A computer-aided diagnostic system for liver tumor detection using modified U-Net architecture
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Kalsoom, A. | - |
dc.contributor.author | Maqsood, M. | - |
dc.contributor.author | Yasmin, S. | - |
dc.contributor.author | Bukhari, M. | - |
dc.contributor.author | Shin, Z. | - |
dc.contributor.author | Rho, Seungmin | - |
dc.date.accessioned | 2023-03-08T07:23:57Z | - |
dc.date.available | 2023-03-08T07:23:57Z | - |
dc.date.issued | 2022-05 | - |
dc.identifier.issn | 0920-8542 | - |
dc.identifier.issn | 1573-0484 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/61407 | - |
dc.description.abstract | A multimedia-based medical decision-making system is an ultimate requirement in the medical imaging domain. In the healthcare sector, achieving quick and efficient results is one of the primary needs as well as a necessity for both doctors and patients. Liver cancer is increasingly spreading disease across the globe, and its timely diagnosis requires an accurate and precise tumor segmentation. However, it is very challenging to segment the tumors because of the variability in appearance, hazy borders, diverse densities, and sizes of tumors. Currently, deep learning-based approaches are being applied in a variety of domains which results in better performances. In this paper, a deep learning method based on modified U-Net referred to as Residual-Atrous U-Net (RA-Net) is proposed to segment the liver tumors. The suggested model is implemented by employing U-Net as a base model and extracts the tumor’s features utilizing a parallel structure-based Atrous convolution block embedded in the original U-Net. This addresses the heterogeneity of tumors in terms of sizes as well as retaining the wider context without introducing any parameters to the model due to the various scales dilated kernels. Moreover, a strong, smooth, and non-monoatomic Mish activation function is deployed in this parallel Atrous block to bring nonlinearity. Besides this, the features of tumors are also extracted from Res Block simultaneously which learns the residual between the input and output feature maps via a skip link that jumps directly from input to output. This identity mapping in this block overcomes network degradation and increases performance. Later on, all these extracted features are fused to be sent to the network and hence improve the feature learning of the original U-Net. Furthermore, we have used the strategy of segmenting the tumors directly from the CT-scan instead of the two-stage process used by the majority of existing methods. We have performed experimental analysis over the 3DIRCADb dataset and according to the results shown by the proposed RA-Net, the Jaccard score which is a performance indicator stands out at 72%. © 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature. | - |
dc.format.extent | 23 | - |
dc.language | 영어 | - |
dc.language.iso | ENG | - |
dc.publisher | Springer | - |
dc.title | A computer-aided diagnostic system for liver tumor detection using modified U-Net architecture | - |
dc.type | Article | - |
dc.identifier.doi | 10.1007/s11227-021-04266-6 | - |
dc.identifier.bibliographicCitation | Journal of Supercomputing, v.78, no.7, pp 9668 - 9690 | - |
dc.description.isOpenAccess | N | - |
dc.identifier.wosid | 000744394700002 | - |
dc.identifier.scopusid | 2-s2.0-85123199865 | - |
dc.citation.endPage | 9690 | - |
dc.citation.number | 7 | - |
dc.citation.startPage | 9668 | - |
dc.citation.title | Journal of Supercomputing | - |
dc.citation.volume | 78 | - |
dc.type.docType | Article | - |
dc.publisher.location | 네델란드 | - |
dc.subject.keywordAuthor | Feature fusion | - |
dc.subject.keywordAuthor | Liver tumor detection | - |
dc.subject.keywordAuthor | Mish activation | - |
dc.subject.keywordAuthor | Parallel structure Atrous convolutions | - |
dc.subject.keywordAuthor | RA-Net | - |
dc.subject.keywordAuthor | Tumor segmentation | - |
dc.subject.keywordAuthor | U-Net | - |
dc.subject.keywordPlus | SEGMENTATION | - |
dc.subject.keywordPlus | CT | - |
dc.subject.keywordPlus | NETWORKS | - |
dc.relation.journalResearchArea | Computer Science | - |
dc.relation.journalResearchArea | Engineering | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Hardware & Architecture | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Theory & Methods | - |
dc.relation.journalWebOfScienceCategory | Engineering, Electrical & Electronic | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
84, Heukseok-ro, Dongjak-gu, Seoul, Republic of Korea (06974)02-820-6194
COPYRIGHT 2019 Chung-Ang University All Rights Reserved.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.