Non-invasive evaluation of embryos using mid-infrared attenuated total reflection spectrometry of incubation medium: a preliminary study

Research question: Can mid-infrared attenuated total reflection (MIR ATR) spectroscopy combined with machine learning methods be used as an additional tool to predict embryo quality and IVF treatment outcomes?Design: Spent culture media was collected and analysed. MIR ATR absorbance spectra were measured using an ALPHA II spectrometer equipped with an attenuated total reflection (ATR) spectrometry accessory. Patient and treatment data and results were collected and analysed in combination with machine learning techniques to identify possible correlations. The main outcome measures were to define the characteristics of absorbance spectra of spent culture media and to distinguish the difference in absorbance between top-and low-quality embryos, day 3 and day 5 embryos and implanting embryos versus non-implanting embryos.Results: Spent culture media of 227 embryos was collected and analysed. Absorbance peaks in the culture media were different between day 3 and day 5 embryos. Moreover, significant differences in P-values, spanning from 0.014 to 0.044 in absorbance peaks for day 3 embryos and 0.024 up to 0.04 for day 5 embryos, were seen between implanting and non-implanting embryos. Machine learning techniques offered a pregnancy prediction value of 84.6% for day 3 embryos.Conclusions: MIR ATR may offer an additional parameter for better selection of embryos based on the spectrometric absorbance and secretions of metabolites in the culture media.

Automated summary

This study aims to investigate whether mid-infrared attenuated total reflection (MIR ATR) spectroscopy combined with machine learning methods can be used as an additional tool to predict embryo quality and IVF treatment outcomes. The results showed that MIR ATR technology can better select embryos based on the characteristics of absorbance peaks in the culture media and the differences in metabolite secretions. Machine learning techniques also offered a high pregnancy prediction value for day 3 embryos.