New RadiomicNet architecture enhances medical image segmentation with interpretable AI

Researchers have developed RadiomicNet, a novel deep learning architecture for medical image segmentation that integrates handcrafted radiomics features to enhance interpretability and reduce computational requirements. This hybrid approach uses a Radiomics Attention Gate (RAG) with Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP) features to guide attention in a lightweight MobileNetV2 encoder-decoder. RadiomicNet achieved competitive performance on the Breast Ultrasound Images (BUSI) dataset and Kvasir-SEG, while using significantly fewer parameters than standard U-Net and U-KAN models. AI

IMPACT This research could lead to more efficient and understandable AI tools for medical image analysis, potentially improving diagnostic accuracy and reducing computational costs.

RANK_REASON The item is a research paper detailing a new AI architecture for medical image segmentation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Breast Ultrasound Images dataset
• Gray Level Co-Occurrence Matrix Texture Analysis of Germinal Center Light Zone Lymphocyte Nuclei: Physiology Viewpoint with Focus on Apoptosis
• Kvasir-SEG
• Local binary patterns
• MobileNetV2
• Mohammad Amanour Rahman
• RadiomicNet
• U-Net
• Wilcoxon signed-rank test

• paper
• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →