A sequential geometry-reconstruction-based deep learning approach to improve accuracy and consistence of lumbar spine MRI image segmentation
Qian L, Chen J, Ma L, Urakov T, Liang L. A sequential geometry-reconstruction-based deep learning approach to improve accuracy and consistence of lumbar spine MRI image segmentation. Progress In Biomedical Optics And Imaging 2024, 12926: 1292634-1292634-8. DOI: 10.1117/12.3007064.Peer-Reviewed Original ResearchDeep learning approachImage segmentationLearning approachMedical image segmentationSelf-attention mechanismAccurate semantic segmentationImage feature extractionMRI image segmentationPosition embeddingsSemantic segmentationSwin-UnetFeature extractionErroneous fragmentsSpine MRI imagesShape representationNeural networkSegmentation resultsTexture similaritySegmentation modelAttention UNetMRI imagesGeometry reconstructionInternal informationIrrelevant informationImage features