Residual Convolutional Neural Network for Cardiac Image Segmentation and Heart Disease Diagnosis

Liu, Tao; Tian, Yun; Zhao, Shu; Huang, Xing; Wang, Qingjun

doi:10.1109/access.2020.2991424

Cited by 26 publications

(16 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In terms of the Cardiac MRI modality, there has been a lot of literature on deep learning-based segmentation 22 – 30 with diagnosis typically being a follow-up step, leveraging the segmentation masks and utilizing models such as random forests 31 , 32 , support vector machines 33 or a simple diagnostic rule 34 . These diagnosis models focus on two timepoints of the cardiac MRI per-patient: the phase of End-Diastole (ED) (maximum heart relaxation) and the phase of End-Systole (ES) (maximum heart contraction).…”

Section: Introductionmentioning

confidence: 99%

Federated learning for multi-center imaging diagnostics: a simulation study in cardiovascular disease

Linardos

Kushibar

Walsh

et al. 2022

Sci Rep

View full text Add to dashboard Cite

Deep learning models can enable accurate and efficient disease diagnosis, but have thus far been hampered by the data scarcity present in the medical world. Automated diagnosis studies have been constrained by underpowered single-center datasets, and although some results have shown promise, their generalizability to other institutions remains questionable as the data heterogeneity between institutions is not taken into account. By allowing models to be trained in a distributed manner that preserves patients’ privacy, federated learning promises to alleviate these issues, by enabling diligent multi-center studies. We present the first simulated federated learning study on the modality of cardiovascular magnetic resonance and use four centers derived from subsets of the M&M and ACDC datasets, focusing on the diagnosis of hypertrophic cardiomyopathy. We adapt a 3D-CNN network pretrained on action recognition and explore two different ways of incorporating shape prior information to the model, and four different data augmentation set-ups, systematically analyzing their impact on the different collaborative learning choices. We show that despite the small size of data (180 subjects derived from four centers), the privacy preserving federated learning achieves promising results that are competitive with traditional centralized learning. We further find that federatively trained models exhibit increased robustness and are more sensitive to domain shift effects.

show abstract

Section: Introductionmentioning

confidence: 99%

Federated learning for multi-center imaging diagnostics: a simulation study in cardiovascular disease

Linardos

Kushibar

Walsh

et al. 2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Automatic CMR diagnosis has thus far focused on isolated centers, with training happening locally and evaluation limited to IID subsets of the data [31][32][33][34] . We have presented the first federated CMR diagnosis study and showcased two distinct evaluation set-ups to quantify both IID and non-IID performance.…”

Section: Discussionmentioning

confidence: 99%

“…In terms of the Cardiac MRI modality, there has been a lot of literature on deep learning-based segmentation [22][23][24][25][26][27][28][29][30] with diagnosis typically being a follow-up step, leveraging the segmentation masks and utilizing models such as random forests 31,32 , support vector machines 33 or a simple diagnostic rule 34 . These diagnosis models focus on two timepoints of the cardiac MRI per-patient: the phase of End-Diastole (ED) (maximum heart relaxation) and the phase of End-Systole (ES) (maximum heart contraction).…”

Section: Introductionmentioning

confidence: 99%

“…On the same task, Cetin et al 33 used an SVM on top of radiomic features derived from manual segmentations, while Wolterink et al 32 evaluated a random forest classifier instead, both using a cross validation scheme. Liu et al 34 used an automated deep-learning based segmentation scheme, following up with a diagnostic rule. Despite the impressive classification performance reported in these studies, these models are both trained and evaluated solely on a single-center dataset (ACDC), and thus, no assumptions can be made regarding their generalization to unseen centers and larger datasets.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Federated Learning for Multi-Center Imaging Diagnostics: A Study in Cardiovascular Disease

Linardos

Kushibar²,

Walsh³

et al. 2021

Preprint

View full text Add to dashboard Cite

Deep learning models can enable accurate and efficient disease diagnosis, but have thus far been hampered by the data scarcity present in the medical world. Automated diagnosis studies have been constrained by underpowered single-center datasets, and although some results have shown promise, their generalizability to other institutions remains questionable as the data heterogeneity between institutions is not taken into account. By allowing models to be trained in a distributed manner that preserves patients’ privacy, federated learning promises to alleviate these issues, by enabling diligent multi-center studies. We present the first federated learning study on the modality of cardiovascular magnetic resonance (CMR) and use four centers derived from subsets of the M&M and ACDC datasets, focusing on the diagnosis of hypertrophic cardiomyopathy (HCM). We adapt a 3D-CNN network pretrained on action recognition and explore two different ways of incorporating shape prior information to the model, and four different data augmentation setups , systematically analyzing their impact on the different collaborative learning choices. We show that despite the small size of data (180 subjects derived from four centers), the privacy preserving federated learning achieves promising results that are competitive with traditional centralized learning. We further find that federatively trained models exhibit increased robustness and are more sensitive to domain shift effects.

show abstract

“…Therefore, neural networks can be used to train large amounts of labeled pneumonia data, learn the characteristics of pneumonia autonomously, and identify pneumonia to help overcome the limitations of human cognition and reduce errors [1,2]. However, in practical applications, although some neural networks such as convolutional neural networks (CNN) [3,4], long shortterm memory networks (LSTM) [5] and residual networks (ResNet) [6] can recognize images, they cannot process data well because the network structures are complex and require a lot of data. There are not too many high-quality pneumonia data available now, which is an important reason that hinders the network from learning semantic features of pictures.…”

Section: Introductionmentioning

confidence: 99%

Pneumonia Identification with Self-supervised Learning and Transfer Learning

Long

2021

Application of Intelligent Systems in Multi-Modal Information Analytics

View full text Add to dashboard Cite

Pneumonia is a common lung disease. Diagnosis through X-rays and computed tomography images is currently the best method for diagnosing pneumonia, but it requires the professional judgment of radiologists. However, with the rise of deep learning, convolutional neural networks can be accurately applied to image recognition and processing, laying the foundation for rapid disease diagnosis, which will ensure that patients receive timely treatment and save them time and money. In this paper, we applied self-supervised learning and transfer learning for high-level semantic features of pictures and for pneumonia identification, in order to learn more discriminative representations of images and improve the expressive ability of the network to identify pneumonia. The experimental result shows that both the accuracy and efficiency of network to identify pneumonia have been improved after applying self-supervised learning and transfer learning. The improvements in time and performance of the pneumonia identification task are of great value in reality, which also make this research full of meaning.

show abstract

Residual Convolutional Neural Network for Cardiac Image Segmentation and Heart Disease Diagnosis

Cited by 26 publications

References 24 publications

Federated learning for multi-center imaging diagnostics: a simulation study in cardiovascular disease

Federated learning for multi-center imaging diagnostics: a simulation study in cardiovascular disease

Federated Learning for Multi-Center Imaging Diagnostics: A Study in Cardiovascular Disease

Pneumonia Identification with Self-supervised Learning and Transfer Learning

Contact Info

Product

Resources

About