Professional Documents
Culture Documents
Itime 2012 6291334
Itime 2012 6291334
LIANG Rongl, WANG Zhao-ping2, YANG Xin-yul 2 Medical Examination Center of Beijing TongRen
REN Yu-jiel, ZHANG Yingl, YAO Xu-yingl Hospital, Beijing, China
1 Beijing University of Traditional Chinese Corresponding author: WANG Zhao-ping
426
Light red tongue data import Matlab software and k
means cluster command was call to carry out classifies
TABLE T. CHROMATlCTTY VALUES COMPARISON OF VARIOUS TYPES
computing. The results showed that when light red tongue
OF TONGUE COLOR (X ± S) was divided into four categories, data classification results
Tongue are stable and good reproducibility. Chromaticity data of
N L* a* h* C* H*
color the light red tongue re-sorting center are as follows:
24. 79 12. :19 27.74 26.71
Light white 68.95 TABLE Ill . L1GIlT RED TONGUE CATEGORY FEATURE VECTOR COLOR
9 ± ± ± ±
tongue ±2.70 VALUES
:1. 98 I. 68 4.1:1 2.5:1
27.62 II. 99 :10.17 2:1.54 Categories L* a* b*
light red 66.27
695 ± ± ± ± 1 68.5976 24.6116 11.7171
tongue ±2.82
2. 97 I. 87 2.99 :1.50
2 67.8594 28.5368 13.4531
29.57 II. 92 :11. 94 22.00
64.18 3 64.7396 27.1670 10.5818
red tongue 426 ± ± ± ±
±2.95
:1.10 I. 95 :1.14 :1. :19 4 62.7042 31.9034 12.6034
:12. :1:1 10.77 :14.II 18.4:1
ecchymosed 59.15
19 ± ± ± ±
tongue ±:1.61 In this study, the same classification were used to
:1. 71 I. 82 :1.81 2.67
classifY the red tongue, dark red tongue, light dark tongue.
28.02 10.54 29.99 20.66
dark red 64.46 Result shown that the red tongue is divided into seven
692 ± ± ± ±
tongue ±:1.14 categories, whether increase or decrease the quantities
:1.06 I. 96 :1.II :1.60
26.05 9.24 27.71 19.55 of data classification may lead the results shown
dark purple 6:1.18 unstable in the cluster analysis, and reproducibility relatively
79 ± ± ± ±
tongue ±:1.7:1
:1.14 2.10 :1. 2:1 4.07 low.
25. 79 II. 62 28. :15 24.:11
light dark 66.85 TABLElV. R E D TONGUE CATEGORY FEATURE VECTOR COLOR
19:1 ± ± ± ±
tongue ±:1.20 VALUES
2.81 I. 85 2.85 :1.67
28.2:1 II. 14 :10. :17 21.95 Categories L* a* b*
light 65.5:1
:1 ± ± ± ± 1 68.1682 26.0392 12.3216
purple tongue ±4.40
7. :17 I. :12 7.:14 2.67
2 65.3928 29.4507 11.5692
C. Tongue color clustering analysis 3 64.9054 32.3409 13.7782
Based on the above characteristics, this study chose the
4 64.0077 26.4450 9.9641
k-means clustering method to calculate the cluster center of
5 62.1437 30.4333 13.1999
each type of tongue color. (Table II )
6 60.1960 36.0884 13.0456
TABLEl! . TilE CENTER COLOR VALUES O F TONGUE COLOR
CLUSTERlNG
7 59.8010 31.6291 10.4610
Chromaticity values
No. Tongue color While the dark red tongue data is divided into four
L* a* b* c* h* categories, the data classification results stable and
Light white repeatable.
1 68.94 24.78 12.39 27.70 26.5T
tongue
2 Light red tongue 66.27 27.61 11.99 30.10 23.4T TABLE V. DARK RED TONGUE CATEGORY FEATURE VECTOR COLOR
VALUES
427
perceived by the human eye [4]. This study based on the
doctors tongue diagnosis; extract the color values of various
types of tongue color, put it into CIE Lab color space,
observe all kinds of tongue color spatial distribution
TABLE V I. L iGlIT DARK TONGUE CATEGORY FEATURE VECTOR
characteristics. Adapt K-means clustering analysis method;
COLOR VALUES
calculate the color values of the cluster center and
Categories L* a* b* eigenvectors of various types of tongue color. Further carry
out the cluster analysis to those tongue color showing a class
I 71.3931 22.3553 12.5840
reunion like light red tongue or red tongue, then obtained a
2 68.6349 22.7073 9.4680
series of tongue color feature vector color values.
3 68.4713 25.2512 12.5211 According to the chromic value, tongue diagnosis pictures
4 66.4625 27.7122 11.6456 could be arranged in accordance with the depth of the
color sequence [5]. In this way, this study not only offer the
5 63.0026 25.5586 10.3600
basis of quantization for the diagnosis of tongue color
6 62.5150 31.0904 12.7929 teaching but also found objective criteria for learning
outcomes assessment.
In this research, sample data collected is insufficient.
Some types of the tongue color sampled data are too little to REFERENCES
be further classified.
[I] Ji Shaoliang Cheng-z haozhi , editor. TCM diagnosis [M] Beijing:
People's Medical Publishing House, 2008.
TABLE VII. TONGUE CATEGORY FEATURE VECTOR COLOR VALUES O F [2] EtehadTavakol M, Sa dri S, Ng EY. Application of K- and fuzzy c
OTIlER TYPES means for color segment ation of t hermal infrared breast images. J
Med Syst. 2010 Feb;34(1):35-42.
Tongue color L* a* b* [3] Ver munt JK. K-means may perfor m as well as mixture model
Light white tongue 68.94 24.78 12.39 clustering but may also be much worse: comment on Steinley and
Br usco (2011). Psychol Methods. 2011 Mar;16(1):82-8; discu ssion
Ecchymosed tongue 59.14 32.33 10.76 89-92.
[4] Tang S hu nqing. Colorimetry [M]. Beijing: Beijing Instit u te of
Dark purple tongue 63.18 26.04 9.24
Technology Press, 1990.
Light purple tongue 65.53 28.23 11.13 [5] Su W, Xu ZY, Wang ZQ, Xu JT. Objectified st u dy on tongue images
of patients wit h lung cancer of different syndromes. Chin J Integr
IV. DISCUSS Med. 2011 Apr ;17(4):272-6. Epub 2011 Apr 21.
250
200
150
100
50
0.00
j 2 0.00 40.00
years old
��
6 0.00 80.00 1 00.00
428
80
75
70
� 65
60
55
.. ,+ 50
50
20 18 16 14 12 10
a'
b'
429