Jalilian H, Afrakhteh S, Iacca G, Demi L. Increasing frame rate of echocardiography based on a novel 2D spatio-temporal meshless interpolation.
ULTRASONICS 2023;
131:106953. [PMID:
36805795 DOI:
10.1016/j.ultras.2023.106953]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Received: 11/25/2022] [Revised: 02/03/2023] [Accepted: 02/09/2023] [Indexed: 06/18/2023]
Abstract
BACKGROUND
Increasing temporal resolution through numerical methods aids clinicians to evaluate fast moving structures of the heart with more confidence.
METHODOLOGY
In this study, a spatio-temporal numerical method is proposed to increase the frame rate based on two-dimensional (2D) interpolation. More specifically, we propose a novel intensity variation time surface (IVTS) strategy to incorporate both temporal and spatial information in the reconstruction. In this regard, we exploit radial basis functions (RBFs) for 2D interpolation. The reason for choosing RBFs for this task is manifold. First, RBFs are able to interpolate on large-scale datasets. Moreover, their mathematical implementation is simple. Another important property of this interpolation technique, which is addressed in this study, is its meshless nature. The meshless property enables higher up-sampling (UpS) rates for echocardiography to improve temporal resolution without noticeably degrading image quality. To evaluate the proposed approach, we tested the RBF interpolation on 2D/3D echocardiography datasets. The reconstructed frames were analyzed using different image quality metrics, and the results were compared with two popular techniques from the literature.
RESULTS
The findings demonstrated that, with a down-sampling rate of 3, the proposed technique outperformed the best existing method by 42%, 87%, 8%, and 11%, respectively, in terms of mean square error (MSE), contrast to noise ratio (CNR), peak signal-to-noise ratio (PSNR), and figure of merit (FOM). It should be noted that the proposed method is comparable to the best available method in terms of structural similarity (SSIM) index. Furthermore, when compared to the original images, the results of employing our technique on radio-frequency (RF) level analysis demonstrated that the reconstruction accuracy is satisfactory in terms of image quality criterion.
CONCLUSION
Finally, it is worthwhile noting that the proposed method is better than (or comparable to) the other methods in terms of reconstruction performance and processing time. Therefore, the RBF interpolation can be a promising alternative to the existing methods.
Collapse