MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation
Objective: For computer-aided Chinese medical diagnosis and aiming at the problem of insufficient segmentation, a novel multi-level method based on the multi-scale fusion residual neural network (MF2ResU-Net) model is proposed. Methods: To obtain refined features of retinal blood vessels, three casc...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
KeAi Communications Co., Ltd.
2022-12-01
|
Series: | Digital Chinese Medicine |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2589377722000787 |
_version_ | 1811160164024188928 |
---|---|
author | Zhenchao CUI Shujie SONG Jing QI |
author_facet | Zhenchao CUI Shujie SONG Jing QI |
author_sort | Zhenchao CUI |
collection | DOAJ |
description | Objective: For computer-aided Chinese medical diagnosis and aiming at the problem of insufficient segmentation, a novel multi-level method based on the multi-scale fusion residual neural network (MF2ResU-Net) model is proposed. Methods: To obtain refined features of retinal blood vessels, three cascade connected U-Net networks are employed. To deal with the problem of difference between the parts of encoder and decoder, in MF2ResU-Net, shortcut connections are used to combine the encoder and decoder layers in the blocks. To refine the feature of segmentation, atrous spatial pyramid pooling (ASPP) is embedded to achieve multi-scale features for the final segmentation networks. Results: The MF2ResU-Net was superior to the existing methods on the criteria of sensitivity (Sen), specificity (Spe), accuracy (ACC), and area under curve (AUC), the values of which are 0.8013 and 0.8102, 0.9842 and 0.9809, 0.9700 and 0.9776, and 0.9797 and 0.9837, respectively for DRIVE and CHASE DB1. The results of experiments demonstrated the effectiveness and robustness of the model in the segmentation of complex curvature and small blood vessels. Conclusion: Based on residual connections and multi-feature fusion, the proposed method can obtain accurate segmentation of retinal blood vessels by refining the segmentation features, which can provide another diagnosis method for computer-aided Chinese medical diagnosis. |
first_indexed | 2024-04-10T05:52:57Z |
format | Article |
id | doaj.art-789b6cfb077e4216af36739ea6162889 |
institution | Directory Open Access Journal |
issn | 2589-3777 |
language | English |
last_indexed | 2024-04-10T05:52:57Z |
publishDate | 2022-12-01 |
publisher | KeAi Communications Co., Ltd. |
record_format | Article |
series | Digital Chinese Medicine |
spelling | doaj.art-789b6cfb077e4216af36739ea61628892023-03-04T04:23:43ZengKeAi Communications Co., Ltd.Digital Chinese Medicine2589-37772022-12-0154406418MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentationZhenchao CUI0Shujie SONG1Jing QI2School of Cyber Security and Computer, Hebei University, Baoding, Hebei 071002, China; Machine Vision Engineering Research Center, Hebei University, Baoding, Hebei 071002, China; Corresponding author: .School of Cyber Security and Computer, Hebei University, Baoding, Hebei 071002, China; Machine Vision Engineering Research Center, Hebei University, Baoding, Hebei 071002, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding, Hebei 071002, China; Machine Vision Engineering Research Center, Hebei University, Baoding, Hebei 071002, ChinaObjective: For computer-aided Chinese medical diagnosis and aiming at the problem of insufficient segmentation, a novel multi-level method based on the multi-scale fusion residual neural network (MF2ResU-Net) model is proposed. Methods: To obtain refined features of retinal blood vessels, three cascade connected U-Net networks are employed. To deal with the problem of difference between the parts of encoder and decoder, in MF2ResU-Net, shortcut connections are used to combine the encoder and decoder layers in the blocks. To refine the feature of segmentation, atrous spatial pyramid pooling (ASPP) is embedded to achieve multi-scale features for the final segmentation networks. Results: The MF2ResU-Net was superior to the existing methods on the criteria of sensitivity (Sen), specificity (Spe), accuracy (ACC), and area under curve (AUC), the values of which are 0.8013 and 0.8102, 0.9842 and 0.9809, 0.9700 and 0.9776, and 0.9797 and 0.9837, respectively for DRIVE and CHASE DB1. The results of experiments demonstrated the effectiveness and robustness of the model in the segmentation of complex curvature and small blood vessels. Conclusion: Based on residual connections and multi-feature fusion, the proposed method can obtain accurate segmentation of retinal blood vessels by refining the segmentation features, which can provide another diagnosis method for computer-aided Chinese medical diagnosis.http://www.sciencedirect.com/science/article/pii/S2589377722000787Medical image processingAtrous space pyramid pooling (ASPP)Residual neural networkMulti-level modelRetinal vessels segmentation |
spellingShingle | Zhenchao CUI Shujie SONG Jing QI MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation Digital Chinese Medicine Medical image processing Atrous space pyramid pooling (ASPP) Residual neural network Multi-level model Retinal vessels segmentation |
title | MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation |
title_full | MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation |
title_fullStr | MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation |
title_full_unstemmed | MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation |
title_short | MF2ResU-Net: a multi-feature fusion deep learning architecture for retinal blood vessel segmentation |
title_sort | mf2resu net a multi feature fusion deep learning architecture for retinal blood vessel segmentation |
topic | Medical image processing Atrous space pyramid pooling (ASPP) Residual neural network Multi-level model Retinal vessels segmentation |
url | http://www.sciencedirect.com/science/article/pii/S2589377722000787 |
work_keys_str_mv | AT zhenchaocui mf2resunetamultifeaturefusiondeeplearningarchitectureforretinalbloodvesselsegmentation AT shujiesong mf2resunetamultifeaturefusiondeeplearningarchitectureforretinalbloodvesselsegmentation AT jingqi mf2resunetamultifeaturefusiondeeplearningarchitectureforretinalbloodvesselsegmentation |