Deep learning pipeline reveals key moments in human embryonic development predictive of live birth in IVF
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (Arcadia Science)
Abstract
Demand for IVF treatment is growing, however success rates remain low partly due to difficulty in selecting the best embryo to be transferred. Current manual assessments are subjective and may not take advantage of the most informative moments in embryo development. Here, we apply convolutional neural networks (CNNs) to identify key windows in preimplantation human development that can be linked to embryo viability and are therefore suitable for the early grading of IVF embryos. We show how machine learning models trained at these developmental time-points can be used to refine overall embryo viability assessment. Exploiting the well-known capabilities of transfer learning, we illustrate the performance of CNN models for very limited data sets, paving the way for the use on a clinic-by-clinic basis, catering for local data heterogeneity.
Article activity feed
-
-
y.
I found this study really exciting, making me hopeful for the future - both as a patient and a scientist. I think your work could lead to really interesting and fundamental advances in understanding some of the key early steps of this developmental process. My curiosity is particularly piqued with regard to the peak performance time point! Thanks for your work.
-
raining set size was increased beyond around 200-400
I'm curious if you think that training set size would need to be increased if you pool together datasets from different clinics, introducing more noise? Perhaps this has already been evaluated elsewhere, but since I'm from outside your field, it made me wonder how this compares to those multiple-clinic ML studies. Thanks!
-
We found that the model performance varied at different time-points and appeared to peak at certain moments in development as shown in graphs 3C-G
I found this observation fascinating. It would be so interesting to follow-up on these peak prediction windows to understand what else is happening in the cell that may be deterministic for cell fate. Do you have any plans to follow up on these time points?
-
shown in Fig. 2B.
Hi there, this is super interesting and promising. Quick note - it would be helpful to have the axes in Figure 2B labeled so that I can better understand the trend.
-