Texture Analysis of Tongue Coating in Traditional Chinese Medicine Based on Transfer Learning and Multi-Model Decision - Details

Author：

Xiao, Qingxin (Xiao, Qingxin.) | Zhang, Hui (Zhang, Hui.) | Zhang, Jing (Zhang, Jing.) | Zhuo, Li (Zhuo, Li.)

Indexed by：

EI Scopus

Abstract：

The　tongue　coating　texture　is　one　of　the　basic　characteristics　of　tongue　manifestation　in　traditional　Chinese　medicine　(TCM),　which　is　generally　divided　into　three　types:　curdy　coating,　greasy　coating,　and　non-corrosive　coating.　In　this　paper,　a　texture　analysis　method　of　tongue　coating　in　TCM　is　proposed　by　using　transfer　learning　and　multi-model　decision.　Firstly,　tongue　coating　texture　is　pre-classified　by　using　the　transfer　learning　strategy　to　pre-train　and　fine-tune　initial　network　model　trained　on　large-scale　dataset　with　the　tongue　coating　dataset.　Then,　the　multi-model　decision　is　made　by　comparing　the　classification　accuracy　of　different　deep　network　models　including　InceptionNet　V3,　ResNet50,　and　MobileNet　V1　to　further　optimize　the　texture　analysis　results　of　tongue　coating.　The　experimental　results　show　that　the　proposed　texture　analysis　method　of　tongue　coating　can　achieve　better　classification　accuracy,　which　has　practical　meanings　for　assisting　the　clinical　diagnosis　and　research　for　TCM.　©　2021,　The　Author(s),　under　exclusive　licence　to　Springer　Science+Business　Media,　LLC　part　of　Springer　Nature.

Keyword：

Textures Coatings Clinical research Computer aided diagnosis Transfer learning Learning systems Large dataset

Author Community：

[ 1 ] [Xiao, Qingxin]Signal and Information Processing Laboratory, Beijing University of Technology, Beijing; 100124, China
[ 2 ] [Zhang, Hui]Signal and Information Processing Laboratory, Beijing University of Technology, Beijing; 100124, China
[ 3 ] [Zhang, Jing]Signal and Information Processing Laboratory, Beijing University of Technology, Beijing; 100124, China
[ 4 ] [Zhuo, Li]Signal and Information Processing Laboratory, Beijing University of Technology, Beijing; 100124, China