Shift Learning by Integrating Image and Subsystem Training for Edge Detection in Medical Images

Author:

Ramapraba P. S.,Kishore Somala Rama,Jayanthi S.,Kumar R.

Abstract

Abstract Several approaches for segmenting clinical data are focused on controlled vertex shade labelling. In particular, such strategies work well if the training set reflects the test pictures per chapter. However, issues can occur in the preparation and testing process, for instance, due to variations in scanners, procedures, otherwise patient classes, lead to distinct concentrations. In these situations, weighing images based on distribution similarities has shown a substantial improvement in inefficiency. To suggest that most of the training examples reflect the quiz information; it should not be similar to the deceiving information. Therefore, we examine the importance of kernel learning to weigh images to minimize discrepancies between training and test results. A local feature measurement scheme has been proposed to minimize the average distance between training and testing data that allows image weights and Kernel to be jointly optimized. Experiments on brain tissues, lesion of white material, and hippocampus division demonstrate because both kernel processing and image calculation boost the efficiency of heterogeneous data dramatically if used separately. MMD weighting here works similarly to the image weighting approaches previously proposed. The combination of image measurement and kernel processing, independently or jointly optimized, could result in a slight additional performance increase.

Publisher

IOP Publishing

Subject

General Physics and Astronomy

Reference17 articles.

1. Gaze-guided object classification using deep neural networks for attention-based computing;Barz,2016

2. Active appearance model segmentation in medical image analysis;Mitchell,2004

3. Imagenet classification with deep convolutional neural networks;Krizhevsky;Advances in neural information processing systems,2012

4. Accuracy and reproducibility study of automatic MRI brain tissue segmentation methods.;De Boer;Neu roimag e,2010

5. Unsupervised representation learning by sorting sequences;Lee,2017

Cited by 1 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Sub-pixel Edge Detection Based on Logistic Function Fitting;2023 3rd International Conference on Electronic Information Engineering and Computer Science (EIECS);2023-09-22

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3