MRI / CT 를 찍게 되면 조영제(Contrast)를 사용하게 되는데 사용한 그림인지 아닌지 구분하는 코드.

CT(Iamages using Contrast Classification CNN with Keras(tensorflow))

  • CT images
  • Total images : 100
  • Training Dataset: 90
  • Test Dataset: 10
import numpy as np # matrix tools
import matplotlib.pyplot as plt # for basic plots
import seaborn as sns # for nicer plots
import pandas as pd
from glob import glob
import re
from skimage.io import imread
import os
import keras

Using TensorFlow backend.

%matplotlib inline

Exploratory data analysis

overview = pd.read_csv('data/overview.csv')
overview.head()
Age Contrast ContrastTag raw_input_path id tiff_name dicom_name
0 60 True NONE ../data/50_50_dicom_cases\Contrast\00001 (1).dcm 0 ID_0000_AGE_0060_CONTRAST_1_CT.tif ID_0000_AGE_0060_CONTRAST_1_CT.dcm
1 69 True NONE ../data/50_50_dicom_cases\Contrast\00001 (10).dcm 1 ID_0001_AGE_0069_CONTRAST_1_CT.tif ID_0001_AGE_0069_CONTRAST_1_CT.dcm
2 74 True APPLIED ../data/50_50_dicom_cases\Contrast\00001 (11).dcm 2 ID_0002_AGE_0074_CONTRAST_1_CT.tif ID_0002_AGE_0074_CONTRAST_1_CT.dcm
3 75 True NONE ../data/50_50_dicom_cases\Contrast\00001 (12).dcm 3 ID_0003_AGE_0075_CONTRAST_1_CT.tif ID_0003_AGE_0075_CONTRAST_1_CT.dcm
4 56 True NONE ../data/50_50_dicom_cases\Contrast\00001 (13).dcm 4 ID_0004_AGE_0056_CONTRAST_1_CT.tif ID_0004_AGE_0056_CONTRAST_1_CT.dcm 
len(overview)

100

overview['Contrast'] = overview['Contrast'].map(lambda x: 1 if x else 0)

plt.figure(figsize=(10,5))
sns.distplot(overview['Age'])

<matplotlib.axes._subplots.AxesSubplot at 0x17aac249f28>

png

g = sns.FacetGrid(overview, col="Contrast", size=8)
g = g.map(sns.distplot, "Age")

png

g = sns.FacetGrid(overview, hue="Contrast",size=6, legend_out=True)
g = g.map(sns.distplot, "Age").add_legend()

png

Read Image Files

BASE_IMAGE_PATH = os.path.join("data","tiff_images")
all_images_list = glob(os.path.join(BASE_IMAGE_PATH,"*.tif"))
all_images_list[:5]

['data\\tiff_images\\ID_0000_AGE_0060_CONTRAST_1_CT.tif',
 'data\\tiff_images\\ID_0001_AGE_0069_CONTRAST_1_CT.tif',
 'data\\tiff_images\\ID_0002_AGE_0074_CONTRAST_1_CT.tif',
 'data\\tiff_images\\ID_0003_AGE_0075_CONTRAST_1_CT.tif',
 'data\\tiff_images\\ID_0004_AGE_0056_CONTRAST_1_CT.tif']
  • np.expand_dims -> Expand the shape of an array.
  • Insert a new axis, corresponding to a given position in the array shape.
imread(all_images_list[0]).shape

(512, 512)

np.array(np.arange(81)).reshape(9,9)

array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8],
       [ 9, 10, 11, 12, 13, 14, 15, 16, 17],
       [18, 19, 20, 21, 22, 23, 24, 25, 26],
       [27, 28, 29, 30, 31, 32, 33, 34, 35],
       [36, 37, 38, 39, 40, 41, 42, 43, 44],
       [45, 46, 47, 48, 49, 50, 51, 52, 53],
       [54, 55, 56, 57, 58, 59, 60, 61, 62],
       [63, 64, 65, 66, 67, 68, 69, 70, 71],
       [72, 73, 74, 75, 76, 77, 78, 79, 80]])

np.array(np.arange(81)).reshape(9,9)[::3,::3]

array([[ 0,  3,  6],
       [27, 30, 33],
       [54, 57, 60]])

np.expand_dims(imread(all_images_list[0])[::4,::4],0).shape

(1, 128, 128)

jimread = lambda x: np.expand_dims(imread(x)[::2,::2],0)

test_image = jimread(all_images_list[0])
plt.imshow(test_image[0])

<matplotlib.image.AxesImage at 0x17aacdb8978>

png

check_contrast = re.compile(r'data\\tiff_images\\ID_([\d]+)_AGE_[\d]+_CONTRAST_([\d]+)_CT.tif')
label = []
id_list = []
for image in all_images_list:
    id_list.append(check_contrast.findall(image)[0][0])
    label.append(check_contrast.findall(image)[0][1])

label_list = pd.DataFrame(label,id_list)

label_list.head()
0
0000 1
0001 1
0002 1
0003 1
0004 1 
images = np.stack([jimread(i) for i in all_images_list],0)

len(images)

100

Split Data to Train, Test

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(images, label_list, test_size=0.1, random_state=0)

Reshape image Data

n_train, depth, width, height = X_train.shape
n_test,_,_,_ = X_test.shape

n_train,depth, width, height

(90, 1, 256, 256)

input_shape = (width,height,depth)

input_shape

(256, 256, 1)

input_train = X_train.reshape((n_train, width,height,depth))
input_train.shape
input_train.astype('float32')
input_train = input_train / np.max(input_train)
input_train.max()

1.0

input_test = X_test.reshape(n_test, *input_shape)
input_test.astype('float32')
input_test = input_test / np.max(input_test)

output_train = keras.utils.to_categorical(y_train, 2)
output_test = keras.utils.to_categorical(y_test, 2)
output_train[5]

array([ 0.,  1.])

input_train.shape

(90, 256, 256, 1)

Network

from keras.models import Sequential
from keras.layers import Dense, Flatten
from keras.optimizers import Adam
from keras.layers import Conv2D, MaxPooling2D

model = Sequential()
model.add(Conv2D(32, (4, 4), activation='relu', input_shape=input_shape))
 # 32개의 4x4 Filter 를 이용하여 Convolutional Network생성
model.add(MaxPooling2D(pool_size=(2, 2))) # 2x2 Maxpooling
model.add(Flatten()) # 쭉풀어서 Fully Connected Neural Network를 만든다.
model.add(Dense(2, activation='softmax'))

model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_5 (Conv2D)            (None, 253, 253, 32)      544
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 126, 126, 32)      0
_________________________________________________________________
flatten_4 (Flatten)          (None, 508032)            0
_________________________________________________________________
dense_4 (Dense)              (None, 2)                 1016066
=================================================================
Total params: 1,016,610
Trainable params: 1,016,610
Non-trainable params: 0
_________________________________________________________________

model.compile(loss='categorical_crossentropy',
              optimizer=Adam(),
              metrics=['accuracy'])

batch_size = 20
epochs = 40

history = model.fit(input_train, output_train,
                    batch_size=batch_size,
                    epochs=epochs,
                    verbose=1,
                    validation_data=(input_test, output_test))

Train on 90 samples, validate on 10 samples
Epoch 1/40
90/90 [==============================] - 1s - loss: 3.5470 - acc: 0.4667 - val_loss: 0.7117 - val_acc: 0.5000
Epoch 2/40
90/90 [==============================] - 0s - loss: 0.8217 - acc: 0.7444 - val_loss: 2.1616 - val_acc: 0.6000
Epoch 3/40
90/90 [==============================] - 0s - loss: 0.5779 - acc: 0.7889 - val_loss: 0.4329 - val_acc: 0.7000
Epoch 4/40
90/90 [==============================] - 0s - loss: 0.4233 - acc: 0.8000 - val_loss: 0.7214 - val_acc: 0.7000
Epoch 5/40
90/90 [==============================] - 0s - loss: 0.2544 - acc: 0.9000 - val_loss: 0.7363 - val_acc: 0.7000
Epoch 6/40
90/90 [==============================] - 0s - loss: 0.0576 - acc: 1.0000 - val_loss: 0.2655 - val_acc: 0.9000
Epoch 7/40
90/90 [==============================] - 0s - loss: 0.0904 - acc: 0.9667 - val_loss: 0.2782 - val_acc: 0.9000
Epoch 8/40
90/90 [==============================] - 0s - loss: 0.0265 - acc: 1.0000 - val_loss: 0.5291 - val_acc: 0.8000
Epoch 9/40
90/90 [==============================] - 0s - loss: 0.0323 - acc: 1.0000 - val_loss: 0.6118 - val_acc: 0.8000
Epoch 10/40
90/90 [==============================] - 0s - loss: 0.0242 - acc: 1.0000 - val_loss: 0.4871 - val_acc: 0.8000
Epoch 11/40
90/90 [==============================] - 0s - loss: 0.0131 - acc: 1.0000 - val_loss: 0.3767 - val_acc: 0.8000
Epoch 12/40
90/90 [==============================] - 0s - loss: 0.0112 - acc: 1.0000 - val_loss: 0.3334 - val_acc: 0.9000
Epoch 13/40
90/90 [==============================] - 0s - loss: 0.0103 - acc: 1.0000 - val_loss: 0.3521 - val_acc: 0.9000
Epoch 14/40
90/90 [==============================] - 0s - loss: 0.0083 - acc: 1.0000 - val_loss: 0.3828 - val_acc: 0.8000
Epoch 15/40
90/90 [==============================] - 0s - loss: 0.0069 - acc: 1.0000 - val_loss: 0.4158 - val_acc: 0.8000
Epoch 16/40
90/90 [==============================] - 0s - loss: 0.0060 - acc: 1.0000 - val_loss: 0.4394 - val_acc: 0.8000
Epoch 17/40
90/90 [==============================] - 0s - loss: 0.0055 - acc: 1.0000 - val_loss: 0.4422 - val_acc: 0.8000
Epoch 18/40
90/90 [==============================] - 0s - loss: 0.0050 - acc: 1.0000 - val_loss: 0.4348 - val_acc: 0.8000
Epoch 19/40
90/90 [==============================] - 0s - loss: 0.0046 - acc: 1.0000 - val_loss: 0.4296 - val_acc: 0.8000
Epoch 20/40
90/90 [==============================] - 0s - loss: 0.0043 - acc: 1.0000 - val_loss: 0.4386 - val_acc: 0.8000
Epoch 21/40
90/90 [==============================] - 0s - loss: 0.0040 - acc: 1.0000 - val_loss: 0.4428 - val_acc: 0.8000
Epoch 22/40
90/90 [==============================] - 0s - loss: 0.0037 - acc: 1.0000 - val_loss: 0.4448 - val_acc: 0.8000
Epoch 23/40
90/90 [==============================] - 0s - loss: 0.0035 - acc: 1.0000 - val_loss: 0.4392 - val_acc: 0.8000
Epoch 24/40
90/90 [==============================] - 0s - loss: 0.0032 - acc: 1.0000 - val_loss: 0.4443 - val_acc: 0.8000
Epoch 25/40
90/90 [==============================] - 0s - loss: 0.0030 - acc: 1.0000 - val_loss: 0.4493 - val_acc: 0.8000
Epoch 26/40
90/90 [==============================] - 0s - loss: 0.0029 - acc: 1.0000 - val_loss: 0.4531 - val_acc: 0.8000
Epoch 27/40
90/90 [==============================] - 0s - loss: 0.0027 - acc: 1.0000 - val_loss: 0.4548 - val_acc: 0.8000
Epoch 28/40
90/90 [==============================] - 0s - loss: 0.0025 - acc: 1.0000 - val_loss: 0.4542 - val_acc: 0.8000
Epoch 29/40
90/90 [==============================] - 0s - loss: 0.0024 - acc: 1.0000 - val_loss: 0.4493 - val_acc: 0.8000
Epoch 30/40
90/90 [==============================] - 0s - loss: 0.0023 - acc: 1.0000 - val_loss: 0.4466 - val_acc: 0.8000
Epoch 31/40
90/90 [==============================] - 0s - loss: 0.0022 - acc: 1.0000 - val_loss: 0.4540 - val_acc: 0.8000
Epoch 32/40
90/90 [==============================] - 0s - loss: 0.0021 - acc: 1.0000 - val_loss: 0.4581 - val_acc: 0.8000
Epoch 33/40
90/90 [==============================] - 0s - loss: 0.0020 - acc: 1.0000 - val_loss: 0.4665 - val_acc: 0.8000
Epoch 34/40
90/90 [==============================] - 0s - loss: 0.0019 - acc: 1.0000 - val_loss: 0.4746 - val_acc: 0.8000
Epoch 35/40
90/90 [==============================] - 0s - loss: 0.0018 - acc: 1.0000 - val_loss: 0.4781 - val_acc: 0.8000
Epoch 36/40
90/90 [==============================] - 0s - loss: 0.0017 - acc: 1.0000 - val_loss: 0.4788 - val_acc: 0.8000
Epoch 37/40
90/90 [==============================] - 0s - loss: 0.0016 - acc: 1.0000 - val_loss: 0.4752 - val_acc: 0.8000
Epoch 38/40
90/90 [==============================] - 0s - loss: 0.0016 - acc: 1.0000 - val_loss: 0.4780 - val_acc: 0.8000
Epoch 39/40
90/90 [==============================] - 0s - loss: 0.0015 - acc: 1.0000 - val_loss: 0.4765 - val_acc: 0.8000
Epoch 40/40
90/90 [==============================] - 0s - loss: 0.0014 - acc: 1.0000 - val_loss: 0.4762 - val_acc: 0.8000

history.history

{'acc': [0.46666666865348816,
  0.74444445636537337,
  0.7888888981607225,
  0.8000000052981906,
  0.9000000026490953,
  1.0,
  0.96666665871938073,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0,
  1.0],
 'loss': [3.5469898117913141,
  0.82166123059060836,
  0.57785466478930581,
  0.42333625257015228,
  0.25443292326397365,
  0.057624416632784739,
  0.090415256718794509,
  0.026549512313471899,
  0.032302634583579168,
  0.024231223803427484,
  0.013135068325532807,
  0.011229113986094793,
  0.010349845927622583,
  0.008280523535278108,
  0.0068697543659557896,
  0.0060073141939938068,
  0.0054577676993277334,
  0.0050347024031604333,
  0.0046245976765122675,
  0.0042539972087575328,
  0.0039954457121590776,
  0.003706983601053556,
  0.0034763599849409526,
  0.0032460537428657212,
  0.0030377271792127025,
  0.0028651561846749652,
  0.0027027277586360774,
  0.0025377526568869748,
  0.0024104245369219119,
  0.0023004989036255414,
  0.0021699113098697532,
  0.0020625931210815907,
  0.0019607565417471859,
  0.0018652590612570445,
  0.0017944105202332139,
  0.0017089455440226528,
  0.0016356468743955095,
  0.0015679198097334141,
  0.0015029058429516023,
  0.0014405033240715663],
 'val_acc': [0.5,
  0.60000002384185791,
  0.70000004768371582,
  0.70000004768371582,
  0.70000004768371582,
  0.89999997615814209,
  0.89999997615814209,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.89999997615814209,
  0.89999997615814209,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896,
  0.80000001192092896],
 'val_loss': [0.71169030666351318,
  2.1615970134735107,
  0.4328601062297821,
  0.72143769264221191,
  0.736328125,
  0.26548981666564941,
  0.27816122770309448,
  0.529060959815979,
  0.61176717281341553,
  0.4870985746383667,
  0.37665331363677979,
  0.33338609337806702,
  0.35212856531143188,
  0.38279733061790466,
  0.41582024097442627,
  0.43944761157035828,
  0.44218528270721436,
  0.43482279777526855,
  0.42955672740936279,
  0.43860852718353271,
  0.44281277060508728,
  0.44484323263168335,
  0.43915966153144836,
  0.44429975748062134,
  0.44934400916099548,
  0.45312052965164185,
  0.45475301146507263,
  0.45418411493301392,
  0.4492676854133606,
  0.44664955139160156,
  0.4540124237537384,
  0.45808041095733643,
  0.46652591228485107,
  0.47463798522949219,
  0.47812485694885254,
  0.47876393795013428,
  0.47516763210296631,
  0.4779818058013916,
  0.47649335861206055,
  0.47622102499008179]}

score = model.evaluate(input_test, output_test, verbose=0)
score

[0.47622102499008179, 0.80000001192092896]

Model 2

model2 = Sequential()
model2.add(Conv2D(50, (5, 5), activation='relu', input_shape=input_shape))
 # 32개의 4x4 Filter 를 이용하여 Convolutional Network생성
model2.add(MaxPooling2D(pool_size=(3, 3))) # 3x3 Maxpooling
model2.add(Conv2D(30, (4, 4), activation='relu', input_shape=input_shape))
model2.add(MaxPooling2D(pool_size=(2, 2))) # 2x2 Maxpooling
model2.add(Flatten()) # 쭉풀어서 Fully Connected Neural Network를 만든다.
model2.add(Dense(2, activation='softmax'))

model2.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_8 (Conv2D)            (None, 252, 252, 50)      1300
_________________________________________________________________
max_pooling2d_8 (MaxPooling2 (None, 84, 84, 50)        0
_________________________________________________________________
conv2d_9 (Conv2D)            (None, 81, 81, 30)        24030
_________________________________________________________________
max_pooling2d_9 (MaxPooling2 (None, 40, 40, 30)        0
_________________________________________________________________
flatten_6 (Flatten)          (None, 48000)             0
_________________________________________________________________
dense_6 (Dense)              (None, 2)                 96002
=================================================================
Total params: 121,332
Trainable params: 121,332
Non-trainable params: 0
_________________________________________________________________

model2.compile(loss='categorical_crossentropy',
              optimizer=Adam(),
              metrics=['accuracy'])

history = model2.fit(input_train, output_train,
                    batch_size=batch_size,
                    epochs=epochs,
                    verbose=1,
                    validation_data=(input_test, output_test))

Train on 90 samples, validate on 10 samples
Epoch 1/40
90/90 [==============================] - 1s - loss: 0.7125 - acc: 0.4667 - val_loss: 0.6814 - val_acc: 0.7000
Epoch 2/40
90/90 [==============================] - 0s - loss: 0.6080 - acc: 0.8556 - val_loss: 0.6586 - val_acc: 0.7000
Epoch 3/40
90/90 [==============================] - 0s - loss: 0.4500 - acc: 0.8667 - val_loss: 0.5858 - val_acc: 0.7000
Epoch 4/40
90/90 [==============================] - 0s - loss: 0.3791 - acc: 0.8333 - val_loss: 0.2891 - val_acc: 0.9000
Epoch 5/40
90/90 [==============================] - 0s - loss: 0.2775 - acc: 0.9000 - val_loss: 0.3965 - val_acc: 0.8000
Epoch 6/40
90/90 [==============================] - 0s - loss: 0.1812 - acc: 0.9556 - val_loss: 0.2022 - val_acc: 0.9000
Epoch 7/40
90/90 [==============================] - 0s - loss: 0.1389 - acc: 0.9667 - val_loss: 0.2486 - val_acc: 0.8000
Epoch 8/40
90/90 [==============================] - 0s - loss: 0.0684 - acc: 0.9778 - val_loss: 0.1031 - val_acc: 1.0000
Epoch 9/40
90/90 [==============================] - 0s - loss: 0.0485 - acc: 0.9778 - val_loss: 0.1829 - val_acc: 0.8000
Epoch 10/40
90/90 [==============================] - 0s - loss: 0.0401 - acc: 1.0000 - val_loss: 0.1561 - val_acc: 0.9000
Epoch 11/40
90/90 [==============================] - 0s - loss: 0.0177 - acc: 1.0000 - val_loss: 0.0454 - val_acc: 1.0000
Epoch 12/40
90/90 [==============================] - 0s - loss: 0.0187 - acc: 1.0000 - val_loss: 0.1317 - val_acc: 0.9000
Epoch 13/40
90/90 [==============================] - 0s - loss: 0.0100 - acc: 1.0000 - val_loss: 0.1339 - val_acc: 0.9000
Epoch 14/40
90/90 [==============================] - 0s - loss: 0.0058 - acc: 1.0000 - val_loss: 0.0540 - val_acc: 1.0000
Epoch 15/40
90/90 [==============================] - 0s - loss: 0.0045 - acc: 1.0000 - val_loss: 0.0564 - val_acc: 1.0000
Epoch 16/40
90/90 [==============================] - 0s - loss: 0.0028 - acc: 1.0000 - val_loss: 0.0769 - val_acc: 1.0000
Epoch 17/40
90/90 [==============================] - 0s - loss: 0.0024 - acc: 1.0000 - val_loss: 0.0755 - val_acc: 1.0000
Epoch 18/40
90/90 [==============================] - 0s - loss: 0.0020 - acc: 1.0000 - val_loss: 0.0629 - val_acc: 1.0000
Epoch 19/40
90/90 [==============================] - 0s - loss: 0.0016 - acc: 1.0000 - val_loss: 0.0488 - val_acc: 1.0000
Epoch 20/40
90/90 [==============================] - 0s - loss: 0.0014 - acc: 1.0000 - val_loss: 0.0434 - val_acc: 1.0000
Epoch 21/40
90/90 [==============================] - 0s - loss: 0.0013 - acc: 1.0000 - val_loss: 0.0466 - val_acc: 1.0000
Epoch 22/40
90/90 [==============================] - 0s - loss: 0.0011 - acc: 1.0000 - val_loss: 0.0457 - val_acc: 1.0000
Epoch 23/40
90/90 [==============================] - 0s - loss: 0.0010 - acc: 1.0000 - val_loss: 0.0441 - val_acc: 1.0000
Epoch 24/40
90/90 [==============================] - 0s - loss: 9.2702e-04 - acc: 1.0000 - val_loss: 0.0426 - val_acc: 1.0000
Epoch 25/40
90/90 [==============================] - 0s - loss: 8.4643e-04 - acc: 1.0000 - val_loss: 0.0412 - val_acc: 1.0000
Epoch 26/40
90/90 [==============================] - 0s - loss: 7.8929e-04 - acc: 1.0000 - val_loss: 0.0378 - val_acc: 1.0000
Epoch 27/40
90/90 [==============================] - 0s - loss: 7.1640e-04 - acc: 1.0000 - val_loss: 0.0397 - val_acc: 1.0000
Epoch 28/40
90/90 [==============================] - 0s - loss: 6.5236e-04 - acc: 1.0000 - val_loss: 0.0443 - val_acc: 1.0000
Epoch 29/40
90/90 [==============================] - 0s - loss: 6.2590e-04 - acc: 1.0000 - val_loss: 0.0465 - val_acc: 1.0000
Epoch 30/40
90/90 [==============================] - 0s - loss: 5.6936e-04 - acc: 1.0000 - val_loss: 0.0383 - val_acc: 1.0000
Epoch 31/40
90/90 [==============================] - 0s - loss: 5.2576e-04 - acc: 1.0000 - val_loss: 0.0343 - val_acc: 1.0000
Epoch 32/40
90/90 [==============================] - 0s - loss: 4.9576e-04 - acc: 1.0000 - val_loss: 0.0327 - val_acc: 1.0000
Epoch 33/40
90/90 [==============================] - 0s - loss: 4.6248e-04 - acc: 1.0000 - val_loss: 0.0337 - val_acc: 1.0000
Epoch 34/40
90/90 [==============================] - 0s - loss: 4.3219e-04 - acc: 1.0000 - val_loss: 0.0352 - val_acc: 1.0000
Epoch 35/40
90/90 [==============================] - 0s - loss: 4.0702e-04 - acc: 1.0000 - val_loss: 0.0352 - val_acc: 1.0000
Epoch 36/40
90/90 [==============================] - 0s - loss: 3.8466e-04 - acc: 1.0000 - val_loss: 0.0332 - val_acc: 1.0000
Epoch 37/40
90/90 [==============================] - 0s - loss: 3.5988e-04 - acc: 1.0000 - val_loss: 0.0334 - val_acc: 1.0000
Epoch 38/40
90/90 [==============================] - 0s - loss: 3.4036e-04 - acc: 1.0000 - val_loss: 0.0332 - val_acc: 1.0000
Epoch 39/40
90/90 [==============================] - 0s - loss: 3.2073e-04 - acc: 1.0000 - val_loss: 0.0318 - val_acc: 1.0000
Epoch 40/40
90/90 [==============================] - 0s - loss: 3.0824e-04 - acc: 1.0000 - val_loss: 0.0332 - val_acc: 1.0000

score = model2.evaluate(input_test, output_test, verbose=0)
score

[0.03319043293595314, 1.0]