Convolutional Feature Noise Reduction for 2D Cardiac MR Image Segmentation

2025-11-28

AI Summary

In 2D cardiac MR image segmentation, convolutional features suffer from unmodeled and unmitigated noise, leading to unstable feature propagation. To address this, we propose the Convolutional Feature Filter (CFF), which—firstly—models convolutional features as Gaussian-distributed signal matrices and introduces a low-amplitude pass filtering mechanism to suppress noise; secondly, incorporates an information-entropy-based binarization equation to enable computationally tractable, quantitative assessment of noise levels. CFF requires no additional parameters or architectural modifications and is plug-and-play compatible with mainstream 2D segmentation networks. Experiments on the ACDC and M&Ms public benchmarks demonstrate that CFF significantly reduces feature noise, improves Dice scores by 1.2–2.4%, and enhances model robustness and generalizability. This work establishes a novel paradigm for feature purification in medical image segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Dimensionality Reduction/Feature Selection
• Natural Language Processing: Summarization

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Web data integration and cleaning
• Search and Retrieval-Augmented AI: Assisted, interactive, and conversational search

Abstract

Noise reduction constitutes a crucial operation within Digital Signal Processing. Regrettably, it frequently remains neglected when dealing with the processing of convolutional features in segmentation networks. This oversight could trigger the butterfly effect, impairing the subsequent outcomes within the entire feature system. To complete this void, we consider convolutional features following Gaussian distributions as feature signal matrices and then present a simple and effective feature filter in this study. The proposed filter is fundamentally a low-amplitude pass filter primarily aimed at minimizing noise in feature signal inputs and is named Convolutional Feature Filter (CFF). We conducted experiments on two established 2D segmentation networks and two public cardiac MR image datasets to validate the effectiveness of the CFF, and the experimental findings demonstrated a decrease in noise within the feature signal matrices. To enable a numerical observation and analysis of this reduction, we developed a binarization equation to calculate the information entropy of feature signals.

Problem

• Reduces noise in convolutional features for segmentation
• Introduces a low-amplitude pass filter for feature signals
• Validates filter on cardiac MR image segmentation networks

Innovation

• Low-amplitude pass filter for convolutional features
• Filters noise in feature signal matrices
• Uses binarization equation to calculate entropy