Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
167 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Diversified and Personalized Multi-rater Medical Image Segmentation (2403.13417v1)

Published 20 Mar 2024 in cs.CV

Abstract: Annotation ambiguity due to inherent data uncertainties such as blurred boundaries in medical scans and different observer expertise and preferences has become a major obstacle for training deep-learning based medical image segmentation models. To address it, the common practice is to gather multiple annotations from different experts, leading to the setting of multi-rater medical image segmentation. Existing works aim to either merge different annotations into the "groundtruth" that is often unattainable in numerous medical contexts, or generate diverse results, or produce personalized results corresponding to individual expert raters. Here, we bring up a more ambitious goal for multi-rater medical image segmentation, i.e., obtaining both diversified and personalized results. Specifically, we propose a two-stage framework named D-Persona (first Diversification and then Personalization). In Stage I, we exploit multiple given annotations to train a Probabilistic U-Net model, with a bound-constrained loss to improve the prediction diversity. In this way, a common latent space is constructed in Stage I, where different latent codes denote diversified expert opinions. Then, in Stage II, we design multiple attention-based projection heads to adaptively query the corresponding expert prompts from the shared latent space, and then perform the personalized medical image segmentation. We evaluated the proposed model on our in-house Nasopharyngeal Carcinoma dataset and the public lung nodule dataset (i.e., LIDC-IDRI). Extensive experiments demonstrated our D-Persona can provide diversified and personalized results at the same time, achieving new SOTA performance for multi-rater medical image segmentation. Our code will be released at https://github.com/ycwu1997/D-Persona.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. The lung image database consortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans. Medical Physics, 38(2):915–931, 2011.
  2. Phiseg: Capturing uncertainty in medical image segmentation. In MICCAI, pages 119–127, 2019.
  3. The cramer distance as a solution to biased wasserstein gradients. arXiv preprint arXiv:1705.10743, 2017.
  4. Deep learning techniques for automatic mri cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE TMI, 37(11):2514–2525, 2018.
  5. Max-mig: an information theoretic approach for joint learning from crowds. In ICLR, 2018.
  6. Longitudinal multiple sclerosis lesion segmentation: resource and challenge. NeuroImage, 148:77–102, 2017.
  7. A survey on deep learning applied to medical images: from simple artificial neural networks to generative models. Neural Computing and Applications, 35(3):2291–2323, 2023.
  8. Instance-dependent label-noise learning with manifold-regularized transition matrix estimation. In CVPR, pages 16630–16639, 2022.
  9. U-net: deep learning for cell counting, detection, and morphometry. Nature Methods, 16(1):67–70, 2019.
  10. Rapid vessel segmentation and reconstruction of head and neck angiograms using 3d convolutional neural network. Nature Communications, 11(1):4829, 2020.
  11. Robust loss functions under label noise for deep neural networks. In AAAI, pages 1919–1925, 2017.
  12. Unetr: Transformers for 3d medical image segmentation. In WACV, pages 574–584, 2022.
  13. nnu-net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods, 18(2):203–211, 2021.
  14. Improving uncertainty estimation in convolutional neural networks using inter-rater agreement. In MICCAI, pages 540–548, 2019.
  15. Learning calibrated medical image segmentation via multi-rater agreement modeling. In CVPR, pages 12341–12351, 2021.
  16. Improving medical images classification with label noise using dual-uncertainty estimation. IEEE TMI, 41(6):1533–1546, 2022.
  17. On the effect of inter-observer variability for a reliable estimation of uncertainty of medical image segmentation. In MICCAI, pages 682–690, 2018.
  18. A probabilistic u-net for segmentation of ambiguous images. In NeurIPS, 2018.
  19. A hierarchical probabilistic u-net for modeling multi-scale ambiguities. arXiv preprint arXiv:1905.13077, 2019.
  20. International guideline for the delineation of the clinical target volumes (ctv) for nasopharyngeal carcinoma. Radiotherapy and Oncology, 126(1):25–36, 2018.
  21. Dividemix: Learning with noisy labels as semi-supervised learning. In ICLR, 2019.
  22. Deep learning segmentation of optical microscopy images improves 3-d neuron reconstruction. IEEE TMI, 36(7):1533–1541, 2017.
  23. Transformer-based annotation bias-aware medical image segmentation. In MICCAI, pages 24–34, 2023.
  24. Deep learning for automated contouring of primary tumor volumes by mri for nasopharyngeal carcinoma. Radiology, 291(3):677–686, 2019.
  25. Classification with noisy labels by importance reweighting. IEEE TPAMI, 38(3):447–461, 2015.
  26. Deep learning-based accurate delineation of primary gross tumor volume of nasopharyngeal carcinoma on heterogeneous magnetic resonance imaging: A large-scale and multi-center study. Radiotherapy and Oncology, 180:109480, 2023.
  27. Cryonuseg: A dataset for nuclei instance segmentation of cryosectioned h&e-stained histological images. Computers in Biology and Medicine, 132:104349, 2021.
  28. Deep learning-based gtv contouring modeling inter-and intra-observer variability in sarcomas. Radiotherapy and Oncology, 167:269–276, 2022.
  29. Learning from corrupted binary labels via class-probability estimation. In ICML, pages 125–134, 2015.
  30. The multimodal brain tumor image segmentation benchmark (brats). IEEE TMI, 34(10):1993–2024, 2014.
  31. A survey of crowdsourcing in medical image analysis. arXiv preprint arXiv:1902.09159, 2019.
  32. Adaptive sample selection for robust learning under label noise. In WACV, pages 3932–3942, 2023.
  33. Making deep neural networks robust to label noise: A loss correction approach. In CVPR, pages 1944–1952, 2017.
  34. Ambiguous medical image segmentation using diffusion models. In CVPR, pages 11536–11546, 2023.
  35. Addressing fairness in artificial intelligence for medical imaging. Nature Communications, 13(1):4581, 2022.
  36. Probabilistic modeling of inter-and intra-observer variability in medical image segmentation. In ICCV, pages 21097–21106, 2023.
  37. Training convolutional networks with noisy labels. arXiv preprint arXiv:1406.2080, 2014.
  38. Learning from noisy labels by regularized estimation of annotator confusion. In CVPR, pages 11244–11253, 2019.
  39. International consensus guidelines on clinical target volume delineation in rectal cancer. Radiotherapy and Oncology, 120(2):195–201, 2016.
  40. Interobserver variability in organ at risk delineation in head and neck cancer. Radiation Oncology, 16:1–11, 2021.
  41. Quantitative cerebral blood volume image synthesis from standard mri using image-to-image translation for brain tumors. Radiology, 308(2):e222471, 2023a.
  42. Medical matting: Medical image segmentation with uncertainty from the matting perspective. Computers in Biology and Medicine, 158:106714, 2023b.
  43. Symmetric cross entropy for robust learning with noisy labels. In ICCV, pages 322–330, 2019.
  44. Simultaneous truth and performance level estimation (staple): an algorithm for the validation of image segmentation. IEEE TMI, 23(7):903–921, 2004.
  45. Self-filtering: A noise-aware sample selection for label noise with confidence penalization. In ECCV, pages 516–532, 2022.
  46. Mutual consistency learning for semi-supervised medical image segmentation. Medical Image Analysis, 81:102530, 2022.
  47. Coactseg: Learning from heterogeneous data for new multiple sclerosis lesion segmentation. In MICCAI, pages 3–13, 2023.
  48. Disentangling human error from the ground truth in segmentation of medical images. In NeurIPS, pages 15750–15762, 2020.
  49. Generalized cross entropy loss for training deep neural networks with noisy labels. In NeurIPS, pages 8778–8788, 2018.
  50. Pluralistic image completion. In CVPR, pages 1438–1447, 2019.
  51. Graph convolutional label noise cleaner: Train a plug-and-play action classifier for anomaly detection. In CVPR, pages 1237–1246, 2019.
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now