Multi-input Unet model based on the integrated block and the aggregation connection for MRI brain tumor segmentation

Fang, Lingling; Wang, Xin

doi:10.1016/j.bspc.2022.104027

Cited by 13 publications

(3 citation statements)

References 84 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Three variants of a lightweight hierarchical convolutional network were constructed [105] using a residual hierarchical convolutional network for fusing the advantages of MRI images of T1c, T2 and FLAIR for segmentation. The U-Net architecture was modified in [106] by the integration of many parallel convolutions with pooling in the down sampling part and aggregation of features obtained from convolution at each layer in up sampling part with the corresponding down sampling part for segmentation of tumor using the masked four types of MRI brain images. The author proposed a transformer based segmentation approach called UNETR [107] in which a transformer encoder connects directly to decoder via skip connection for medical image segmentation.The U-Net was combined with a dual encoder based R -Transformer network in [108] which constituted a feature branch for extracting global context information and a patch branch for extracting semantic features from the different modalities of MRI images.…”

Section: B U-net Based Segmentation Of Brain Tumormentioning

confidence: 99%

“…In this FLAIR MRI scan was used for the segmentation of WT, TC and ET. The difference between the variants of U-Net architectures proposed in [69], [88], [93], [98], [99], [100], [101], [103], [106], [108], [109], [110], [111], [112], [116] and [117] are summarized in Table V. The U-Net architecture was modified by hybridizing the network [120] with residual block and attention block (in between the concatenation of down sampling part with up sampling part) and deep supervision block at the end of the decoder part from multi-resolution T1c, T2 and FLAIR MRI images.…”

Section: B U-net Based Segmentation Of Brain Tumormentioning

confidence: 99%

“…The measured values of Dice score and Hausdroff distance (Hd95) by the methods [112], [117], [88], [101], [110], [120], [132], [93], [117], [121], [122], [123], [134], [135], [103], [106], [116], [123], [130], [108], [116], [126], [131], [133], [144], [99] and [98] are extracted in Table VII.…”

Section: Fig 13 Hausdorff Distance Estimation Curvementioning

confidence: 99%

See 2 more Smart Citations

Comparative Study on Architecture of Deep Neural Networks for Segmentation of Brain Tumor using Magnetic Resonance Images

Preetha,

Priyadarsini,

Nisha

2023

IEEE Access

View full text Add to dashboard Cite

The state-of-the-art works for the segmentation of brain tumor using the images acquired by Magnetic Resonance Imaging (MRI) with their performances are analyzed in this comparative study. First, the architectures of convolutional neural networks (CNN) and the variants of U-shaped Network (U-Net), a kind of Deep Neural Network (DNN) are compared and their differences are highlighted. The publicly available datasets of MRI images specifically Brain Tumor Segmentation (BraTS) are also discussed. Next, the performances of tumor segmentation of various methods in the literature are compared using the parameters such as Dice score and Hausdroff distance ( 95). This study concludes that the U-Net based architectures using the BraTS-2019 dataset outperform well compared with other CNN based architectures.

show abstract