Gradient-guided multi-task framework with genetic optimization for medical image segmentation and classification

doi:10.1016/j.patcog.2026.113276

Gradient-guided multi-task framework with genetic optimization for medical image segmentation and classification

2026-02-10

0

PRE

AI

DOI：10.1016/j.patcog.2026.113276

Original

Original request for help

Save

Abstract

En 中文

• A two-stage training strategy uses classification gradients to guide segmentation feature selection. • Multi-scale encoder features are integrated using a genetic algorithm for enhanced classification. • Grad-CAM heatmaps identify discriminative regions that improve segmentation accuracy. • Outperforms state-of-the-art methods in classification and segmentation tasks on ultrasound and magnetic resonance imaging datasets.

Keywords:

Gradient-guided training

Multi-task framework

Genetic algorithm

Medical image segmentation

Classification accuracy

Gradient-guided multi-task framework with genetic optimization for medical image segmentation and classification

Abstract

Journal

Organization