TY - JOUR
T1 - Parallel deep solutions for image retrieval from imbalanced medical imaging archives
AU - Khatami, Amin
AU - Babaie, Morteza
AU - Khosravi, Abbas
AU - Tizhoosh, H. R.
AU - Nahavandi, Saeid
N1 - Publisher Copyright:
© 2017 Elsevier B.V.
PY - 2018/2
Y1 - 2018/2
N2 - Learning and extracting representative features along with similarity measurements in high dimensional feature spaces is a critical task. Moreover, the problem of how to bridge the semantic gap, between the low-level information captured by a machine learning model and the high-level one interpreted by a human operator, is still a practical challenge, especially in medicine. In medical applications, retrieving similar images from archives of past cases can be immensely beneficial in diagnostic imaging. However, large and balanced datasets may not be available for many reasons. Exploring the ways of using deep networks, for classification to retrieval, to fill this semantic gap was a key question for this research. In this work, we propose a parallel deep solution approach based on convolutional neural networks followed by a local search using LBP, HOG and Radon features. The IRMA dataset, from ImageCLEF initiative, containing 14,400 X-ray images, is employed to validate the proposed scheme. With a total IRMA error of 165.55, the performance of our scheme surpasses the dictionary approach and many other learning methods applied on the same dataset.
AB - Learning and extracting representative features along with similarity measurements in high dimensional feature spaces is a critical task. Moreover, the problem of how to bridge the semantic gap, between the low-level information captured by a machine learning model and the high-level one interpreted by a human operator, is still a practical challenge, especially in medicine. In medical applications, retrieving similar images from archives of past cases can be immensely beneficial in diagnostic imaging. However, large and balanced datasets may not be available for many reasons. Exploring the ways of using deep networks, for classification to retrieval, to fill this semantic gap was a key question for this research. In this work, we propose a parallel deep solution approach based on convolutional neural networks followed by a local search using LBP, HOG and Radon features. The IRMA dataset, from ImageCLEF initiative, containing 14,400 X-ray images, is employed to validate the proposed scheme. With a total IRMA error of 165.55, the performance of our scheme surpasses the dictionary approach and many other learning methods applied on the same dataset.
KW - CBIR
KW - Content-based image retrieval
KW - Deep learning
KW - HOG
KW - LBP
KW - Medical imaging
KW - Radon
UR - http://www.scopus.com/inward/record.url?scp=85037532632&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85037532632&partnerID=8YFLogxK
U2 - 10.1016/j.asoc.2017.11.024
DO - 10.1016/j.asoc.2017.11.024
M3 - Article
AN - SCOPUS:85037532632
SN - 1568-4946
VL - 63
SP - 197
EP - 205
JO - Applied Soft Computing Journal
JF - Applied Soft Computing Journal
ER -