Increasing the Frame Rate of Echocardiography Images Based on a Novel Interpolation Technique
Sajjad Afrakhteh, Hamed Jalilian, Giovanni Iacca, Libertario Demi · 2022 IEEE International Ultrasonics Symposium (IUS) · 2022
High-temporal cardiac imaging allows for more accurate diagnostics for heart diseases. The objective of the present work is to propose a technique to improve the temporal resolution of echocardiography images. More specifically, we use a new non-polynomial interpolation method to increase echocardiographic image temporal resolution without significantly (i.e., less than 5%) reducing the spatial resolution or size of the reconstructed images. The proposed method involves two steps. In the first step, analytically derived interpolation functions are constructed according to the problem conditions. In the second step, the unknown pixel intensity variation time curves (IVTC) values of the unknown frames are approximated by the functions created in the first step, to increase the temporal resolution. The method presented has been tested on a clinical B-mode echocardiographic dataset consisting of 114 frames with a size of 411 by 481 pixels. The quantitative results show that our proposed interpolation approach is effective, robust, and able to reconstruct images at different times of cardiac cycles. For example, the mean squared error is reduced from 0.6 to 0.3 compared to methods based on polynomials (e.g., cubic spline and B-spline). Although our method has a similar temporal complexity to the cubic spline, the time processing analysis for different down-sampling ratios shows that not only our method is more accurate in terms of quantitative and qualitative results, but it also does not increase the computational time.