Cross-attention learning enables real-time nonuniform rotational distortion correction in OCT

TL;DR

This paper introduces a cross-attention learning approach for real-time correction of nonuniform rotational distortion in OCT imaging, significantly improving speed and accuracy over traditional methods.

Contribution

It presents an end-to-end stacked cross-attention network that models long-range dependencies for fast, accurate NURD correction in OCT, outperforming existing techniques.

Findings

Achieved approximately 3x speedup to real-time processing (26 fps)

Demonstrated superior correction performance on multiple datasets

Outperformed traditional feature-based and CNN-based methods

Abstract

Nonuniform rotational distortion (NURD) correction is vital for endoscopic optical coherence tomography (OCT) imaging and its functional extensions, such as angiography and elastography. Current NURD correction methods require time-consuming feature tracking or cross-correlation calculations and thus sacrifice temporal resolution. Here we propose a cross-attention learning method for the NURD correction in OCT. Our method is inspired by the recent success of the self-attention mechanism in natural language processing and computer vision. By leveraging its ability to model long-range dependencies, we can directly obtain the correlation between OCT A-lines at any distance, thus accelerating the NURD correction. We develop an end-to-end stacked cross-attention network and design three types of optimization constraints. We compare our method with two traditional feature-based methods and a CNN-based method, on two publicly-available endoscopic OCT datasets and a private dataset collected on our home-built endoscopic OCT system. Our method achieved a $\sim3\times$ speedup to real time ($26\pm 3$ fps), and superior correction performance.