Hierarchical Spatio-Channel Clustering for Efficient Model Compression in Medical Image Analysis

2026-04-25

AI Summary

This work addresses the challenge of deploying bulky convolutional neural networks on resource-constrained devices for medical image analysis, where existing low-rank compression methods fail to effectively exploit local spatiotemporal-channel redundancies in feature maps. To overcome this limitation, the authors propose a hierarchical spatiotemporal-channel clustering compression framework that first partitions feature maps spatially and then clusters channels within each region based on co-activation patterns, followed by rank-adaptive singular value decomposition (SVD) applied per cluster. By jointly modeling spatial and channel-wise redundancies, the method transcends the constraints of conventional global uniform compression. Evaluated on brain tumor MRI classification, it achieves an 81.1% reduction in FLOPs (from 8.21G to 1.55G), a 1.38× inference speedup, and an improved accuracy of 89.80%, while substantially enhancing the F₁ score for challenging classes such as meningioma.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Medical and Biological Imaging
• Data Mining & Knowledge Management: Data Compression

Application Category

• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Convolutional neural networks (CNNs) have become increasingly difficult to deploy in resource-constrained environments due to their large memory and computational requirements. Although low-rank compression methods can reduce this burden, most existing approaches compress spatial and channel redundancy independently and therefore do not fully exploit the localised structure within convolutional feature maps. This paper proposes a hierarchical spatio-channel low-rank compression framework for CNNs that exploits redundancy across spatial regions and channel activations. Unlike conventional methods, which apply a uniform decomposition across an entire layer, the proposed approach first partitions feature maps into spatial regions, then groups channels according to their co-activation patterns within each region, and finally applies rank-adaptive SVD to each resulting spatio-channel cluster. The method is evaluated on an AlexNet-based brain tumour MRI classification model and compared with Global SVD and Tucker decomposition under \(3\times\) and \(6\times\) compression budgets. Our method outperforms both baselines, reducing FLOPs from \(8.21\,\mathrm{G}\) to \(1.55\,\mathrm{G}\) (\(81.1\%\) reduction), achieving a \(1.38\times\) inference speed-up, and increasing classification accuracy from \(87.76\%\) to \(89.80\%\). The method also improves the macro \(F_1\)-score and performance on challenging classes such as meningioma. A hyper-parameter trade-off analysis demonstrates that the framework provides Pareto-optimal configurations, enabling control over the balance between compression and predictive performance. Moderate clustering with adaptive rank selection yields strong results. Bootstrap standard errors are reported for all classification metrics.

Problem

• model compression
• spatio-channel redundancy
• convolutional neural networks
• medical image analysis
• low-rank decomposition

Innovation

• hierarchical spatio-channel clustering
• low-rank compression
• adaptive SVD
• CNN model compression
• medical image analysis