Federated Learning for Site Aware Chest Radiograph Screening
Abstract
The shortage of Radiologists is inspiring the development of Deep Learning (DL) based solutions for detecting cardio, thoracic and pulmonary pathologies in Chest radiographs through multi-institutional collaborations. However, sharing the training data across multiple sites is often impossible due to privacy, ownership and technical challenges. Although Federated Learning (FL) has emerged as a solution to this, the large variations in disease prevalence and co-morbidity distributions across the sites may hinder proper training. We propose a DL architecture with a Convolutional Neural Network (CNN) followed by a Graph Neural Network (GNN) to address this issue. The CNN-GNN model is trained by modifying the Federated Averaging algorithm. The CNN weights are shared across all sites to extract robust features while separate GNN models are trained at each site to leverage the local co-morbidity dependencies for multi-label disease classification. The CheXpert dataset is partitioned across five sites to simulate the FL set up. Federated training did not show any significant drop in performance over centralized training. The site-specific GNN models also demonstrated their efficacy in modelling local disease co-occurrence statistics leading to an average area under the ROC curve of 0.79 with a 1.74% improvement.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Six out of 17 reviewed articles used chest X-ray images.
“According to our investigation, six articles [18], [23], [25], [27], [29], [32] have used chest X-ray images out of 17.”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
Chakravarty et al. [23] proposed a multilabel cardiac diseases classification where 14 classes were examined.
“In addition to that, a multilabel cardiac diseases classification has been proposed by Chakravarty et al. [23], where 14 classes were examined.”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
CheXpert, the largest dataset found under the investigation (overall 223,414 CXR images), was considered by Chakravarty et al. [23].
“CheXpert, the largest dataset (overall 223,414 CXR image) found under our investigation was considered by Chakravarty et al. [23].”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
ResNet18 was used in [23], [26], [27], and [29]; ResNet50 in [18] and [24]; ResNet101 in [24].
“Residual Network (ResNet) is also a commonly used algorithm that can be constructed by different numbers of layers, e.g., ResNet18 ( [23], [26], [27], [29]), ResNet50 ( [18], [24]), ResNet101 ( [24]).”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
The remaining three articles performed multi-class classification: Chakravarty et al. [23] used 14 classes, Połap et al. [17] seven classes, and Li et al. [20] four classes; their performances were AUC 80%, accuracy 70%, and 76%, respectively.
“Low: The rest three articles performed multi-class classification, where Chakravarty et al. [23] 14, Połap et al. [17] seven, and Li et al. [20] have considered four classes, their acquired performances are AUC 80%, accuracy 70% and 76% respectively.”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
Chakravarty et al. [23] had an AUC score of 80% with FL, while with the same dataset and ML algorithm articles [67] and [68] had 86% and 87% AUC, respectively.
“Then as well Chakravarty et al. [23] has an AUC score of 80% with FL environment, but with same dataset and ML algorithm article [67] and [68] have 86% and 87% AUC respectively.”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
In Chakravarty et al. [23], performance comparison with state-of-the-art methods is not given.
“Performance comparison with”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
In Chakravarty et al. [23], non-IID data is not considered.
“non-IID data is not consid-”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
Chakravarty et al. [23] applied FL to chest radiograph screening using deep learning methods.
“FL applied to chest radiograph screening”
From A systematic review on federated learning in medical image analysis (2023) · cites this paper as [23]
The citation lists haven’t been fetched yet
References 21 · Cited by 24
- Authors
- Chakravarty, A., Kar, A., Sethuraman, R., & Sheet, D.
- Affiliations
- Indian Institute of Technology Kharagpur (IN)
- Intel (India) (IN)
- Venue
- 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)
- Publisher
- IEEE
- Published
- 2021-04-13
- Pages
- 1077-1081
- OpenAlex
- W3165352259
- Cited by
- 24
- References
- 21
- Citations by year
- 2026: 3 · 2025: 10 · 2024: 4 · 2023: 3 · 2022: 4
- FWCI
- 3.46
- Type
- conference paper
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE International Symposium on Biomedical Imaging (ISBI) (publishedVersion)
- Topics
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › COVID-19 diagnosis using AI
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- federated learning, graph neural networks, convolutional neural networks, disease co-occurrence, FedAvg, CheXpert, radiologist shortage
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
