Market Need
Despite growth in industry demand, IVF cycles currently result in low success rates, with reports indicating less than 20% of cycles result in a live birth. While part of this is attributable to the physical environment of the individual patients, part of the high fail rate of IVF is attributable to the clinical elements of the IVF process, including the quality of embryos selected for transfer to the patient.
Current methods for embryo screening are highly subjective, reliant on the individual, physical assessment by human embryologists.
The Technology
Our team, including third party collaborators, is developing a technology to improve the success of IVF cycles. This technology is intended to be used as a decision support tool to objectively improve the processes currently used by embryologists when assessing the best (and worst) embryos for transfer to female patients.
Utilising non invasive techniques and a suite of unique algorithms, our team is looking at grey level co-occurrence matrices (GLCM), in combination with unique embryo features as a predictor for embyro viability. Existing proof of concept studies have shown a high degree of correlation between the algorithm prediction rates and patient outcome data, with further refinement proposed for the remainder of 2018.
The team benefits from an extensive data set of more than 50,000 images gathered from 3,500 different embryos across a number of treatment cycles (including fresh and frozen embryos, and PGD tested and untested embryos), providing an extensive data set and deep statistical based image analysis.
A set of algorithms have been developed into a prediction model using foetal heart as the predictor of pregnancy potential of the embryos. This model has been validated in a blinded, retrospective study against scoring systems used by trained embryologists, and demonstrated strong predictive potential. Recent studies demonstrate the model's ability to successfully identify the bottom 20% of embryos, (those least suitable for transfer), at an accuracy of above 90%.
Research Team
We have brought together expertise across reproductive biology, clinical embryology, software development and biostatisticians to deliver on this project. This diverse skillset ensures the algorithms developed reflect the biological features of embryo quality, and the technology will nicely complement current clinical practices.
IP Position
The combination of algorithms and embryo characteristics remain trade secret and are not subject to patent protection.
Next Steps
The team is now launching a short term cross validation study, with a view of further refining the existing algorithms. This is expected to be completed by the end of 2018. We are actively looking for collaborators to contribute to future validation studies, as well as funding partners to support scale up and commercialisation of this technology.