Abstract
Machine learning has been used in many retinal image processing applications such as optic disc segmentation. It assumes that the training and testing data sets have the same feature distribution. However, retinal images are often collected under different conditions and may have different feature distributions. Therefore, the models trained from one data set may not work well for another data set. However, it is often too expensive and time consuming to label the needed training data and rebuild the models for all different data sets. In this paper, we propose a novel quadratic divergence regularized support vector machine (QDSVM) to transfer the knowledge from domains with sufficient training data to domains with limited or even no training data. The proposed method simultaneously minimizes the distribution difference between the source domain and target domain while training the classifier. Experimental results show that the proposed transfer learning based method reduces the classification error in superpixel level from 14.2% without transfer learning to 2.4% with transfer learning. The proposed method is effective to transfer the label knowledge from source to target domain, which enables it to be used for optic disc segmentation in data sets with different feature distributions.
Highlights
The optic disc (OD) is the location where ganglion cell axons exit the eye to form the optic nerve
We propose a new transfer learning method, referred to as quadratic divergence regularized SVM (QDSVM), for transfer learning in OD segmentation
We evaluate the performance of the proposed method when different numbers of labelled images are used
Summary
The optic disc (OD) is the location where ganglion cell axons exit the eye to form the optic nerve. Template based approaches often model the disc as a circle [3] or an ellipse [4] as the discs are approximately a circle or an ellipse In these approaches, the disc is often detected by Hough transform from the edges in the image, while different authors propose different methods to detect the edges from the discs. Deformable model based approaches usually start with an initial disc contour and deform toward the disc edges based on various energy terms such as boundary energy, shape energy, region energy [13]. These energy terms are often defined using image intensity, image gradient, and/or boundary smoothness, etc. The limitation of the pixel classification is that the high number of pixels makes the optimization difficult
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
