Joint Sequence Learning and Cross-Modality Convolution for 3D Biomedical Segmentation
Citations
1,561 citations
Cites methods from "Joint Sequence Learning and Cross-M..."
...To solve this problem, some researchers proposed to use tri-planar schemes or RNN to probe the 3D contexts [4], [20], [21]....
[...]
277 citations
138 citations
Cites methods from "Joint Sequence Learning and Cross-M..."
...For each type of regions, we compute DSC [34], PPV, and sensitivity [35] as quantitative evaluation metrics....
[...]
136 citations
Cites background from "Joint Sequence Learning and Cross-M..."
...and other multi-modality CNN research [10], [47], [48], [50]....
[...]
105 citations
References
72,897 citations
Additional excerpts
...Different from traditional LSTM [11], convLSTM replaces the matrix multiplication by the convolution operators in state-to-state and input-to-state transitions, which preservers the spatial information for long-term sequences....
[...]
49,590 citations
30,843 citations
"Joint Sequence Learning and Cross-M..." refers methods in this paper
...Each convolution layer uses the kernel size 3 × 3 to produce a set of feature maps, which are further applied by a batch normalization layer [12] and an element-wise rectified-linear non-linearity (ReLU)....
[...]
28,225 citations
19,534 citations
"Joint Sequence Learning and Cross-M..." refers background or methods or result in this paper
...U-Net [20] consists of a contracting path that contains multiple convolutions for downsampling, and a expansive path that has several deconvolution layers to up-sample the features and concatenate the cropped feature maps from the contracting path....
[...]
...Recent image segmentation models [8, 15, 9, 20] fuse multi-resolution feature maps with the concatenation....
[...]
...Some methods apply 2D segmentation to 3D biomedical data [20] [5]....
[...]
...The results show that the proposed model: MME + MRF + CMC and MME + MRF + CMC + convLSTM achieve the best performance and outperform the baseline method U-Net [20]....
[...]
...Segmentation results of variants of our method and U-Net [20] baseline....
[...]