142 have been introduced in this study. In the future, these features can be integrated into a machine learning framework with additional standard features, such as contraction frequency and amplitude, to further improve the prediction of successful embryo implantation [18]. With this approach, clinicians could be supported with critical decision-making, such as whether to proceed with the ET or to freeze the embryo and wait for more favourable UP characteristics. In this way, increased IVF success rates may be possibly achieved. To test this hypothesis, dedicated clinical trials should be performed where predictive modelling and subsequent decisions are integrated into the clinical workflow. The mechanical activity of the uterus is not the only determinant of successful IVF. Extensive research has focused on the assessment of embryo quality as a predictor of successful embryo implantation [35]. Also, in this context, predictive models based on machine learning are being developed and investigated [36]. In addition, it is worth investigating the relationship between patient responses to hormone injection and related changes in UP characteristics. The combination of features reflecting embryo quality, hormone response, and UP characteristics can be envisaged to improve the prediction of IVF success throughout a machine learning model. In this work, 2-D TVUS recordings were all acquired in the sagittal plane, and the main focus was on measuring the uterine contraction in the radial direction. However, with the advent of 3-D US options, complex UP patterns and OOP motion can be analysed and elucidated accounting for all spatial dimensions. This will also contribute to improve our knowledge of the uterine dynamics by exploring its longitudinal and circumferential deformation, opening up new possibilities for characterising UP and understanding the underlying physiological processes. Although this study focussed on IVF, UP assessment and characterization can represent a valuable diagnostic tool also in the context of widespread pathological conditions of the uterus, such as adenomyosis and endometriosis, which may be reflected in altered UP patterns. Dedicated clinical trials can be designed in the future to investigate the potential of the proposed features for the diagnosis of uterine diseases and dysfunctions. Conclusion A new method for quantitative analysis and characterization of UP based on ultrasound speckle tracking is proposed. Features related to the velocity,
RkJQdWJsaXNoZXIy MTk4NDMw