
143/143 [==============================] - 0s 826us/step - loss: 0.5539 - acc: 0.6573 - val_loss: 0.6047 - val_acc: 0.5323
Epoch 22/30
143/143 [==============================] - 0s 814us/step - loss: 0.5467 - acc: 0.6783 - val_loss: 0.5961 - val_acc: 0.5645
Epoch 23/30
143/143 [==============================] - 0s 849us/step - loss: 0.5391 - acc: 0.6923 - val_loss: 0.5890 - val_acc: 0.5645
Epoch 24/30
143/143 [==============================] - 0s 821us/step - loss: 0.5313 - acc: 0.7063 - val_loss: 0.5827 - val_acc: 0.5645
Epoch 25/30
143/143 [==============================] - 0s 848us/step - loss: 0.5238 - acc: 0.7133 - val_loss: 0.5754 - val_acc: 0.5806
Epoch 26/30
143/143 [==============================] - 0s 864us/step - loss: 0.5160 - acc: 0.7552 - val_loss: 0.5677 - val_acc: 0.6129
Epoch 27/30
143/143 [==============================] - 0s 832us/step - loss: 0.5085 - acc: 0.7832 - val_loss: 0.5591 - val_acc: 0.6774
Epoch 28/30
143/143 [==============================] - 0s 811us/step - loss: 0.5004 - acc: 0.8252 - val_loss: 0.5514 - val_acc: 0.6774
Epoch 29/30
143/143 [==============================] - 0s 855us/step - loss: 0.4924 - acc: 0.8252 - val_loss: 0.5440 - val_acc: 0.6935
Epoch 30/30
143/143 [==============================] - 0s 750us/step - loss: 0.4846 - acc: 0.8601 - val_loss: 0.5334 - val_acc: 0.7581
Train on 143 samples, validate on 62 samples
Epoch 1/30
143/143 [==============================] - 1s 4ms/step - loss: 0.6923 - acc: 0.5245 - val_loss: 0.6942 - val_acc: 0.4032
Epoch 2/30
143/143 [==============================] - 0s 678us/step - loss: 0.6903 - acc: 0.5524 - val_loss: 0.6944 - val_acc: 0.4355
Epoch 3/30
143/143 [==============================] - 0s 601us/step - loss: 0.6882 - acc: 0.6154 - val_loss: 0.6947 - val_acc: 0.5161
Epoch 4/30
143/143 [==============================] - 0s 640us/step - loss: 0.6861 - acc: 0.6224 - val_loss: 0.6949 - val_acc: 0.5000
Epoch 5/30
143/143 [==============================] - 0s 715us/step - loss: 0.6838 - acc: 0.6154 - val_loss: 0.6952 - val_acc: 0.5161
Epoch 6/30
143/143 [==============================] - 0s 684us/step - loss: 0.6817 - acc: 0.6154 - val_loss: 0.6954 - val_acc: 0.5161
Epoch 7/30
143/143 [==============================] - 0s 660us/step - loss: 0.6794 - acc: 0.6154 - val_loss: 0.6956 - val_acc: 0.5161
Epoch 8/30
143/143 [==============================] - 0s 706us/step - loss: 0.6770 - acc: 0.6154 - val_loss: 0.6958 - val_acc: 0.5161
Epoch 9/30
143/143 [==============================] - 0s 740us/step - loss: 0.6745 - acc: 0.6154 - val_loss: 0.6959 - val_acc: 0.5161
Epoch 10/30
143/143 [==============================] - 0s 621us/step - loss: 0.6723 - acc: 0.6154 - val_loss: 0.6962 - val_acc: 0.5161
Epoch 11/30
143/143 [==============================] - 0s 582us/step - loss: 0.6695 - acc: 0.6154 - val_loss: 0.6961 - val_acc: 0.5161
Epoch 12/30
143/143 [==============================] - 0s 581us/step - loss: 0.6672 - acc: 0.6154 - val_loss: 0.6962 - val_acc: 0.5161
Epoch 13/30
143/143 [==============================] - 0s 666us/step - loss: 0.6644 - acc: 0.6154 - val_loss: 0.6964 - val_acc: 0.5161
Epoch 14/30
143/143 [==============================] - 0s 665us/step - loss: 0.6616 - acc: 0.6154 - val_loss: 0.6966 - val_acc: 0.5161
Epoch 15/30
143/143 [==============================] - 0s 709us/step - loss: 0.6587 - acc: 0.6154 - val_loss: 0.6968 - val_acc: 0.5161
Epoch 16/30
143/143 [==============================] - 0s 692us/step - loss: 0.6560 - acc: 0.6154 - val_loss: 0.6972 - val_acc: 0.5161
Epoch 17/30
143/143 [==============================] - 0s 625us/step - loss: 0.6530 - acc: 0.6154 - val_loss: 0.6974 - val_acc: 0.5161
Epoch 18/30
143/143 [==============================] - 0s 650us/step - loss: 0.6502 - acc: 0.6154 - val_loss: 0.6979 - val_acc: 0.5161
Epoch 19/30
143/143 [==============================] - 0s 807us/step - loss: 0.6472 - acc: 0.6154 - val_loss: 0.6982 - val_acc: 0.5161
Epoch 20/30
143/143 [==============================] - 0s 639us/step - loss: 0.6441 - acc: 0.6154 - val_loss: 0.6984 - val_acc: 0.5161
Epoch 21/30
143/143 [==============================] - 0s 652us/step - loss: 0.6411 - acc: 0.6154 - val_loss: 0.6985 - val_acc: 0.5161
Epoch 22/30
143/143 [==============================] - ETA: 0s - loss: 0.6344 - acc: 0.632 - 0s 734us/step - loss: 0.6382 - acc: 0.6154 - val_loss: 0.6986 - val_acc: 0.5161
Epoch 23/30
143/143 [==============================] - 0s 687us/step - loss: 0.6349 - acc: 0.6154 - val_loss: 0.6982 - val_acc: 0.5161
Epoch 24/30
143/143 [==============================] - 0s 743us/step - loss: 0.6323 - acc: 0.6154 - val_loss: 0.6982 - val_acc: 0.5161
Epoch 25/30
143/143 [==============================] - 0s 747us/step - loss: 0.6290 - acc: 0.6154 - val_loss: 0.6976 - val_acc: 0.5161
Epoch 26/30
143/143 [==============================] - 0s 632us/step - loss: 0.6261 - acc: 0.6154 - val_loss: 0.6968 - val_acc: 0.5161
Epoch 27/30
143/143 [==============================] - 0s 864us/step - loss: 0.6229 - acc: 0.6154 - val_loss: 0.6956 - val_acc: 0.5161
Epoch 28/30
143/143 [==============================] - 0s 865us/step - loss: 0.6199 - acc: 0.6154 - val_loss: 0.6949 - val_acc: 0.5161
Epoch 29/30
143/143 [==============================] - 0s 661us/step - loss: 0.6167 - acc: 0.6154 - val_loss: 0.6943 - val_acc: 0.5161
Epoch 30/30
143/143 [==============================] - 0s 731us/step - loss: 0.6134 - acc: 0.6154 - val_loss: 0.6931 - val_acc: 0.5161
Train on 143 samples, validate on 62 samples
Epoch 1/30
143/143 [==============================] - 1s 4ms/step - loss: 0.6688 - acc: 0.6713 - val_loss: 0.7584 - val_acc: 0.5645
Epoch 2/30
143/143 [==============================] - 0s 898us/step - loss: 0.6615 - acc: 0.6713 - val_loss: 0.7669 - val_acc: 0.5645
Epoch 3/30
143/143 [==============================] - 0s 870us/step - loss: 0.6502 - acc: 0.6713 - val_loss: 0.7273 - val_acc: 0.5645
Epoch 4/30
143/143 [==============================] - 0s 908us/step - loss: 0.6400 - acc: 0.6713 - val_loss: 0.7230 - val_acc: 0.5645
Epoch 5/30
143/143 [==============================] - 0s 838us/step - loss: 0.6340 - acc: 0.6713 - val_loss: 0.7121 - val_acc: 0.5645
Epoch 6/30
143/143 [==============================] - 0s 828us/step - loss: 0.6266 - acc: 0.6713 - val_loss: 0.7065 - val_acc: 0.5645
Epoch 7/30
143/143 [==============================] - 0s 816us/step - loss: 0.6202 - acc: 0.6713 - val_loss: 0.7013 - val_acc: 0.5645
Epoch 8/30
143/143 [==============================] - 0s 807us/step - loss: 0.6169 - acc: 0.6713 - val_loss: 0.6839 - val_acc: 0.5645
Epoch 9/30
143/143 [==============================] - 0s 777us/step - loss: 0.6090 - acc: 0.6713 - val_loss: 0.6784 - val_acc: 0.5645
Epoch 10/30
143/143 [==============================] - 0s 841us/step - loss: 0.5996 - acc: 0.6713 - val_loss: 0.6886 - val_acc: 0.5645
Epoch 11/30
143/143 [==============================] - 0s 1ms/step - loss: 0.5948 - acc: 0.6713 - val_loss: 0.6986 - val_acc: 0.5645
Epoch 12/30
143/143 [==============================] - 0s 814us/step - loss: 0.5902 - acc: 0.6713 - val_loss: 0.6866 - val_acc: 0.5645
Epoch 13/30
143/143 [==============================] - 0s 885us/step - loss: 0.5836 - acc: 0.6713 - val_loss: 0.6792 - val_acc: 0.5645
Epoch 14/30
143/143 [==============================] - 0s 815us/step - loss: 0.5744 - acc: 0.6713 - val_loss: 0.6522 - val_acc: 0.5645
Epoch 15/30
143/143 [==============================] - 0s 833us/step - loss: 0.5708 - acc: 0.6713 - val_loss: 0.6340 - val_acc: 0.5645
Epoch 16/30
143/143 [==============================] - 0s 875us/step - loss: 0.5650 - acc: 0.6713 - val_loss: 0.6322 - val_acc: 0.5645
Epoch 17/30
143/143 [==============================] - 0s 801us/step - loss: 0.5576 - acc: 0.6713 - val_loss: 0.6350 - val_acc: 0.5645
Epoch 18/30
143/143 [==============================] - 0s 877us/step - loss: 0.5514 - acc: 0.6713 - val_loss: 0.6381 - val_acc: 0.5645
Epoch 19/30
143/143 [==============================] - 0s 822us/step - loss: 0.5472 - acc: 0.6713 - val_loss: 0.6347 - val_acc: 0.5645
Epoch 20/30
143/143 [==============================] - 0s 886us/step - loss: 0.5398 - acc: 0.6713 - val_loss: 0.6198 - val_acc: 0.5645
Epoch 21/30
143/143 [==============================] - 0s 1ms/step - loss: 0.5364 - acc: 0.6713 - val_loss: 0.5980 - val_acc: 0.5645
Epoch 22/30
143/143 [==============================] - 0s 832us/step - loss: 0.5302 - acc: 0.6853 - val_loss: 0.5905 - val_acc: 0.5806
Epoch 23/30
143/143 [==============================] - 0s 852us/step - loss: 0.5255 - acc: 0.7063 - val_loss: 0.5849 - val_acc: 0.6129
Epoch 24/30
143/143 [==============================] - 0s 828us/step - loss: 0.5172 - acc: 0.6783 - val_loss: 0.5946 - val_acc: 0.5645
Epoch 25/30
143/143 [==============================] - 0s 822us/step - loss: 0.5129 - acc: 0.6713 - val_loss: 0.6004 - val_acc: 0.5645
Epoch 26/30
143/143 [==============================] - 0s 892us/step - loss: 0.5067 - acc: 0.6713 - val_loss: 0.5862 - val_acc: 0.5645
Epoch 27/30
143/143 [==============================] - 0s 854us/step - loss: 0.5004 - acc: 0.6923 - val_loss: 0.5652 - val_acc: 0.6613
Epoch 28/30
143/143 [==============================] - 0s 859us/step - loss: 0.4977 - acc: 0.7902 - val_loss: 0.5542 - val_acc: 0.6935
Epoch 29/30
143/143 [==============================] - 0s 874us/step - loss: 0.4906 - acc: 0.8182 - val_loss: 0.5558 - val_acc: 0.6935
Epoch 30/30
143/143 [==============================] - 0s 1ms/step - loss: 0.4842 - acc: 0.7552 - val_loss: 0.5563 - val_acc: 0.6774
Train on 143 samples, validate on 62 samples
Epoch 1/30
143/143 [==============================] - 1s 4ms/step - loss: 0.6682 - acc: 0.6643 - val_loss: 0.7258 - val_acc: 0.4839
Epoch 2/30
143/143 [==============================] - 0s 774us/step - loss: 0.6358 - acc: 0.6713 - val_loss: 0.7755 - val_acc: 0.4839
Epoch 3/30
143/143 [==============================] - 0s 786us/step - loss: 0.6300 - acc: 0.6713 - val_loss: 0.7771 - val_acc: 0.4839
Epoch 4/30
143/143 [==============================] - 0s 713us/step - loss: 0.6253 - acc: 0.6713 - val_loss: 0.7634 - val_acc: 0.4839
Epoch 5/30
143/143 [==============================] - 0s 811us/step - loss: 0.6153 - acc: 0.6713 - val_loss: 0.7258 - val_acc: 0.4839
Epoch 6/30
143/143 [==============================] - 0s 817us/step - loss: 0.6084 - acc: 0.6713 - val_loss: 0.7010 - val_acc: 0.4839
Epoch 7/30
143/143 [==============================] - 0s 957us/step - loss: 0.6037 - acc: 0.6713 - val_loss: 0.6951 - val_acc: 0.4839
Epoch 8/30
143/143 [==============================] - 0s 748us/step - loss: 0.5958 - acc: 0.6713 - val_loss: 0.6969 - val_acc: 0.4839
Epoch 9/30
143/143 [==============================] - 0s 842us/step - loss: 0.5884 - acc: 0.6713 - val_loss: 0.7086 - val_acc: 0.4839
Epoch 10/30
143/143 [==============================] - 0s 826us/step - loss: 0.5883 - acc: 0.6713 - val_loss: 0.7194 - val_acc: 0.4839
Epoch 11/30
143/143 [==============================] - 0s 850us/step - loss: 0.5780 - acc: 0.6713 - val_loss: 0.6947 - val_acc: 0.4839
Epoch 12/30
143/143 [==============================] - 0s 784us/step - loss: 0.5709 - acc: 0.6713 - val_loss: 0.6670 - val_acc: 0.4839
Epoch 13/30
143/143 [==============================] - 0s 853us/step - loss: 0.5633 - acc: 0.6713 - val_loss: 0.6605 - val_acc: 0.4839
Epoch 14/30
143/143 [==============================] - 0s 1ms/step - loss: 0.5571 - acc: 0.6713 - val_loss: 0.6549 - val_acc: 0.4839
Epoch 15/30
143/143 [==============================] - 0s 811us/step - loss: 0.5523 - acc: 0.6713 - val_loss: 0.6401 - val_acc: 0.4839
Epoch 16/30
143/143 [==============================] - 0s 865us/step - loss: 0.5447 - acc: 0.6713 - val_loss: 0.6428 - val_acc: 0.4839
Epoch 17/30
143/143 [==============================] - 0s 827us/step - loss: 0.5383 - acc: 0.6713 - val_loss: 0.6447 - val_acc: 0.4839
Epoch 18/30
143/143 [==============================] - 0s 817us/step - loss: 0.5322 - acc: 0.6713 - val_loss: 0.6335 - val_acc: 0.4839
Epoch 19/30
143/143 [==============================] - 0s 834us/step - loss: 0.5256 - acc: 0.6713 - val_loss: 0.6301 - val_acc: 0.4839
Epoch 20/30
143/143 [==============================] - 0s 822us/step - loss: 0.5198 - acc: 0.6783 - val_loss: 0.6132 - val_acc: 0.5000
Epoch 21/30
143/143 [==============================] - 0s 847us/step - loss: 0.5140 - acc: 0.6783 - val_loss: 0.6141 - val_acc: 0.4839
Epoch 22/30
143/143 [==============================] - 0s 835us/step - loss: 0.5096 - acc: 0.6783 - val_loss: 0.5978 - val_acc: 0.5161
Epoch 23/30
143/143 [==============================] - 0s 706us/step - loss: 0.5015 - acc: 0.6923 - val_loss: 0.5930 - val_acc: 0.5323
Epoch 24/30
143/143 [==============================] - 0s 686us/step - loss: 0.4947 - acc: 0.6853 - val_loss: 0.6010 - val_acc: 0.5000
Epoch 25/30
143/143 [==============================] - 0s 882us/step - loss: 0.4907 - acc: 0.6783 - val_loss: 0.6018 - val_acc: 0.4839
Epoch 26/30
143/143 [==============================] - 0s 797us/step - loss: 0.4854 - acc: 0.6783 - val_loss: 0.5906 - val_acc: 0.5000
Epoch 27/30
143/143 [==============================] - 0s 857us/step - loss: 0.4787 - acc: 0.6853 - val_loss: 0.5746 - val_acc: 0.6129
Epoch 28/30
143/143 [==============================] - 0s 821us/step - loss: 0.4726 - acc: 0.7483 - val_loss: 0.5589 - val_acc: 0.7097
Epoch 29/30
143/143 [==============================] - 0s 808us/step - loss: 0.4672 - acc: 0.8112 - val_loss: 0.5495 - val_acc: 0.7742
Epoch 30/30
143/143 [==============================] - 0s 825us/step - loss: 0.4624 - acc: 0.8531 - val_loss: 0.5401 - val_acc: 0.8065

#[! png

y_pred = model.predict(X_test)

y_pred[y_pred < 0.5] = 0
y_pred[y_pred > 0.5] = 1

y_pred - Y_test

array([[ 1., -1.],
       [ 0.,  0.],
       [ 1., -1.],
       [ 1., -1.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 1., -1.],
       [ 1., -1.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 1., -1.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 1., -1.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 1., -1.],
       [ 1., -1.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.],
       [ 0.,  0.]])

Y_test

array([[0, 1],
       [1, 0],
       [0, 1],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [0, 1],
       [1, 0],
       [0, 1],
       [1, 0],
       [0, 1],
       [0, 1],
       [1, 0],
       [0, 1],
       [1, 0],
       [0, 1],
       [1, 0],
       [0, 1],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0],
       [0, 1],
       [0, 1],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [1, 0],
       [0, 1],
       [0, 1],
       [1, 0],
       [1, 0]])

目标：利用长短期记忆网络实现AD诊断
使用了数据增强
数据集划分：共得到1024个样本数据，820个用于训练，204个用于测试
结果：训练30次后，测试集准确度为84.96%
在这里插入图片描述测试结果：
利用测试集数据进行测试，204例数据可以正确识别出156例。
测试集准确率为76.47%

#对时间序列数据进行样本扩充
#方法:先划分数据集,再利用nitime进行滤波
#https://mp.weixin.qq.com/s?__biz=MzU1MTkwNzIyOQ==&mid=2247486561&idx=1&sn=91df24bc64059f785856fc8f82945f58&chksm=fb8b7693ccfcff853936faccf0b42409c3dd7b2cef416fc623e051d4d81ee85d78702bd20325&scene=21#wechat_redirect

#对测试集进行样本扩充

#利用3种滤波方式增强数据

#训练集和验证集分开增强
#五折交叉验证怎样划分数据?-不使用交叉验证

网络搭建：

在这里插入图片描述

import warnings
warnings.filterwarnings("ignore")
import matplotlib
from matplotlib import pylab as plt
import nibabel as nib
import os
import numpy as np
import cv2
import keras
from keras.layers import Conv3D,Conv2D,Dense,Input,MaxPool2D,MaxPool3D,Flatten
from keras.models import Model
import os
import numpy as np
from nilearn import datasets
from nilearn import input_data
import cv2
from sklearn.utils import shuffle
from keras import models
from keras.models import load_model
import nilearn
from nilearn.image import new_img_like
from nilearn import datasets
from nilearn import input_data
from nilearn.connectome import ConnectivityMeasure
import matplotlib.pyplot as plt
from nilearn import image
import nibabel as nib
import skimage.io as io
import numpy as np
from nilearn import datasets,plotting,image
import random
#nilearn:一个用于神经影像的机器学习工具包
#nibabel:用于读取nii文件,提取图像数据的工具包
#numpy:用于进行矩阵操作
#skimage:用于进行图像处理的模块
import keras
from keras.layers import Conv3D,Conv2D,Dense,Input,MaxPool2D,MaxPool3D,Flatten
from keras.models import Model
from keras.layers import Dense,Input,Add,LSTM,Flatten,GRU
from keras.optimizers import Adam,SGD
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.layers import concatenate
from keras.layers import Conv1D, GlobalAveragePooling1D, MaxPooling1D
from keras.layers import Conv2D, GlobalAveragePooling2D, MaxPooling2D
from keras.layers import TimeDistributed
from keras.layers import BatchNormalization
from keras import regularizers
from keras.layers.core import Dense, Dropout, Activation, Flatten
from keras.optimizers import Adam,RMSprop
from keras.layers import Dense,Input,Add,LSTM,Flatten,GRU

import nitime
# Import the time-series objects:
from nitime.timeseries import TimeSeries
# Import the analysis objects:
from nitime.analysis import SpectralAnalyzer, FilterAnalyzer, NormalizationAnalyzer

Using TensorFlow backend.
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
/usr/local/lib/python3.5/dist-packages/skimage/__init__.py:71: ResourceWarning: unclosed file <_io.TextIOWrapper name='/usr/local/lib/python3.5/dist-packages/pytest.py' mode='r' encoding='utf-8'>
  imp.find_module('pytest')
/usr/local/lib/python3.5/dist-packages/nilearn/plotting/__init__.py:20: UserWarning: 
This call to matplotlib.use() has no effect because the backend has already
been chosen; matplotlib.use() must be called *before* pylab, matplotlib.pyplot,
or matplotlib.backends is imported for the first time.

The backend was *originally* set to 'module://ipykernel.pylab.backend_inline' by the following code:
  File "/usr/lib/python3.5/runpy.py", line 184, in _run_module_as_main
    "__main__", mod_spec)
  File "/usr/lib/python3.5/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py", line 16, in <module>
    app.launch_new_instance()
  File "/usr/local/lib/python3.5/dist-packages/traitlets/config/application.py", line 658, in launch_instance
    app.start()
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelapp.py", line 486, in start
    self.io_loop.start()
  File "/usr/local/lib/python3.5/dist-packages/tornado/platform/asyncio.py", line 132, in start
    self.asyncio_loop.run_forever()
  File "/usr/lib/python3.5/asyncio/base_events.py", line 345, in run_forever
    self._run_once()
  File "/usr/lib/python3.5/asyncio/base_events.py", line 1312, in _run_once
    handle._run()
  File "/usr/lib/python3.5/asyncio/events.py", line 125, in _run
    self._callback(*self._args)
  File "/usr/local/lib/python3.5/dist-packages/tornado/ioloop.py", line 758, in _run_callback
    ret = callback()
  File "/usr/local/lib/python3.5/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
    return fn(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 536, in <lambda>
    self.io_loop.add_callback(lambda : self._handle_events(self.socket, 0))
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 450, in _handle_events
    self._handle_recv()
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 480, in _handle_recv
    self._run_callback(callback, msg)
  File "/usr/local/lib/python3.5/dist-packages/zmq/eventloop/zmqstream.py", line 432, in _run_callback
    callback(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
    return fn(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 283, in dispatcher
    return self.dispatch_shell(stream, msg)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 233, in dispatch_shell
    handler(stream, idents, msg)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/kernelbase.py", line 399, in execute_request
    user_expressions, allow_stdin)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/ipkernel.py", line 208, in do_execute
    res = shell.run_cell(code, store_history=store_history, silent=silent)
  File "/usr/local/lib/python3.5/dist-packages/ipykernel/zmqshell.py", line 537, in run_cell
    return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2662, in run_cell
    raw_cell, store_history, silent, shell_futures)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2785, in _run_cell
    interactivity=interactivity, compiler=compiler, result=result)
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2901, in run_ast_nodes
    if self.run_code(code, result):
  File "/usr/local/lib/python3.5/dist-packages/IPython/core/interactiveshell.py", line 2961, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-5-a2293080d056>", line 4, in <module>
    from matplotlib import pylab as plt
  File "/usr/local/lib/python3.5/dist-packages/matplotlib/pylab.py", line 252, in <module>
    from matplotlib import cbook, mlab, pyplot as plt
  File "/usr/local/lib/python3.5/dist-packages/matplotlib/pyplot.py", line 71, in <module>
    from matplotlib.backends import pylab_setup
  File "/usr/local/lib/python3.5/dist-packages/matplotlib/backends/__init__.py", line 16, in <module>
    line for line in traceback.format_stack()


  matplotlib.use('Agg')
/usr/lib/python3.5/importlib/_bootstrap.py:222: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)

#从本地载入文件
#打乱过label的数据
data = np.load('AD_CN_102_154_LSTM_data_1.npy')
label = np.load('AD_CN_102_154_LSTM_label.npy')
data = np.squeeze(data)

#先做标准化再划分数据集
from sklearn import preprocessing
from sklearn.preprocessing import MinMaxScaler

num_data = 256
minmax = preprocessing.MinMaxScaler()

for i in range(num_data):
    data[i] = minmax.fit_transform(data[i])

#标准化处理2
from sklearn import preprocessing
from sklearn.preprocessing import MaxAbsScaler

num_data = 256
maxabs = preprocessing.MaxAbsScaler()

for i in range(num_data):
    data[i] = maxabs.fit_transform(data[i])

#标准化处理3
from sklearn import preprocessing
from sklearn.preprocessing import StandardScaler

num_data = 256
standard = preprocessing.StandardScaler()

for i in range(num_data):
    data[i] = standard.fit_transform(data[i])

#划分数据集
train_num = 205

data_train = data[0:train_num,:,:]
label_train = label[0:train_num,:]
data_test = data[train_num:data.shape[0],:,:]
label_test = label[train_num:data.shape[0],:]

#滤波变换示意图
#非必需
TR = 1.4
plot_mun = 2
T = TimeSeries(data_train[plot_mun], sampling_interval=TR)
F = FilterAnalyzer(T, ub=0.15, lb=0.02)
fig01 = plt.figure()
ax01 = fig01.add_subplot(1, 1, 1)

ax01.plot(F.data[plot_mun],label='unfiltered')
ax01.plot(F.filtered_fourier.data[plot_mun],label='Fourier')
ax01.plot(F.iir.data[plot_mun],label='IIR')
ax01.plot(F.filtered_boxcar.data[plot_mun],label='Boxcar filter')


ax01.set_xlabel('Time (TR)')
ax01.set_ylabel('Signal amplitude (a.u.)')

ax01.legend()

/usr/local/lib/python3.5/dist-packages/nitime/utils.py:1028: DeprecationWarning: object of type <class 'float'> cannot be safely interpreted as an integer.
  return np.linspace(0, float(Fs) / 2, float(n) / 2 + 1)
/usr/local/lib/python3.5/dist-packages/scipy/signal/filter_design.py:3462: RuntimeWarning: divide by zero encountered in true_divide
  (stopb * (passb[0] - passb[1])))
/usr/local/lib/python3.5/dist-packages/scipy/signal/_arraytools.py:45: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  b = a[a_slice]





<matplotlib.legend.Legend at 0x7fcf97d2f5f8>

data_train.shape

(205, 140, 116)

data_aug = []
label_aug = []
num_train = 205
for i in range(num_train):
    T = TimeSeries(data_train[i], sampling_interval=TR)
    F = FilterAnalyzer(T, ub=0.15, lb=0.02)
    data_aug.append(F.filtered_fourier.data)
    label_aug.append(label_train[i])
    data_aug.append(F.iir.data)
    label_aug.append(label_train[i])
    data_aug.append(F.filtered_boxcar.data)
    label_aug.append(label_train[i])

/usr/local/lib/python3.5/dist-packages/nitime/utils.py:1028: DeprecationWarning: object of type <class 'float'> cannot be safely interpreted as an integer.
  return np.linspace(0, float(Fs) / 2, float(n) / 2 + 1)
/usr/local/lib/python3.5/dist-packages/scipy/signal/filter_design.py:3462: RuntimeWarning: divide by zero encountered in true_divide
  (stopb * (passb[0] - passb[1])))
/usr/local/lib/python3.5/dist-packages/scipy/signal/_arraytools.py:45: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  b = a[a_slice]

data_aug = np.array(data_aug)
label_aug = np.array(label_aug)

data_aug.shape,label_aug.shape

((615, 140, 116), (615, 2))

data_all = np.concatenate((data_train,data_aug),axis = 0)
label_all = np.concatenate((label_train,label_aug),axis = 0)
data_all = np.expand_dims(data_all, axis = 3)

data_all.shape,label_all.shape

((820, 140, 116, 1), (820, 2))

测试集数据增强

data_test_aug = []
label_test_aug = []
num_test = 51
for i in range(num_test):
    T = TimeSeries(data_test[i], sampling_interval=TR)
    F = FilterAnalyzer(T, ub=0.15, lb=0.02)
    data_test_aug.append(F.filtered_fourier.data)
    label_test_aug.append(label_test[i])
    data_test_aug.append(F.iir.data)
    label_test_aug.append(label_test[i])
    data_test_aug.append(F.filtered_boxcar.data)
    label_test_aug.append(label_test[i])
#转换格式
data_test_aug = np.array(data_test_aug)
label_test_aug = np.array(label_test_aug)
#拼接
data_test_all = np.concatenate((data_test,data_test_aug),axis = 0)
label_test_all = np.concatenate((label_test,label_test_aug),axis = 0)
data_test_all = np.expand_dims(data_test_all, axis = 3)
#打乱
data_test_all,label_test_all = shuffle(data_test_all,label_test_all)
data_test_all.shape,label_test_all.shape

/usr/local/lib/python3.5/dist-packages/nitime/utils.py:1028: DeprecationWarning: object of type <class 'float'> cannot be safely interpreted as an integer.
  return np.linspace(0, float(Fs) / 2, float(n) / 2 + 1)
/usr/local/lib/python3.5/dist-packages/scipy/signal/filter_design.py:3462: RuntimeWarning: divide by zero encountered in true_divide
  (stopb * (passb[0] - passb[1])))
/usr/local/lib/python3.5/dist-packages/scipy/signal/_arraytools.py:45: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.
  b = a[a_slice]





((204, 140, 116, 1), (204, 2))

#画结果图用
#写一个LossHistory类，保存loss和acc
class LossHistory(keras.callbacks.Callback):
    def on_train_begin(self, logs={}):
        self.losses = {'batch':[], 'epoch':[]}
        self.accuracy = {'batch':[], 'epoch':[]}
        self.val_loss = {'batch':[], 'epoch':[]}
        self.val_acc = {'batch':[], 'epoch':[]}

    def on_batch_end(self, batch, logs={}):
        self.losses['batch'].append(logs.get('loss'))
        self.accuracy['batch'].append(logs.get('acc'))
        self.val_loss['batch'].append(logs.get('val_loss'))
        self.val_acc['batch'].append(logs.get('val_acc'))

    def on_epoch_end(self, batch, logs={}):
        self.losses['epoch'].append(logs.get('loss'))
        self.accuracy['epoch'].append(logs.get('acc'))
        self.val_loss['epoch'].append(logs.get('val_loss'))
        self.val_acc['epoch'].append(logs.get('val_acc'))

    def loss_plot(self, loss_type):
        iters = range(len(self.losses[loss_type]))
        plt.figure()
        # acc
        plt.plot(iters, self.accuracy[loss_type], 'r', label='train acc')
        # loss
        plt.plot(iters, self.losses[loss_type], 'g', label='train loss')
        if loss_type == 'epoch':
            # val_acc
            plt.plot(iters, self.val_acc[loss_type], 'b', label='val acc')
            # val_loss
            plt.plot(iters, self.val_loss[loss_type], 'k', label='val loss')
        plt.grid(True)
        plt.xlabel(loss_type)
        plt.ylabel('acc-loss')
        plt.legend(loc="upper right")
        plt.show()

size1 = 140
size2 = 116
model = Sequential()
model.add(TimeDistributed(Conv1D(filters=2,kernel_size=2,activation='relu'), input_shape=(size1,size2,1)))
model.add(TimeDistributed(Flatten()))
model.add(LSTM(16,return_sequences=True))
model.add(Flatten())
model.add(Dropout(0.8))
model.add(Dense(2, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['acc'])
print(model.summary())

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
time_distributed_1 (TimeDist (None, 140, 115, 2)       6         
_________________________________________________________________
time_distributed_2 (TimeDist (None, 140, 230)          0         
_________________________________________________________________
lstm_1 (LSTM)                (None, 140, 16)           15808     
_________________________________________________________________
flatten_2 (Flatten)          (None, 2240)              0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 2240)              0         
_________________________________________________________________
dense_1 (Dense)              (None, 2)                 4482      
=================================================================
Total params: 20,296
Trainable params: 20,296
Non-trainable params: 0
_________________________________________________________________
None

#创建一个实例history
history = LossHistory()

X, y = data_all,label_all
model.fit(X, y, batch_size=16, epochs=30,validation_split = 0.3,callbacks=[history])
#绘制acc-loss曲线
history.loss_plot('epoch')

Train on 574 samples, validate on 246 samples
Epoch 1/30
574/574 [==============================] - 30s 53ms/step - loss: 0.7603 - acc: 0.5906 - val_loss: 0.6899 - val_acc: 0.6098
Epoch 2/30
574/574 [==============================] - 25s 44ms/step - loss: 0.6172 - acc: 0.6672 - val_loss: 0.6334 - val_acc: 0.6260
Epoch 3/30
574/574 [==============================] - 24s 42ms/step - loss: 0.5209 - acc: 0.7404 - val_loss: 0.5912 - val_acc: 0.6545
Epoch 4/30
574/574 [==============================] - 27s 47ms/step - loss: 0.4298 - acc: 0.8188 - val_loss: 0.5260 - val_acc: 0.7398
Epoch 5/30
574/574 [==============================] - 26s 46ms/step - loss: 0.3918 - acc: 0.8223 - val_loss: 0.4847 - val_acc: 0.7805
Epoch 6/30
574/574 [==============================] - 27s 47ms/step - loss: 0.3411 - acc: 0.8554 - val_loss: 0.4703 - val_acc: 0.8089
Epoch 7/30
574/574 [==============================] - 28s 48ms/step - loss: 0.3219 - acc: 0.8746 - val_loss: 0.4335 - val_acc: 0.8008
Epoch 8/30
574/574 [==============================] - 26s 46ms/step - loss: 0.2235 - acc: 0.9077 - val_loss: 0.4634 - val_acc: 0.7683
Epoch 9/30
574/574 [==============================] - 27s 47ms/step - loss: 0.2467 - acc: 0.8972 - val_loss: 0.4185 - val_acc: 0.8171
Epoch 10/30
574/574 [==============================] - 26s 46ms/step - loss: 0.1952 - acc: 0.9268 - val_loss: 0.3867 - val_acc: 0.8374
Epoch 11/30
574/574 [==============================] - 26s 46ms/step - loss: 0.2236 - acc: 0.9199 - val_loss: 0.3699 - val_acc: 0.8130
Epoch 12/30
574/574 [==============================] - 27s 46ms/step - loss: 0.1392 - acc: 0.9460 - val_loss: 0.3485 - val_acc: 0.8374
Epoch 13/30
574/574 [==============================] - 26s 46ms/step - loss: 0.1197 - acc: 0.9582 - val_loss: 0.3428 - val_acc: 0.8333
Epoch 14/30
574/574 [==============================] - 27s 47ms/step - loss: 0.0972 - acc: 0.9652 - val_loss: 0.3566 - val_acc: 0.8333
Epoch 15/30
574/574 [==============================] - 26s 46ms/step - loss: 0.0935 - acc: 0.9617 - val_loss: 0.3420 - val_acc: 0.8374
Epoch 16/30
574/574 [==============================] - 27s 47ms/step - loss: 0.0624 - acc: 0.9739 - val_loss: 0.3002 - val_acc: 0.8618
Epoch 17/30
574/574 [==============================] - 27s 46ms/step - loss: 0.0699 - acc: 0.9791 - val_loss: 0.3374 - val_acc: 0.8455
Epoch 18/30
574/574 [==============================] - 27s 47ms/step - loss: 0.0633 - acc: 0.9756 - val_loss: 0.2348 - val_acc: 0.8740
Epoch 19/30
574/574 [==============================] - 26s 45ms/step - loss: 0.0353 - acc: 0.9878 - val_loss: 0.2971 - val_acc: 0.8740
Epoch 20/30
574/574 [==============================] - 28s 48ms/step - loss: 0.0646 - acc: 0.9721 - val_loss: 0.2745 - val_acc: 0.8699
Epoch 21/30
574/574 [==============================] - 27s 48ms/step - loss: 0.0802 - acc: 0.9669 - val_loss: 0.3214 - val_acc: 0.8577
Epoch 22/30
574/574 [==============================] - 26s 46ms/step - loss: 0.0647 - acc: 0.9774 - val_loss: 0.3369 - val_acc: 0.8577
Epoch 23/30
574/574 [==============================] - 26s 46ms/step - loss: 0.0622 - acc: 0.9791 - val_loss: 0.3870 - val_acc: 0.8415
Epoch 24/30
574/574 [==============================] - 26s 45ms/step - loss: 0.0506 - acc: 0.9791 - val_loss: 0.2629 - val_acc: 0.8780
Epoch 25/30
574/574 [==============================] - 24s 42ms/step - loss: 0.0390 - acc: 0.9861 - val_loss: 0.3160 - val_acc: 0.8618
Epoch 26/30
574/574 [==============================] - 23s 41ms/step - loss: 0.0254 - acc: 0.9930 - val_loss: 0.4207 - val_acc: 0.8415
Epoch 27/30
574/574 [==============================] - 23s 40ms/step - loss: 0.0351 - acc: 0.9895 - val_loss: 0.3660 - val_acc: 0.8699
Epoch 28/30
574/574 [==============================] - 22s 39ms/step - loss: 0.0477 - acc: 0.9791 - val_loss: 0.4207 - val_acc: 0.8659
Epoch 29/30
574/574 [==============================] - 23s 40ms/step - loss: 0.0419 - acc: 0.9843 - val_loss: 0.4053 - val_acc: 0.8618
Epoch 30/30
574/574 [==============================] - 23s 39ms/step - loss: 0.0371 - acc: 0.9861 - val_loss: 0.4967 - val_acc: 0.8496

#model.save('model_time_series_dataaugmentation.h5')

label_pred = model.predict(data_test_all)

label_pred[label_pred > 0.5] = 1
label_pred[label_pred < 0.5] = 0

from collections import Counter
x = ((label_pred - label_test_all)[:,0])
x = x.tolist()
x.count(0)
print(x.count(0))
acc = (x.count(0))/204
print(acc)

156
0.7647058823529411

#150/204 73.53%
#LSTM层输出参数设置为16：156/204 76.47%