MorphSeek: Fine-grained Latent Representation-Level Policy Optimization for Deformable Image Registration
This work addresses the problem of deformable image registration for medical image analysis, offering a scalable and data-efficient solution, though it appears incremental as it builds on existing reinforcement learning frameworks with novel refinements.
The paper tackles the challenge of high-dimensional deformation space and limited supervision in deformable image registration by proposing MorphSeek, a fine-grained representation-level policy optimization paradigm that reformulates it as a spatially continuous process in latent feature space, achieving consistent Dice improvements across three 3D medical image benchmarks.
Deformable image registration (DIR) remains a fundamental yet challenging problem in medical image analysis, largely due to the prohibitively high-dimensional deformation space of dense displacement fields and the scarcity of voxel-level supervision. Existing reinforcement learning frameworks often project this space into coarse, low-dimensional representations, limiting their ability to capture spatially variant deformations. We propose MorphSeek, a fine-grained representation-level policy optimization paradigm that reformulates DIR as a spatially continuous optimization process in the latent feature space. MorphSeek introduces a stochastic Gaussian policy head atop the encoder to model a distribution over latent features, facilitating efficient exploration and coarse-to-fine refinement. The framework integrates unsupervised warm-up with weakly supervised fine-tuning through Group Relative Policy Optimization, where multi-trajectory sampling stabilizes training and improves label efficiency. Across three 3D registration benchmarks (OASIS brain MRI, LiTS liver CT, and Abdomen MR-CT), MorphSeek achieves consistent Dice improvements over competitive baselines while maintaining high label efficiency with minimal parameter cost and low step-level latency overhead. Beyond optimizer specifics, MorphSeek advances a representation-level policy learning paradigm that achieves spatially coherent and data-efficient deformation optimization, offering a principled, backbone-agnostic, and optimizer-agnostic solution for scalable visual alignment in high-dimensional settings.