OmniMamba4D: Spatio-temporal Mamba for longitudinal CT lesion segmentation
This work addresses the need for accurate segmentation in longitudinal CT scans to monitor tumor progression and evaluate treatment responses, representing an incremental advancement by extending 3D models to 4D data.
The paper tackles the problem of segmenting longitudinal CT scans for monitoring tumor progression by proposing OmniMamba4D, a novel model that processes 4D medical images to capture spatio-temporal features, achieving a Dice score of 0.682 on an internal dataset of 3,252 CT scans, which is competitive with state-of-the-art models.
Accurate segmentation of longitudinal CT scans is important for monitoring tumor progression and evaluating treatment responses. However, existing 3D segmentation models solely focus on spatial information. To address this gap, we propose OmniMamba4D, a novel segmentation model designed for 4D medical images (3D images over time). OmniMamba4D utilizes a spatio-temporal tetra-orientated Mamba block to effectively capture both spatial and temporal features. Unlike traditional 3D models, which analyze single-time points, OmniMamba4D processes 4D CT data, providing comprehensive spatio-temporal information on lesion progression. Evaluated on an internal dataset comprising of 3,252 CT scans, OmniMamba4D achieves a competitive Dice score of 0.682, comparable to state-of-the-arts (SOTA) models, while maintaining computational efficiency and better detecting disappeared lesions. This work demonstrates a new framework to leverage spatio-temporal information for longitudinal CT lesion segmentation.