🤖 AI Summary
To address the loss of protein marker specificity and poor rare-cell discrimination caused by early channel fusion in multiplexed tissue imaging data, this work proposes a lightweight self-supervised representation learning paradigm that explicitly preserves channel independence. Methodologically, we design CIM-S—a shallow, channel-independent model with only 5.5K parameters—incorporating contrastive self-supervised pretraining, an explicitly disentangled CNN architecture, customized spatial augmentations, and a linear evaluation protocol to rigorously maintain discriminative representations for each protein channel. Evaluated on 49- and 18-plex CODEX datasets, CIM-S significantly outperforms mainstream early-fusion CNNs and large foundation models, achieving a 12.6% absolute gain in rare-cell classification accuracy. It further demonstrates strong generalizability and high reproducibility across diverse experimental settings. This study provides the first empirical validation of the critical importance of both channel independence and architectural lightness for effective representation learning in multiplexed biological imaging.
📝 Abstract
Multiplexed tissue imaging measures dozens of protein markers per cell, yet most deep learning models still apply early channel fusion, assuming shared structure across markers. We investigate whether preserving marker independence, combined with deliberately shallow architectures, provides a more suitable inductive bias for self-supervised representation learning in multiplex data than increasing model scale. Using a Hodgkin lymphoma CODEX dataset with 145,000 cells and 49 markers, we compare standard early-fusion CNNs with channel-separated architectures, including a marker-aware baseline and our novel shallow Channel-Independent Model (CIM-S) with 5.5K parameters. After contrastive pretraining and linear evaluation, early-fusion models show limited ability to retain marker-specific information and struggle particularly with rare-cell discrimination. Channel-independent architectures, and CIM-S in particular, achieve substantially stronger representations despite their compact size. These findings are consistent across multiple self-supervised frameworks, remain stable across augmentation settings, and are reproducible across both the 49-marker and reduced 18-marker settings. These results show that lightweight, channel-independent architectures can match or surpass deep early-fusion CNNs and foundation models for multiplex representation learning. Code is available at https://github.com/SimonBon/CIM-S.