Research Progress on Content-Based Medical Image Retrieval

Feng Yang; Guohui Wei; Hui Cao; Mengmeng Xing; Jing Liu; Junzhong Zhang

doi:10.3788/LOP57.060003

Journals >Laser & Optoelectronics Progress >Volume 57 >Issue 6 >Page 060003 > Article

Laser & Optoelectronics Progress
Vol. 57, Issue 6, 060003 (2020)

Research Progress on Content-Based Medical Image Retrieval

Feng Yang, Guohui Wei, Hui Cao^*, Mengmeng Xing, Jing Liu, and Junzhong Zhang

Author Affiliations

School of Science and Technology, Shandong University of Traditional Chinese Medicine, Jinan, Shandong 250355, China

show less

DOI: 10.3788/LOP57.060003 Cite this Article Set citation alerts

Feng Yang, Guohui Wei, Hui Cao, Mengmeng Xing, Jing Liu, Junzhong Zhang. Research Progress on Content-Based Medical Image Retrieval[J]. Laser & Optoelectronics Progress, 2020, 57(6): 060003 Copy Citation Text

show less

Fig. 1. Flow chart of image retrieval using deep global features

Download full size

Fig. 2. Flow chart of image retrieval based on deep local feature aggregation

Download full size

Fig. 3. Diagram of IRMA system structure

Download full size

Time	Representative method	Category
1996	Methods featured by color^[4],edge^[5], and texture^[6]	Hand-craft global features
2001	Method featured by GIST (generalized search trees)^[7]
2003	BoW (bag of word)^[8]
2004	Method featured by SIFT (scale-invariant feature transform)^[9]
2005	Method featured by HOG (histogram of oriented gradients)^[10]
2006	SURF (speeded up robust features)^[11],LBP (local binary pattern)^[12]	Hand-craft local features
2007	FV (fisher vector)^[13]
2012	VLAD (vector of locally aggregated descriptor)^[14]
2014	Triangulation embedding^[15]
2014	Neural code^[16]	Deep global features
2014	MOP-CNN (multiscale orderlesspooling-convolutional neural network)^[17]
2015	SPoC (sum-pooled convolutional features)^[18]
2016	R-MAC (regional maximum activation of convolutions)^[19],CroW (cross weight aggregation code)^[20]	Deep local features
2017	Class weighted^[21]
2018	PWA (progressive web app)^[22]

Table 1. Representative methods of image feature extraction and their development stages

Type	Common model
	Convolutionalneural network,
Superviseddeep network	deep stackingnetwork,
	deep-structuredconditional random fields
Unsuperviseddeep network	Auto encoders, restrictedBoltzmann machines,sparse coding, K-means
Semi-superviseddeep network	Pre-trained deepneural networks

Table 2. Classification of depth feature extraction methods

System name	Application object	Feature extraction	Similarity measure	Related feedback
ASSERT^[75]	CT image of lung	Labeled area feature	Based on classification	×
NHANES III^[76]	Spinal X-ray image	Contour shape	Contour matching	×
MRIAGE^[77]	MRI imageof brain	3D Texture	Histogram	×
FICBDS^[78]	FunctionalPET	Physiologicalinformation	Vectordistance	×
IRMA^[79]	Integrated medicalimaging	Multiple featuredescription	Multipleranging metric	√
VisMed ^[80]	Integratedmedical imaging	Visual word	Wordmatching	×
medGIFT^[81]	Integratedmedical imaging	Text andvisual features	Multimodal informationsorting fusion	×
NovaMedSearech^[82]	Integrated medicalimaging	Text andvisual features	Multimodal informationsorting fusion	√
iMedline^[83]	Clinical case	Text and visualfeatures	Linear weighting ofmultiple features	√

Table 3. Common CBMIR systems

Feng Yang, Guohui Wei, Hui Cao, Mengmeng Xing, Jing Liu, Junzhong Zhang. Research Progress on Content-Based Medical Image Retrieval[J]. Laser & Optoelectronics Progress, 2020, 57(6): 060003

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information