Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation
In this paper, we propose a dense multi-scale adaptive graph convolutional network (<i>DMA-GCN</i>) method for automatic segmentation of the knee joint cartilage from MR images. Under the multi-atlas setting, the suggested approach exhibits several novelties, as described in the followin...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-03-01
|
| Series: | Bioengineering |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2306-5354/11/3/278 |
| _version_ | 1797242006602252288 |
|---|---|
| author | Christos Chadoulos Dimitrios Tsaopoulos Andreas Symeonidis Serafeim Moustakidis John Theocharis |
| author_facet | Christos Chadoulos Dimitrios Tsaopoulos Andreas Symeonidis Serafeim Moustakidis John Theocharis |
| author_sort | Christos Chadoulos |
| collection | DOAJ |
| description | In this paper, we propose a dense multi-scale adaptive graph convolutional network (<i>DMA-GCN</i>) method for automatic segmentation of the knee joint cartilage from MR images. Under the multi-atlas setting, the suggested approach exhibits several novelties, as described in the following. First, our models integrate both local-level and global-level learning simultaneously. The local learning task aggregates spatial contextual information from aligned spatial neighborhoods of nodes, at multiple scales, while global learning explores pairwise affinities between nodes, located globally at different positions in the image. We propose two different structures of building models, whereby the local and global convolutional units are combined by following an alternating or a sequential manner. Secondly, based on the previous models, we develop the <i>DMA-GCN</i> network, by utilizing a densely connected architecture with residual skip connections. This is a deeper <i>GCN</i> structure, expanded over different block layers, thus being capable of providing more expressive node feature representations. Third, all units pertaining to the overall network are equipped with their individual adaptive graph learning mechanism, which allows the graph structures to be automatically learned during training. The proposed cartilage segmentation method is evaluated on the entire publicly available Osteoarthritis Initiative (<i>OAI</i>) cohort. To this end, we have devised a thorough experimental setup, with the goal of investigating the effect of several factors of our approach on the classification rates. Furthermore, we present exhaustive comparative results, considering traditional existing methods, six deep learning segmentation methods, and seven graph-based convolution methods, including the currently most representative models from this field. The obtained results demonstrate that the <i>DMA-GCN</i> outperforms all competing methods across all evaluation measures, providing <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mi>S</mi><mi>C</mi><mo>=</mo><mn>95.71</mn><mo>%</mo></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mi>S</mi><mi>C</mi><mo>=</mo><mn>94.02</mn><mo>%</mo></mrow></semantics></math></inline-formula> for the segmentation of femoral and tibial cartilage, respectively. |
| first_indexed | 2024-04-24T18:32:22Z |
| format | Article |
| id | doaj.art-72841006d678476e98ca965919c3e842 |
| institution | Directory Open Access Journal |
| issn | 2306-5354 |
| language | English |
| last_indexed | 2024-04-24T18:32:22Z |
| publishDate | 2024-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Bioengineering |
| spelling | doaj.art-72841006d678476e98ca965919c3e8422024-03-27T13:21:57ZengMDPI AGBioengineering2306-53542024-03-0111327810.3390/bioengineering11030278Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage SegmentationChristos Chadoulos0Dimitrios Tsaopoulos1Andreas Symeonidis2Serafeim Moustakidis3John Theocharis4Department of Electrical & Computer Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceInstitute for Bio-Economy and Agri-Technology, Centre for Research and Technology—Hellas, 38333 Volos, GreeceDepartment of Electrical & Computer Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceDepartment of Electrical & Computer Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceDepartment of Electrical & Computer Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceIn this paper, we propose a dense multi-scale adaptive graph convolutional network (<i>DMA-GCN</i>) method for automatic segmentation of the knee joint cartilage from MR images. Under the multi-atlas setting, the suggested approach exhibits several novelties, as described in the following. First, our models integrate both local-level and global-level learning simultaneously. The local learning task aggregates spatial contextual information from aligned spatial neighborhoods of nodes, at multiple scales, while global learning explores pairwise affinities between nodes, located globally at different positions in the image. We propose two different structures of building models, whereby the local and global convolutional units are combined by following an alternating or a sequential manner. Secondly, based on the previous models, we develop the <i>DMA-GCN</i> network, by utilizing a densely connected architecture with residual skip connections. This is a deeper <i>GCN</i> structure, expanded over different block layers, thus being capable of providing more expressive node feature representations. Third, all units pertaining to the overall network are equipped with their individual adaptive graph learning mechanism, which allows the graph structures to be automatically learned during training. The proposed cartilage segmentation method is evaluated on the entire publicly available Osteoarthritis Initiative (<i>OAI</i>) cohort. To this end, we have devised a thorough experimental setup, with the goal of investigating the effect of several factors of our approach on the classification rates. Furthermore, we present exhaustive comparative results, considering traditional existing methods, six deep learning segmentation methods, and seven graph-based convolution methods, including the currently most representative models from this field. The obtained results demonstrate that the <i>DMA-GCN</i> outperforms all competing methods across all evaluation measures, providing <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mi>S</mi><mi>C</mi><mo>=</mo><mn>95.71</mn><mo>%</mo></mrow></semantics></math></inline-formula> and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mi>S</mi><mi>C</mi><mo>=</mo><mn>94.02</mn><mo>%</mo></mrow></semantics></math></inline-formula> for the segmentation of femoral and tibial cartilage, respectively.https://www.mdpi.com/2306-5354/11/3/278knee cartilage osteoarthritis (<i>KOA</i>)magnetic resonance imaging (<i>MRI</i>) segmentationmulti-atlasgraph neural networks (<i>GNNs</i>)deep learninggraph learning |
| spellingShingle | Christos Chadoulos Dimitrios Tsaopoulos Andreas Symeonidis Serafeim Moustakidis John Theocharis Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation Bioengineering knee cartilage osteoarthritis (<i>KOA</i>) magnetic resonance imaging (<i>MRI</i>) segmentation multi-atlas graph neural networks (<i>GNNs</i>) deep learning graph learning |
| title | Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation |
| title_full | Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation |
| title_fullStr | Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation |
| title_full_unstemmed | Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation |
| title_short | Dense Multi-Scale Graph Convolutional Network for Knee Joint Cartilage Segmentation |
| title_sort | dense multi scale graph convolutional network for knee joint cartilage segmentation |
| topic | knee cartilage osteoarthritis (<i>KOA</i>) magnetic resonance imaging (<i>MRI</i>) segmentation multi-atlas graph neural networks (<i>GNNs</i>) deep learning graph learning |
| url | https://www.mdpi.com/2306-5354/11/3/278 |
| work_keys_str_mv | AT christoschadoulos densemultiscalegraphconvolutionalnetworkforkneejointcartilagesegmentation AT dimitriostsaopoulos densemultiscalegraphconvolutionalnetworkforkneejointcartilagesegmentation AT andreassymeonidis densemultiscalegraphconvolutionalnetworkforkneejointcartilagesegmentation AT serafeimmoustakidis densemultiscalegraphconvolutionalnetworkforkneejointcartilagesegmentation AT johntheocharis densemultiscalegraphconvolutionalnetworkforkneejointcartilagesegmentation |