A Comprehensive Survey on Federated Learning Techniques for Healthcare Informatics

Table 3

FL applications for medical image processing.

Ref. No	Technologies used	Key contributions	Limitations
[56]	Differential privacy techniques	Using the BraTS dataset, assess the usefulness of practical FL methods for segmenting brain tumors	It is impossible to collect and share patient data in a centralized data lake
[58]	CFL-based collaborative learning framework	To highlight the potential of intelligent processing of clinical data at the edge, open research issues related to deploying ML at the edge for healthcare applications that re- quire further investigation	(i) Image size and quality
			(ii) Contrast and brightness level, and
			(iii) Positioning of subjects
[59]	Skin imaging technology	The proposed model contains two core contributions:	Health practitioners usually apply manual or computer vision-based tools for skin tumor diagnosis, which may cause misinterpretation of the disease and lead to a longer analysis time
		(i) The model was deployed on the cloud server, and
		(ii) Its deployment on the edges majorly contributes toward adaptability by continuously updating
[60]	FL techniques	The performance of FL may be enhanced with more images or data augmentation	Comparisons of FL with unequal data distribution, data augmentation, and one-shot learning are required to explore the implications of data imbalance
[62]	3D-convolutional neural network technique	FL study on cardiovascular magnetic resonance diagnosis and demonstrate that FL performance is comparable to central database server	Patient privacy