Evaluating iDA on day 5
Research type
Research Study
Full title
eValuating iDA Selection Ability. The VISA study. Will embryo selection through use of artificial intelligence (iDA) perform equally compared to day 5 morphology?
IRAS ID
301947
Contact name
Scott Nelson
Contact email
Sponsor organisation
Vitrolife A/S
Clinicaltrials.gov Identifier
ACTRN12620000197932, Australian New Zealand Clinical Trials Registry
Duration of Study in the UK
0 years, 7 months, 31 days
Research summary
During IVF (in vitro fertilization), embryos are created in the laboratory and cultured for up to five days before transfer. On transfer day, an embryologist assesses the morphologic appearance of the embryos and selects the embryo(s) with the highest likelihood of success for transfer.
Whether morphology assessment is the optimal embryo selection method has been studied extensively over the past twenty years. Recently, iDAScore, a new decision support tool for embryo assessment has been developed. Using deep learning, iDAScore objectively “compares” a given embryo with others having similar development patterns, and generates a score correlating with the likelihood of implantation. This approach differs from previous algorithms in that it is completely based on deep learning, is independent of any assumptions from previous knowledge of embryology standards and does not require any user input.
In the current trial, deep learning and conventional assessment are compared using a randomised controlled setup, where the embryo selection technique is allocated at random. Half of the patients will be randomly allocated to having the embryologist choose the transferable embryo and half to the treatment arm, with embryo selection performed with support from iDAScore. This is an ongoing multi-centre study conducted in IVF clinics in Australia, Denmark, Hungary, and the UK. Over 300 participants have already been randomized. Eligible participants are patients undergoing IVF treatments at the recruiting clinics. The maximum age of the embryo recipient is 42 years. A minimum of two early blastocysts need to be available on transfer day (day 5 of culture) to enable embryo selection. One single blastocyst is transferred either in the fresh cycle or in a frozen cycle. Surplus good quality blastocysts are cryopreserved.
The overall chance of pregnancy will be similar in patients receiving embryos from the treatment group compared to the group not receiving any treatment.Summary of Rsearch Results
During IVF (in vitro fertilisation), embryos are created in the laboratory and cultured for up to five days before embryo transfer. By this time, embryos with a favourable prognosis have developed into blastocysts. On transfer day, an embryologist assesses the embryos’ morphologic, or structural, appearance and selects the one(s) with the highest chance of success for transfer.Whether or not morphological assessment is the optimal method of embryo selection has been studied extensively over the past twenty years. Recently, a new decision support tool for embryo evaluation has been developed. Using deep learning, iDAScore objectively “compares” a given embryo with others having similar development patterns and generates a score correlating with the likelihood of implantation. The main purpose of the investigation was to investigate if embryo selection using iDAScore can generate as many pregnancies as when the selection is performed by trained embryologists using conventional assessment only.
The study population consists of IVF patients seeking treatment at the 14 participating clinics in Australia, Denmark, Sweden and the UK. All patients’ embryos were cultured to day 5 and if at least 2 blastocysts were available the patient was randomly assigned to either the control or study group. In both groups, the highest-ranking embryo, either by morphology (control group) or by iDAScore (study group) was transferred to the patient.
In total, 1066 patients were included. The clinical pregnancy rate was 48.2% (257 of 533 patients) in the control group and 46.5% (248 of 533 patients) in the iDAScore group. The live birth rates were 43.5% (232 of 533 patients) for control and 39.8% (212 of 533 patients) for iDAScore. The iDAScore was 10 times faster in ranking the embryos compared to the manual assessment in the control group (mean: 21.3 seconds versus 208.3 seconds). No adverse events were reported.
This is the world’s first randomised controlled trial comparing the use of a deep-learning-based algorithm for embryo selection with standard visual assessment of embryos by a trained embryologist. The use of iDAScore yielded clinical pregnancy and live birth rates comparable to standard assessment. The significant reduction in evaluation time highlights iDAScore’s potential to enhance efficiency in IVF clinics without compromising clinical outcomes.
REC name
North West - Greater Manchester South Research Ethics Committee
REC reference
21/NW/0223
Date of REC Opinion
30 Sep 2021
REC opinion
Further Information Favourable Opinion