Blood Estradiol level from transvaginal ultrasounds
Keeping blood Estradiol level under control is crucial during fertility treatments in order to avoid risk of overstimulation. To this end blood samples are typically taken every one or two days.
As during fertility treatments Estradiol is mainly produced by the growing follicles, E2E allows physicians to estimate Estradiol level from number and size of the growing follicles as a result of transvaginal ultrasounds.
E2E enables reduction of the number of required blood samples during treatment to check that Estradiol level is below its safety threshold.
This contributes to reduce costs of fertility treatments.
As patients can perform transvaginal ultrasounds by themselves at home with portable devices (see for example Sonaura), E2E allows remote Estradiol estimation, thus enabling reduction of the number of patient visits per treatment.
This also contributes to reduce costs of fertility treatments and to increase patient comfort.
E2E computes reliable individualised Estradiol estimations by taking into account patient external factors (for example, AFC, pathologies, treatment protocol in use).
Accuracy improves after any actual measurement. For most patient groups two measurements suffice to have fully reliable estimations.
Follicles growing in the ovary have the 3-layer structure shown in the figure. They consist in a fluid-filled Antrum, the Granulosa layer and the Theca layer.
During the Stimulation phase of a fertility treatment, Estradiol is mainly produced by the cells of the Granulosa layer of the growing follicles.
The diameter of the follicles, as measured from transvaginal ultrasounds, is the diameter of the Antrum (as granulosa and theca layers are not visible through ultrasounds).
It is known that the tickness of the Granulosa layer is constant during follicle growth (around \(45\mu m\)).
From this follows that the volume of the Granulosa layer can be easily estimated from the diameter of the Antrum using standard geometry arguments, assuming that the follicles have a spherical shape.
Here we show experimental results on the accuracy assessment of the Estradiol estimations performed by E2E both when some actual Estradiol measurements are available for the patient at hand (patient-specific estimations) and when only information of the patient group is known (inter-patient estimations).
Experiments use retrospective clinical data on thousands of patients.
Data have been kindly provided by University Hospital of Zurich and Hannover Medical School (Paeon partners) as well as University Hospital of Lausanne and University Hospital of Basel as a result of networking activities.
The figure shows the accuracy achieved by the patient-specific Estradiol estimation performed by E2E. In particular, for each patient class, the figure shows the average Estradiol estimation error and the standard deviation on the patients in our database, when feeding the system with 2, 3 and 4 actual measurements of Estradiol blood level and number and size of follicles.
It can be seen that 3 actual measurements are enough to make Estradiol patient-specific estimation very accurate.
As expected, for some groups like the group consisting in women affected by PCOS, standard deviation is higher. This is in agreement with the fact that PCOS patients form a heterogeneous group with high inter-patient variability. It is also expected that standard deviation tends to increase as the number of follicles increases.
When no actual Estradiol measurements are available for the patient at hand to tune the parameters, E2E can still estimate Estradiol level using patient group parameters.
Of course, by using group parameters, the achieved estimation accuracy is lower. In particular, as shown in the figure, standard deviation on all groups is higher than that achieved with patient-specific parameters and 3 measurements.
Summing up, experimental results show that patient-specific Estradiol estimation is very accurate when 2 or 3 actual measurements are perfomed. This means that several blood samples can be avoided to measure Estradiol level during a fertility treatment. As Estradiol level is currently measured every one or two days during a fertility treatment, E2E might yield substantial reduction of treatment costs and improvements of patient comfort.
Although results show that group parameters are not enough to accurately estimate Estradiol level in a patient when no actual measurements have been taken, estimating Estradiol by using group parameters is still very useful as a safety check. In fact, a patient under treatment whose measured Estradiol level is too different from the expected behaviour (for that patient group) can be a symptom that something wrong is happening during the treatment. This can be an early warning that deeper investigations are needed.