Connie Rees

140 worse than the intra-observer variability. This can be further improved by either conducting a more detailed TM positioning protocol for user guidance or investigating automatic segmentation techniques. The ability to quantify and characterize the uterine activity with the proposed method can also aid the diagnosis of uterine dysfunction and diseases. This will be investigated in future dedicated studies. B. Validation in IVF Patients 1) Estimation of UP Velocity, Direction, and Coordination: Significantly higher velocities in both directions were found in the failure group compared to the success group. However, if we compare the velocities in the F2C direction extracted 1 h before ET and during the LF phase from healthy volunteers, the results are very similar. This finding suggest that, even though these patients are suffering from infertility problems and treated with hormones, the UP velocity is similar to that in healthy women in a comparable phase of the menstrual cycle. Our findings also suggest UP direction to be a poor predictor of the success of embryo implantation. This result may be ascribed to the complex propagation patterns observed before embryo implantation in both the successful and unsuccessful pregnancy groups. Significant differences in UP coordination were found between the success and failure groups from all metrics. Well-coordinated UP (higher CC and lower Hd and MSE) was found in the success group. This result suggests that proper coordination of uterine muscle contraction may be a relevant factor determining successful embryo implantation. C. Study Limitations This feasibility study was carried out with a limited dataset. To confirm the predictive value of the proposed features, larger datasets should be acquired and investigated in the future. The analysis of larger datasets would then benefit from automatic segmentation of the uterus, reported, e.g., in [31], enabling fully automatic positioning of the TMs rather than the semiautomatic approach employed in this study.

RkJQdWJsaXNoZXIy MTk4NDMw