Sharing Massive Biomedical Data at Magnitudes Lower Bandwidth Using Implicit Neural Function

Abstract

Efficient storage and sharing of massive biomedical data would open up their wide accessibility to different institutions and disciplines. However, compressors tailored for natural photos/videos are rapidly limited for biomedical data, while emerging deep learning based methods demand huge training data and are difficult to generalize. Here we propose to conduct biomedical data compRession with Implicit nEural Function (BRIEF) by representing the original data with compact deep neural networks, which are data specific and thus have no generalization issues. Benefiting from the strong representation capability of implicit neural function, BRIEF achieves significantly higher-fidelity on diverse biomedical data than existing techniques. Besides, BRIEF is of consistent performance across the whole data volume, supports customized spatially-varying fidelity. BRIEF's multi-fold advantageous features also serve reliable downstream tasks at low bandwidth. Our approach will facilitate biomedical data sharing at low bandwidth and maintenance costs, and promote collaboration and progress in the biomedical field.

Semantic redundancy-aware implicit neural compression for multidimensional biomedical image data

“…3b. Meanwhile, we compared our method with JPEG compression, conventional INR compression and autoencoder-based compression methods (CAE) 14,[22][23][24] , demonstrating the superior visual fidelity achieved by our approach.…”

“…We further enlarged the regions of interest (ROIs) within these patches and displayed them side-by-side with their corresponding probability saliency maps in Figure 3(b). Furthermore, we compared our method with JPEG compression and conventional INR compression methods applied to the entire dataset [24][25][26] , demonstrating superior visual fidelity achieved by our approach.…”

“…Additionally, with the universal approximation theorem of neural networks 23,24 , INR implemented using Multi-Layer Perceptron (MLP) can fit any complex function with a sufficient number of parameters, resulting in high-fidelity compression representation. Currently, INR has been demonstrated to be applicable for compressing natural image scenes 25 as well as various multi-dimensional biomedical data for biological research and clinical diagnosis, including 2D images, 3D volumes 26 , and 4D data. By controlling the network parameters across different samples, stable compression rates and compression quality can be achieved.…”