TY - GEN
T1 - Membranous nephropathy identification using hyperspectral microscopic images
AU - Wei, Xueling
AU - Tu, Tianqi
AU - Zhang, Nianrong
AU - Yang, Yue
AU - Li, Wenge
AU - Li, Wei
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2019.
PY - 2019
Y1 - 2019
N2 - In clinical diagnosis of membranous nephropathy (MN), separating hepatitis B virus-associated membranous nephropathy (HBV-MN) and primary membranous nephropathy (PMN) is an important step. Currently, most diagnostic technique is to conduct immunofluo-rescence on kidney biopsy samples with high false positive probability. In this paper, an automatic MN identification approach using medical hyperspectral microscopic images is developed. The proposed framework, denoted as local fisher discriminant analysis-deep neural network (LFDA-DNN), firstly constructs a subspace with well separability for HBV-MN and PMN through projection, and then obtains high-level features that are beneficial for final classification via a DNN-based network. To evaluate the effectiveness of LFDA-DNN, experiments are implemented on a real MN dataset, and the results confirm the superiority of LFDA-DNN for recognising HBV-MN and PMN precisely.
AB - In clinical diagnosis of membranous nephropathy (MN), separating hepatitis B virus-associated membranous nephropathy (HBV-MN) and primary membranous nephropathy (PMN) is an important step. Currently, most diagnostic technique is to conduct immunofluo-rescence on kidney biopsy samples with high false positive probability. In this paper, an automatic MN identification approach using medical hyperspectral microscopic images is developed. The proposed framework, denoted as local fisher discriminant analysis-deep neural network (LFDA-DNN), firstly constructs a subspace with well separability for HBV-MN and PMN through projection, and then obtains high-level features that are beneficial for final classification via a DNN-based network. To evaluate the effectiveness of LFDA-DNN, experiments are implemented on a real MN dataset, and the results confirm the superiority of LFDA-DNN for recognising HBV-MN and PMN precisely.
KW - Deep neural network
KW - Hyperspectral microscopic images
KW - MN Identification
UR - http://www.scopus.com/inward/record.url?scp=85076977476&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-31723-2_15
DO - 10.1007/978-3-030-31723-2_15
M3 - Conference contribution
AN - SCOPUS:85076977476
SN - 9783030317225
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 173
EP - 184
BT - Pattern Recognition and Computer Vision 2nd Chinese Conference, PRCV 2019, Proceedings, Part II
A2 - Lin, Zhouchen
A2 - Wang, Liang
A2 - Tan, Tieniu
A2 - Yang, Jian
A2 - Shi, Guangming
A2 - Zheng, Nanning
A2 - Chen, Xilin
A2 - Zhang, Yanning
PB - Springer
T2 - 2nd Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2019
Y2 - 8 November 2019 through 11 November 2019
ER -
