Motion Correction of 3D Dynamic Contrast-Enhanced Ultrasound Imaging without Anatomical Bmode Images

TL;DR

This paper presents a novel motion correction algorithm for 3D DCE-US imaging that does not rely on anatomical Bmode images, demonstrating significant improvements in image stability and quantitative analysis in clinical liver data.

Contribution

The work introduces a pyramidal motion correction method for 3D DCE-US that operates without Bmode images, enhancing motion correction capabilities in clinical settings.

Findings

Significant increase in lesion overlap from 68% to 83%.

Improved similarity metrics (NCC) from 0.694 to 0.862.

Enhanced time-intensity curve fitting with decreased RMSE and increased R2.

Abstract

In conventional 2D DCE-US, motion correction algorithms take advantage of accompanying side-by-side anatomical Bmode images that contain time-stable features. However, current commercial models of 3D DCE-US do not provide side-by-side Bmode images, which makes motion correction challenging. This work introduces a novel motion correction (MC) algorithm for 3D DCE-US and assesses its efficacy when handling clinical data sets. In brief, the algorithm uses a pyramidal approach whereby short temporal windows consisting of 3-6 consecutive frames are created to perform local registrations, which are then registered to a master reference derived from a weighted average of all frames. We evaluated the algorithm in 8 patients with metastatic lesions in the liver using the Philips X6-1 matrix transducer at a frame rate of 1-3 Hz. We assessed improvements in original vs. motion corrected 3D DCE-US cine using: i) frame-to-frame volumetric overlap of segmented lesions, ii) normalized correlation coefficient (NCC) between frames (similarity analysis), and iii) sum of squared errors (SSE), root-mean-squared error (RMSE), and r-squared (R2) quality-of-fit from fitted time-intensity curves (TIC) extracted from a segmented lesion. Overall, results demonstrate significant decreases in 3D DCE-US motion after applying the proposed algorithm. We noted significant improvements in frame-to-frame lesion overlap across all patients, from 68% without correction to 83% with motion correction (p = 0.023). Frame-to-frame similarity as assessed by NCC also significantly improved on two different sets of time points from 0.694 (original cine) to 0.862 (corresponding MC cine) and 0.723 to 0.886. TIC analysis displayed a significant decrease in RMSE (p = 0.018) and a significant increase in R2 goodness-of-fit (p = 0.029) for the patient cohort.