kerasによるUNetの実装ができない

現在、kerasによるUNetの実装に取り組んでいます。そこで下記のプログラムを使用し、モデルの作成を試みたのですが、mergeの部分でエラーが発生して困っています。keras2ではmergeLayerは使用できないとのことですが、どなたか代替方法を教えて頂きたいです。よろしくお願いします。

Python
1def create_fcn(input_size):
2    inputs = Input((input_size[1], input_size[0], 3))
3
4    conv1 = Convolution2D(32, 3, 3, activation='relu', border_mode='same')(inputs)
5    conv1 = Convolution2D(32, 3, 3, activation='relu', border_mode='same')(conv1)
6    pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)
7
8    conv2 = Convolution2D(64, 3, 3, activation='relu', border_mode='same')(pool1)
9    conv2 = Convolution2D(64, 3, 3, activation='relu', border_mode='same')(conv2)
10    pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)
11
12    conv3 = Convolution2D(128, 3, 3, activation='relu', border_mode='same')(pool2)
13    conv3 = Convolution2D(128, 3, 3, activation='relu', border_mode='same')(conv3)
14    pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)
15
16    conv4 = Convolution2D(256, 3, 3, activation='relu', border_mode='same')(pool3)
17    conv4 = Convolution2D(256, 3, 3, activation='relu', border_mode='same')(conv4)
18    pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)
19
20    conv5 = Convolution2D(512, 3, 3, activation='relu', border_mode='same')(pool4)
21    conv5 = Convolution2D(512, 3, 3, activation='relu', border_mode='same')(conv5)
22    pool5 = MaxPooling2D(pool_size=(2, 2))(conv5)
23
24    conv6 = Convolution2D(1024, 3, 3, activation='relu', border_mode='same')(pool5)
25    conv6 = Convolution2D(1024, 3, 3, activation='relu', border_mode='same')(conv6)
26
27    up7 = merge([UpSampling2D(size=(2, 2))(conv6), conv5], mode='concat', concat_axis=3)
28    conv7 = Convolution2D(512, 3, 3, activation='relu', border_mode='same')(up7)
29    conv7 = Convolution2D(512, 3, 3, activation='relu', border_mode='same')(conv7)
30
31    up8 = merge([UpSampling2D(size=(2, 2))(conv7), conv4], mode='concat', concat_axis=3)
32    conv8 = Convolution2D(256, 3, 3, activation='relu', border_mode='same')(up8)
33    conv8 = Convolution2D(256, 3, 3, activation='relu', border_mode='same')(conv8)
34
35    up9 = merge([UpSampling2D(size=(2, 2))(conv8), conv3], mode='concat', concat_axis=3)
36    conv9 = Convolution2D(128, 3, 3, activation='relu', border_mode='same')(up9)
37    conv9 = Convolution2D(128, 3, 3, activation='relu', border_mode='same')(conv9)
38
39    up10 = merge([UpSampling2D(size=(2, 2))(conv9), conv2], mode='concat', concat_axis=3)
40    conv10 = Convolution2D(64, 3, 3, activation='relu', border_mode='same')(up10)
41    conv10 = Convolution2D(64, 3, 3, activation='relu', border_mode='same')(conv10)
42
43    up11 = merge([UpSampling2D(size=(2, 2))(conv10), conv1], mode='concat', concat_axis=3)
44    conv11 = Convolution2D(32, 3, 3, activation='relu', border_mode='same')(up11)
45    conv11 = Convolution2D(32, 3, 3, activation='relu', border_mode='same')(conv11)
46
47    conv12 = Convolution2D(1, 1, 1, activation='sigmoid')(conv11)
48
49    fcn = Model(input=inputs, output=conv12)
50
51    return fcn