By Topic

Performance of neural network architectures: Cascaded MLP versus extreme learning machine on cervical cell image classification

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

5 Author(s)
Yusoff, I.A. ; Sch. of Electr. & Electron. Eng., Univ. Sains Malaysia, Nibong Tebal, Malaysia ; Isa, N.A.M. ; Othman, N.H. ; Sulaiman, S.N.
more authors

In Malaysia, the screening coverage for cervical cancer is poor, which was at 2% in 1992, 3.5% in 1995, and at 6.2% in 1996, due to the shortage in pathologist workforce being one of the major cause. Study has been done before to overcome this by developing a diagnosis system based on neural networks, so that diagnosis can be done by an automated system with pathologist-like knowledge. Cell's features were used as input to the neural network architecture, and cell's classification into NORMAL, Low-Squamous Intraepithelial Lession (LSIL), or High-Squamous Intraepithelial Lession (HSIL) were used as output target. This paper focused on finding the best neural network to be used as classifier tool for cervical cancer diagnostic system with cervical cells' features as input. Two architectures of neural network system were proposed; Cascaded Multilayered Perceptron (c-MLP) and Extreme Learning Machine (ELM). Result suggests that all the features selected which are area, grey level, perimeter, red, green, blue, intensity1, intensity2 and saturation are more suitable to be used with c-MLP neural network architecture compared to ELM, with the accuracy of 96.02%.

Published in:

Information Sciences Signal Processing and their Applications (ISSPA), 2010 10th International Conference on

Date of Conference:

10-13 May 2010