NL-MambaXCT: Self-Supervised Nested-Learning Mamba for Nomex Honeycomb X-ray CT Defect Classification

2026-05-25

AI Summary

This study addresses the challenges of manual interpretation dependency and scarce annotated data in X-ray computed tomography (CT) defect classification of Nomex honeycomb structures in aerospace manufacturing. To this end, the authors propose NL-MambaXCT, a novel framework that integrates the Mamba architecture with nested learning, featuring a dual-timescale fast-slow parameter update mechanism. The model employs a four-stage 2D encoder—comprising RegNet convolutional blocks in early stages and a fusion of Mamba sequence modeling with attention mechanisms in later stages—and leverages masked image modeling for self-supervised pretraining, exponential moving average projection, and a deep momentum optimizer to achieve label-efficient learning. On an independent test set, NL-MambaXCT attains 96.91% accuracy and a 96.8% macro F1-score, significantly outperforming CNN, pure attention-based, and single-timescale Mamba baselines by 3.11–10.31 percentage points, thereby enhancing generalization in small-sample industrial settings.

Abstract

X-ray computed tomography (XCT) is widely used for non-destructive testing of Nomex honeycomb structures in aerospace manufacturing, but industrial inspection still relies heavily on manual interpretation and supervised models trained on limited labeled data. This work introduces NL-MambaXCT, a Mamba-based framework that combines self-supervised masked image modelling with a Nested Learning (NL) formulation for automated, label-efficient defect classification from production XCT slices. The backbone is a four-stage 2D encoder with RegNet convolutional blocks in the early stages and Mamba-based sequence mixing with attention in the deeper stages. It is pretrained by masked image modelling on 19,961 unlabeled industrial XCT slices and fine-tuned on 2,000 relabeled Nomex XCT slices split by production order. NL is instantiated through two-timescale parameter dynamics: selected projections maintain slow exponential-moving-average traces alongside fast weights, while a deep-momentum optimizer introduces an additional slow parameter-update trajectory. On the held-out test set, the MIM-pretrained NL-MambaXCT model achieves 96.91% accuracy and 96.8% macro F1, outperforming CNN, attention, and single-timescale Mamba baselines by 3.11--10.31 percentage points in accuracy. The results suggest that combining masked self-supervision with NL-style fast/ slow learning dynamics is a promising strategy for robust defect classification in Nomex honeycomb XCT inspection.

Problem

• X-ray computed tomography
• Nomex honeycomb
• defect classification
• self-supervised learning
• label-efficient

Innovation

• Mamba
• Nested Learning
• Masked Image Modeling
• Self-Supervised Learning
• X-ray CT Defect Classification