Collaborative Unsupervised Domain Adaptation for Medical Image Diagnosis

Zhang, Yifan; Wei, Ying; Wu, Qingyao; Zhao, Peilin; Niu, Shuaicheng; Huang, Junzhou; Tan, Mingkui

doi:10.1109/tip.2020.3006377

Cited by 142 publications

(60 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A related framework is unsupervised domain adaptation, where we are only given unlabelled data for the target domain. Unsupervised domain adaptation has also been applied to medical imaging [26,58,86]. Our benchmark focuses on the domain generalization setting, where only labeled data from multiple training environments are available, and the goal is to be able to generalize to all unseen test domains.…”

Section: Model Transferrability In Medical Settingsmentioning

confidence: 99%

An empirical framework for domain generalization in clinical settings

Zhang

Dullerud

Seyyed-Kalantari

et al. 2021

Proceedings of the Conference on Health, Inference, and Learning

View full text Add to dashboard Cite

Clinical machine learning models experience significantly degraded performance in datasets not seen during training, e.g., new hospitals or populations. Recent developments in domain generalization offer a promising solution to this problem by creating models that learn invariances across environments. In this work, we benchmark the performance of eight domain generalization methods on multi-site clinical time series and medical imaging data. We introduce a framework to induce synthetic but realistic domain shifts and sampling bias to stress-test these methods over existing nonhealthcare benchmarks. We find that current domain generalization methods do not achieve significant gains in out-of-distribution performance over empirical risk minimization on real-world medical imaging data, in line with prior work on general imaging datasets. However, a subset of realistic induced-shift scenarios in clinical time series data exhibit limited performance gains. We characterize these scenarios in detail, and recommend best practices for domain generalization in the clinical setting. CCS CONCEPTS• Computing methodologies → Machine learning; • Applied computing → Health informatics; • General and reference → Empirical studies.

show abstract

Section: Model Transferrability In Medical Settingsmentioning

confidence: 99%

An empirical framework for domain generalization in clinical settings

Zhang

Dullerud

Seyyed-Kalantari

et al. 2021

Proceedings of the Conference on Health, Inference, and Learning

View full text Add to dashboard Cite

show abstract

“…There may be heterogeneity in the imaging data due to the different parameters in the scanning process at different centers, which can reduce the generalization ability of the prediction models. As indicated in some recent studies (29)(30)(31), domain adaptive technology based on deep learning may be applied to reduce the difference in data distribution to improve the generalization ability of the method in further studies.…”

Section: Discussionmentioning

confidence: 99%

Predicting the recurrence risk of pancreatic neuroendocrine neoplasms after radical resection using deep learning radiomics with preoperative computed tomography images

et al. 2021

View full text Add to dashboard Cite

Background: To establish and validate a prediction model for pancreatic neuroendocrine neoplasms (pNENs) recurrence after radical surgery with preoperative computed tomography (CT) images. Methods:We retrospectively collected data from 74 patients with pathologically confirmed pNENs (internal group: 56 patients, Hospital I; external validation group: 18 patients, Hospital II). Using the internal group, models were trained with CT findings evaluated by radiologists, radiomics, and deep learning radiomics (DLR) to predict 5-year pNEN recurrence. Radiomics and DLR models were established for arterial (A), venous (V), and arterial and venous (A&V) contrast phases. The model with the optimal performance was further combined with clinical information, and all patients were divided into high-and low-risk groups to analyze survival with the Kaplan-Meier method.Results: In the internal group, the areas under the curves (AUCs) of DLR-A, DLR-V, and DLR-A&V models were 0.80, 0.58, and 0.72, respectively. The corresponding radiomics AUCs were 0.74, 0.68, and 0.70.The AUC of the CT findings model was 0.53. The DLR-A model represented the optimum; added clinical information improved the AUC from 0.80 to 0.83. In the validation group, the AUCs of DLR-A, DLR-V, and DLR-A&V models were 0.77, 0.48, and 0.64, respectively, and those of radiomics-A, radiomics-V, and radiomics-A&V models were 0.56, 0.52, and 0.56, respectively. The AUC of the CT findings model was 0.52. In the validation group, the comparison between the DLR-A and the random models showed a trend of significant difference (P=0.058). Recurrence-free survival differed significantly between high-and low-risk groups (P=0.003).Conclusions: Using DLR, we successfully established a preoperative recurrence prediction model for pNEN patients after radical surgery. This allows a risk evaluation of pNEN recurrence, optimizing clinical decision-making.

show abstract

“…Gholami et al [175] trained a model based on synthetic brain tumorous dataset (generated from healthy brains) and tested in the real brain tumor dataset BraTS'18 [176]- [178]. Zhang et al [179] extended the model from labeled set (source) to unlabeled set (target) in adenoma diagnosis. Mahmood et al [180] trained the model learning from synthetic organ model images and tested in the monocular depth estimation for endoscopy images.…”

Section: G Transfer Learning and Domain Adaptationmentioning

confidence: 99%

Application and Construction of Deep Learning Networks in Medical Imaging

Torres-Velázquez

Chen

et al. 2021

IEEE Trans. Radiat. Plasma Med. Sci.

View full text Add to dashboard Cite

Deep learning (DL) approaches are part of the machine learning (ML) subfield concerned with the development of computational models to train artificial intelligence systems. DL models are characterized by automatically extracting high-level features from the input data to learn the relationship between matching datasets. Thus, its implementation offers an advantage over common ML methods that often require the practitioner to have some domain knowledge of the input data to select the best latent representation. As a result of this advantage, DL has been successfully applied within the medical imaging field to address problems such as disease classification and tumor segmentation for which it is difficult or impossible to determine which image features are relevant. Therefore, taking into consideration the positive impact of DL on the medical imaging field, this paper reviews the key concepts associated with its evolution and implementation. The sections of this review summarize the milestones related to the development of the DL field, followed by a description of the elements of deep neural network and an overview of its application within the medical imaging field. Subsequently, the key steps necessary to implement a supervised DL application are defined, and associated limitations are discussed.

show abstract

Collaborative Unsupervised Domain Adaptation for Medical Image Diagnosis

Cited by 142 publications

References 40 publications

An empirical framework for domain generalization in clinical settings

An empirical framework for domain generalization in clinical settings

Predicting the recurrence risk of pancreatic neuroendocrine neoplasms after radical resection using deep learning radiomics with preoperative computed tomography images

Application and Construction of Deep Learning Networks in Medical Imaging

Contact Info

Product

Resources

About