Matches in SemOpenAlex for { <https://semopenalex.org/work/W2896808624> ?p ?o ?g. }
Showing items 1 to 78 of
78
with 100 items per page.
- W2896808624 endingPage "e379" @default.
- W2896808624 startingPage "e379" @default.
- W2896808624 abstract "To develop a novel deep-learning based prostate segmentation method with magnetic resonance images for radiation therapy treatment planning. Accurate delineation of the prostate gland is important for diagnosis, management and prognosis of radiation treatment of the prostate cancer. Currently, MR prostate segmentation is mainly done by radiation oncologists based almost entirely on visual inspection on a slice-by-slice basis, which is time-consuming, subject to inter- and intra-reader variations. We proposed a novel Encoder-Decoder Densely Connected Convolutional Network (DenseNet) to automatic segment the prostate gland with T2 MR images. Our model consists of two interconnected pathways, a dense encoder pathway, which learns discriminative high-level image features and a dense decoder pathway, which predicts the final segmentation in the pixel level. Instead of using the convolutional network as the basic unit in the encoder-decoder framework, the DenseNet preserves the maximum information flow between layers by a densely-connected mechanism. In addition, a novel loss function that jointly consider the encoder-deconder reconstruction error and the prediction error is applied to optimize the feature learning and segmentation result. The performance of prostate segmentation method is evaluated using the Dice Similarity Coefficient (DSC). The robustness and effectiveness of the proposed deep learning model were tested on two cohorts: 112 cases from the PROSTATEx-2 Challenge; and 132 cases from our institutional review board approved patient database. Our automatic segmentation result shows high agreement to the clinical segmentation results by multiple experienced radiation oncologists. We compared our proposed model with two baseline methods including the Encoder-Decoder convolutional Network and Encoder-Decoder DenseNet using traditional cross entropy. Our method achieved the highest DSC result of 0.945 among other baseline models. Better performance was also achieved the in comparison with three recent segmentation methods: Fully Convolutional Networks, U-Net and V-Net. The novel deep learning based automatic prostate segmentation method produces the best performance and provided valuable information for treatment planning and precision radiation therapy." @default.
- W2896808624 created "2018-10-26" @default.
- W2896808624 creator A5023976830 @default.
- W2896808624 creator A5061512571 @default.
- W2896808624 creator A5068509121 @default.
- W2896808624 creator A5073968803 @default.
- W2896808624 creator A5078041216 @default.
- W2896808624 creator A5082528902 @default.
- W2896808624 creator A5084187424 @default.
- W2896808624 date "2018-11-01" @default.
- W2896808624 modified "2023-10-16" @default.
- W2896808624 title "Automatic Prostate Segmentation using Deep Learning and MR Images" @default.
- W2896808624 doi "https://doi.org/10.1016/j.ijrobp.2018.07.1129" @default.
- W2896808624 hasPublicationYear "2018" @default.
- W2896808624 type Work @default.
- W2896808624 sameAs 2896808624 @default.
- W2896808624 citedByCount "0" @default.
- W2896808624 crossrefType "journal-article" @default.
- W2896808624 hasAuthorship W2896808624A5023976830 @default.
- W2896808624 hasAuthorship W2896808624A5061512571 @default.
- W2896808624 hasAuthorship W2896808624A5068509121 @default.
- W2896808624 hasAuthorship W2896808624A5073968803 @default.
- W2896808624 hasAuthorship W2896808624A5078041216 @default.
- W2896808624 hasAuthorship W2896808624A5082528902 @default.
- W2896808624 hasAuthorship W2896808624A5084187424 @default.
- W2896808624 hasConcept C108583219 @default.
- W2896808624 hasConcept C111919701 @default.
- W2896808624 hasConcept C118505674 @default.
- W2896808624 hasConcept C121608353 @default.
- W2896808624 hasConcept C126322002 @default.
- W2896808624 hasConcept C138885662 @default.
- W2896808624 hasConcept C153180895 @default.
- W2896808624 hasConcept C154945302 @default.
- W2896808624 hasConcept C2776235491 @default.
- W2896808624 hasConcept C2776401178 @default.
- W2896808624 hasConcept C31972630 @default.
- W2896808624 hasConcept C41008148 @default.
- W2896808624 hasConcept C41895202 @default.
- W2896808624 hasConcept C71924100 @default.
- W2896808624 hasConcept C89600930 @default.
- W2896808624 hasConcept C97931131 @default.
- W2896808624 hasConceptScore W2896808624C108583219 @default.
- W2896808624 hasConceptScore W2896808624C111919701 @default.
- W2896808624 hasConceptScore W2896808624C118505674 @default.
- W2896808624 hasConceptScore W2896808624C121608353 @default.
- W2896808624 hasConceptScore W2896808624C126322002 @default.
- W2896808624 hasConceptScore W2896808624C138885662 @default.
- W2896808624 hasConceptScore W2896808624C153180895 @default.
- W2896808624 hasConceptScore W2896808624C154945302 @default.
- W2896808624 hasConceptScore W2896808624C2776235491 @default.
- W2896808624 hasConceptScore W2896808624C2776401178 @default.
- W2896808624 hasConceptScore W2896808624C31972630 @default.
- W2896808624 hasConceptScore W2896808624C41008148 @default.
- W2896808624 hasConceptScore W2896808624C41895202 @default.
- W2896808624 hasConceptScore W2896808624C71924100 @default.
- W2896808624 hasConceptScore W2896808624C89600930 @default.
- W2896808624 hasConceptScore W2896808624C97931131 @default.
- W2896808624 hasIssue "3" @default.
- W2896808624 hasLocation W28968086241 @default.
- W2896808624 hasOpenAccess W2896808624 @default.
- W2896808624 hasPrimaryLocation W28968086241 @default.
- W2896808624 hasRelatedWork W1971623867 @default.
- W2896808624 hasRelatedWork W1982770690 @default.
- W2896808624 hasRelatedWork W2467256294 @default.
- W2896808624 hasRelatedWork W2510758617 @default.
- W2896808624 hasRelatedWork W2546942002 @default.
- W2896808624 hasRelatedWork W2554403468 @default.
- W2896808624 hasRelatedWork W2798217096 @default.
- W2896808624 hasRelatedWork W3102644505 @default.
- W2896808624 hasRelatedWork W4281756384 @default.
- W2896808624 hasRelatedWork W4300535155 @default.
- W2896808624 hasVolume "102" @default.
- W2896808624 isParatext "false" @default.
- W2896808624 isRetracted "false" @default.
- W2896808624 magId "2896808624" @default.
- W2896808624 workType "article" @default.