MAUNext: a lightweight segmentation network for medical images.

Area of Science:

• Medical Imaging
• Computer Vision
• Artificial Intelligence

Background:

• Automated medical image segmentation is crucial for clinical diagnosis and treatment.
• Manual segmentation is time-consuming and resource-intensive.
• Enhancing accuracy and efficiency in segmentation is a key research objective.

Purpose of the Study:

• To develop an advanced medical image segmentation technique prioritizing accuracy and parameter efficiency.
• To introduce the novel MAUNext approach for improved clinical research applications.

Main Methods:

• A novel codec-based MAUNext approach was devised.
• Key components include a lightweight backbone, multiscale attention, and collaborative neighborhood-attention MLP.
• Three core modules were integrated: multi-scale attentional convolution, collaborative neighborhood-attention MLP encoding, and skip-connected cross-layer semantic fusion.

Main Results:

• MAUNext was evaluated against eight state-of-the-art methods on Kagglelung, ISIC, and Brain datasets.
• The proposed approach demonstrated superior performance in both accuracy and parameter count.
• Experimental outcomes confirm MAUNext's effectiveness in medical image segmentation tasks.

Conclusions:

• The MAUNext approach offers substantial improvements in accuracy and efficiency for medical image segmentation.
• This innovation holds significant promise for aiding clinical decision-making and patient treatment.
• The study highlights the clinical value of advanced automated segmentation solutions.