Spectral Discrimination of Rabbit Liver VX2 Tumor and Normal Tissue Based on Genetic Algorithm-Support Vector Machine

Chen-yang LIU; Huang-rong XU; Feng DUAN; Tai-sheng WANG; Zhen-wu LU; Wei-xing YU

doi:10.3964/j.issn.1000-0593(2021)10-3123-06

Journals >Spectroscopy and Spectral Analysis >Volume 41 >Issue 10 >Page 3123 > Article

Spectroscopy and Spectral Analysis
Vol. 41, Issue 10, 3123 (2021)

Spectral Discrimination of Rabbit Liver VX2 Tumor and Normal Tissue Based on Genetic Algorithm-Support Vector Machine

Chen-yang LIU^1、*, Huang-rong XU^{2、2; 3;}, Feng DUAN^4、4;, Tai-sheng WANG^1、1;, Zhen-wu LU^1、1;, and Wei-xing YU^{3、3; *;}

Author Affiliations

¹1. State Key Laboratory of Applied Optics, Changchun Institute of Optics, Fine Mechanics & Physics, Chinese Academy of Sciences, Changchun 130033, China

²2. University of Chinese Academy of Sciences, Beijing 100049, China

³3. Key Laboratory of Spectral Imaging Technology, Xi’an Institute of Optics and Precision Mechanics, Chinese Academy of Science, Xi’an 710119, China

⁴4. Department of Interventional Radiology, the General Hospital of Chinese People’s Liberation Army, Beijing 100853, China

show less

DOI: 10.3964/j.issn.1000-0593(2021)10-3123-06 Cite this Article

Chen-yang LIU, Huang-rong XU, Feng DUAN, Tai-sheng WANG, Zhen-wu LU, Wei-xing YU. Spectral Discrimination of Rabbit Liver VX2 Tumor and Normal Tissue Based on Genetic Algorithm-Support Vector Machine[J]. Spectroscopy and Spectral Analysis, 2021, 41(10): 3123 Copy Citation Text

show less

Fig. 1. Experimental apparatus for measuring VX2 tumor tissue and normal tissue in rabbit liver

Download full size | View in the Article

Reflection of VX2 tumor tissue and normal tissue in rabbit liver (a) and spectral reflection of normal liver tissue in non-bleeding living body, VX2 tumor tissue in non-bleeding living body, normal liver tissue in bleeding isolated and VX2 tumor tissue in bleeding isolated (b)

Fig. 2. Reflection of VX2 tumor tissue and normal tissue in rabbit liver (a) and spectral reflection of normal liver tissue in non-bleeding living body, VX2 tumor tissue in non-bleeding living body, normal liver tissue in bleeding isolated and VX2 tumor tissue in bleeding isolated (b)

Download full size | View in the Article

Fig. 3. The predicted and true values of two categories (a) and four categories (b) of SVM parameters are optimized by using 5-K cross validation

Download full size | View in the Article

Fig. 4. Fitness curves (a) and (b), classification results (c) and (d) of Two categories and Four categories optimized by genetic algorithm

Download full size | View in the Article

Classification method	Parameter optimization method	Optimal values of c parameters	Optimal values of kernel function parameter g	The accuracy of the calibration set	The accuracy of the prediction set
Two categories	5-fold cross validation	4	0.125 0	100%(130/130)	100%(30/30)
Two categories	Genetic algorithm	0.845 6	0.121 1	100%(130/130)	100%(30/30)
Four categories	5-fold cross validation	8	0.062 5	99.242 4%(130/129)	93.333%(30/27)
Four categories	Genetic algorithm	5.530 7	0.068 5	99.242 4%(130/129)	100%(30/30)

Table 1. Compares the results of Two categories and Four categories of SVM parameters optimized by two methods

View in the Article

Classification method	Optimal values of c parameters	Optimal values of kernel function parameter g	The number of variables	Running time of algorithm /s	The accuracy of the calibration set/%	The accuracy of the prediction set/%
Genetic algorithm- Two categories	0.845 6	0.121 1	1 401	340.26	100	100
	5.220 6	0.294 4	902	221.026	100	100
	1.213 2	1.463 4	280	83.74	100	96.67
	2.604 1	1.539 8	140	44.28	100	96.67
	5.600 7	1.857 5	94	32.90	100	96.67
	5.532 1	2.919 4	70	27.36	100	96.67
	3.341 6	1.686 3	56	19.51	100	96.67
	7.952 8	2.465 0	47	20.22	100	100
	8.252 6	1.590 5	35	17.54	99.24	96.67
	9.911 7	1.363 3	14	11.41	99.24	93.33
Genetic algorithm- Four categories	5.530 7	0.068 5	1 401	490.99	99.24	100
	7.670 8	0.062 3	902	298.97	99.24	93.33
	7.177 7	0.292 6	280	83.64	100	93.33
	3.604 4	0.957 1	140	72.05	99.24	93.33
	5.538 0	0.813 5	94	48.59	99.24	93.33
	6.360 3	2.011 9	70	45.17	100	96.67
	9.107 5	1.019 6	56	35.37	99.24	96.67
	3.909 3	2.436 1	47	31.84	98.48	96.67
	7.675 2	1.633 2	35	27.87	98.48	93.33
	5.741 3	4.335 5	14	21.02	99.24	93.33

Table 2. The results of Two categories and Four categories under different number variables of SVM parameter optimized by Genetic algorithm

Download Citation

Save the article for my favorites

Paper Information