Semi-supervised Medical Image Segmentation through Dual-task Consistency

By Xiangde Luo, Jieneng Chen, Tao Song, Yinan Chen, Guotai Wang, Shaoting Zhang. AAAI Conference on Artificial Intelligence, 2020

796 citations recorded by the source corpus.

Deep learning-based semi-supervised learning (SSL) algorithms have led to promising results in medical images segmentation and can alleviate doctors' expensive annotations by leveraging unlabeled data. However, most of the existing SSL algorithms in literature tend to regularize the model training by perturbing networks and/or data. Observing that multi/dual-task learning attends to various levels of information which have inherent prediction perturbation, we ask the question in this work: can we explicitly build task-level regularization rather than implicitly constructing networks- and/or data-level perturbation and then regularization for SSL? To answer this question, we propose a novel dual-task-consistency semi-supervised framework for the first time. Concretely, we use a dual-task deep network that jointly predicts a pixel-wise segmentation map and a geometry-aware level set representation of the target. The level set representation is converted to an approximated segmentation map through a differentiable task transform layer. Simultaneously, we introduce a dual-task consistency regularization between the level set-derived segmentation maps and directly predicted segmentation maps for both labeled and unlabeled data. Extensive experiments on two public datasets show that our method can largely improve the performance by incorporating the unlabeled data. Meanwhile, our framework outperforms the state-of-the-art semi-supervised learning methods.

Subfields: Artificial Intelligence, Biomedical Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/f356edffeb43bc2e1050071e59cd718f5b0f60a5

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jianfei Cai
2. Alan L. Yuille
3. Yuyin Zhou
4. Qihang Yu
5. Le Lu
6. Guotai Wang
7. Shaoting Zhang
8. Puyang Wang

In Biomedical Engineering

1. Puyang Wang
2. Alan L. Yuille
3. James S. Duncan
4. Jianfei Cai
5. Guotai Wang
6. Rushi Jiao
7. Jieneng Chen

In Computer Vision and Pattern Recognition

1. Jianfei Cai
2. Alan L. Yuille
3. Qihang Yu
4. Le Lu
5. Shaoting Zhang
6. Yuyin Zhou
7. Dacheng Tao
8. Luc Van Gool

Citations in the corpus

• SimCVD: Simple Contrastive Voxel-Wise Representation Distillation for Semi-Supervised Medical Image Segmentation (2021, methodology): “[23], we adopt a sliding window strategy, which uses a stride with 18×18×4 for the LA and 16×16×16 for the pancreas.”
• TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation (2021, background): “magnetic resonance (MR) [17], organ segmentation from computed tomography (CT) [7, 9, 18,20] and polyp segmentation [21] from colonoscopy videos.”
• Bidirectional Copy-Paste for Semi-Supervised Medical Image Segmentation (2023, neutral): “LA dataset We compare our framework on LA dataset with various competitors: UA-MT [41], SASSNet [12], DTC [16], URPC [17], MC-Net [36] and SS-Net [35].”
• Mutual consistency learning for semi-supervised medical image segmentation (2021, methodology): “They also investigated a semi-supervised model to achieve the multi-scale consistency for the gross target volume segmentation (Luo et al., 2021b).”
• Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation (2022, background): “[68] combine the level set function regression task with”
• Efficient Semi-supervised Gross Target Volume of Nasopharyngeal Carcinoma Segmentation via Uncertainty Rectified Pyramid Consistency (2020, methodology): “[15] proposed a dual-task consistency framework for SSL by representing segmentation as a pixel-wise classification task and a level set regression task simultaneously, the difference between which was minimized during training.”
• Semi-supervised Left Atrium Segmentation with Mutual Consistency Training (2021, methodology): “Most of the experiment settings were set same as the recent SOTA methods [6,7,17].”
• LViT: Language Meets Vision Transformer in Medical Image Segmentation (2022, methodology): “We compare our method with both the traditional SOTA semi-supervised medical image segmentation methods, such as DTC [50], PLCT [51], and MC-Net+ [52], and the multimodal methods, such as LAVT [27] and GLoRIA [49].”