WeaveSeg: Iterative Contrast-weaving and Spectral Feature-refining for Nuclei Instance Segmentation

Automatic and accurate nuclei instance segmentation in histopathology images is a fundamental task in computational pathology. It is also a very challenging task due to complex nuclei morphologies, ambiguous boundaries, and staining variations. Existing methods often struggle to precisely delineate overlapping nuclei and handle class imbalance. We introduce WeaveSeg, a novel deep learning framework that synergistically integrates two key innovations: an adaptive spectral refinement (ASR) module to enhance high-frequency boundary details often blurred by standard convolutions, and an iterative contrast-weaving (ICW) module. Guided by a specialized uncertainty-aware loss, this module leverages contrastive attention and a novel local self-adaptive mechanism to progressively resolve ambiguous instances. Extensive experiments on MoNuSeg, CoNSeP, and CPM17 demonstrate WeaveSeg's SOTA performance over existing models. The code is available at https://github.com/jj-sterne/WeaveSeg.