Combinatorial deep learning improves medical image analysis accuracy
Combinatorial Network-Based Manifold Topological Deep Learning for Image Analysis
Computer Vision and Pattern Recognition
Summary
Medical images are complex because they have shapes and structures that are hard for normal AI models to understand. The authors developed a new deep learning method called CNMTDL that treats medical images like geometric shapes and breaks them down into parts to better capture their structure. This helps the AI better analyze the images by paying attention to details between different geometric components. They tested their method on several medical image datasets and showed it works well for both 2D and 3D images.
What this means in practice
- •For medical imaging teams: Improve diagnostic accuracy by integrating CNMTDL to analyze complex geometric structures in medical scans.
- •For computer vision developers: Develop new image analysis tools that harness discrete manifold structures for better feature extraction in medical datasets.
Authors
Alice Wachira, Xiang Liu, Zhe Su, Yiying Tong, Ge Wang, Guo-Wei Wei
Abstract
Medical image analysis remains fundamentally challenging because of the intricate geometric and topological structures present in medical data. Conventional convolutional neural networks model images as regular Euclidean grids, limiting their ability to preserve geometric relationships and higher-order structural information. Recently, manifold topological deep learning (MTDL) has emerged as a promising paradigm that integrates deep learning with geometric and topological representations. Nevertheless, existing methods have not yet fully exploited discrete manifold structures within combinatorial complex neural networks. To bridge this gap, we introduce CNMTDL, a MTDL framework that integrates Hodge decomposition with a combinatorial attention mechanism. In our approach, medical images are represented as discrete manifolds and decomposed into three Hodge components. Features extracted from these components are concatenated and embedded into a combinatorial complex architecture, enabling enhanced higher-order message passing between $0$-cells and $2$-cells through attention-based blocks. We evaluate CNMTDL on six two-dimensional and three-dimensional datasets from the MedMNIST v2 benchmark, demonstrating its effectiveness for medical image analysis.