LoSA-Net: A Localized and Scale-Adaptive Network for Boundary-Sensitive Prediction of Perineural Invasion in 3D MRI

2026-07-12

AI Summary

This study addresses the challenge of predicting perineural invasion (PNI) in 3D MRI, where conventional downsampling and global feature aggregation often lead to the loss of fine neural structural details. To overcome this limitation, the authors propose a localized, scale-adaptive 3D network architecture that preserves nerve-aligned features through Talking Neighborhood Attention, dynamically adjusts receptive fields via Scale-Adaptive Feature Mixing, and maintains cross-scale consistency between multi-stage semantics and high-resolution boundaries using a Cross-Scale Refinement and Alignment mechanism. Evaluated on contrast-enhanced MRI data from 168 patients with cholangiocarcinoma, the model achieves an AUC of 0.7567, significantly outperforming existing convolutional and Transformer-based baselines, thereby demonstrating its effectiveness and innovation in boundary-sensitive PNI prediction.

Abstract

Perineural invasion (PNI) is a clinically relevant indicator of tumor aggressiveness and can influence surgical decision-making, motivating interest in reliable preoperative assessment. The subtle MRI features of PNI, however, often resemble nearby anatomy, complicating noninvasive prediction. These fine perineural cues are easily attenuated by routine downsampling or overly global feature aggregation, reducing the effectiveness of conventional volumetric models. We present LoSA-Net, a localized and scale-adaptive architecture for boundary-sensitive PNI prediction in 3D MRI. Talking Neighborhood Attention (TNA) preserves nerve-aligned detail through localized self-attention with head-wise mixing, and Scale-Adaptive Feature Mixing (SAFM) modulates the receptive field using multi-scale depthwise processing. Cross-Scale Refinement and Alignment (CSRA) maintains consistency between semantic context and high-resolution boundaries across stages. In contrast-enhanced MRI scans from 168 patients with cholangiocarcinoma, LoSA-Net achieves an AUC of 0.7567 and outperforms representative convolutional and transformer baselines under matched preprocessing and optimization settings.

Problem

• Perineural Invasion
• 3D MRI
• Boundary-Sensitive Prediction
• Localized Features
• Scale Adaptation

Innovation

• Localized Attention
• Scale-Adaptive Processing
• Boundary-Sensitive Prediction
• Perineural Invasion
• 3D MRI