Breast Cancer Classification from Histopathological Images Based on Improved Inception Model

Zhaoxu Li; Tao Song; Mengfei Ge; Jiaxin Liu; Hongwei Wang; Jia Wang

doi:10.3788/LOP202158.0817001

Laser & Optoelectronics Progress, Volume. 58, Issue 8, 0817001(2021)

Breast Cancer Classification from Histopathological Images Based on Improved Inception Model

Zhaoxu Li¹, Tao Song², Mengfei Ge¹, Jiaxin Liu¹, Hongwei Wang^1,3, and Jia Wang^2、*

Author Affiliations

¹School of Electrical Engineering, Xinjiang University, Urumqi, Xinjiang 830000, China

²School of Basic Medicine Science, Dalian Medical University, Dalian, Liaoning 110041, China

³School of Control Science and Engineering, Dalian University of Technology, Dalian, Liaoning 116023, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Histopathological patches after image preprocessing

Download full size

Fig. 2. Architecture of convolutional neural network

Download full size

Fig. 3. Architecture of model and internal structure. (a) Architecture of improved model; (b) architecture of Pre-treatment module; (c) architecture of Inception module

Download full size

Fig. 4. Overall strategy of experiment. (a) Process of training; (b) process of testing

Download full size

Fig. 5. Loss change during training iterative process

Download full size

Table 1. Convergence time of model under different the image sizes
View table
Table 1. Convergence time of model under different the image sizes
Picture size Convergence times /min
299 pixel×299 pixel 967
512 pixel×512 pixel 713
700 pixel×700 pixel 1219

Table 2. Test results corresponding different learning_rate
View table
Table 2. Test results corresponding different learning_rate
learning_rate Precision /% Recall /%
0.1 86.3 75.1
0.01 91.7 89.4
0.001 96.1 93.1

Table 3. Convergence time of different models
View table
Table 3. Convergence time of different models
Model Convergence times/min
LeNet 1346
AlexNet 1120
VGG-16 1219
GoogleNet 927
Proposed model 713

Table 4. Performance of the different models
View table
Table 4. Performance of the different models
Model Sensitivity /% Specificity /% Accuracy /%
LeNet 76.32 77.83 74.57
AlexNet 89.23 91.22 90.10
VGG-16 93.15 96.17 93.21
GoogleNet 96.1 93.1 95.19
Proposed model 96.13 98.47 96.12

Tools

Get Citation

Copy Citation Text

Zhaoxu Li, Tao Song, Mengfei Ge, Jiaxin Liu, Hongwei Wang, Jia Wang. Breast Cancer Classification from Histopathological Images Based on Improved Inception Model[J]. Laser & Optoelectronics Progress, 2021, 58(8): 0817001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites