Deep learning for clinical decision support in in vitro fertilisation, IVF

Primary supervisor

Hamid Rezatofighi

Co-supervisors

  • Dr. Fabrizzio Horta, Department of Obstetrics and Gynaecology, Monash University, Melbourne, Australia.

In vitro fertilisation (IVF) is a process of fertilisation where an egg is combined with sperm outside the female body, in vitro ("in glass"). The process involves monitoring and stimulating a person's ovulatory process, removing an ovum or ova (egg or eggs) from their ovaries and letting sperm fertilise them in a culture medium in a laboratory. After the fertilised egg undergoes embryo culture for 2–6 days, it is implanted in the same or another person's uterus, with the intention of establishing a successful pregnancy [1].

Background
Since the first report of successful clinical pregnancy with the use of IVF, several patients
characteristics such as female/male age [2], type of infertility, body mass index (BMI) [3] as well as embryology characteristics like embryo quality are known to play a critical role in clinical outcomes and patient management in IVF [4]. For instance, embryo quality grading systems have been developed based on embryo morphology showing significantly improved clinical outcomes [5]. However, their accuracy remains limited and does not consider individual clinical characteristics of each couple, with no improvements in methodologies applied to date. Indeed, in women younger than 35 years old, 60% of fresh embryos transferred do not result in clinical pregnancy. This failure even rises with increased female age to 75% for women 41–42 years old, and to 85-90% for those who are 43–44 years old [6, 7]. These outcomes are consistent with the fact that normal embryos, as assessed morphologically, fail to implant because of unknown reasons that could be explained due to a lack of integration between clinical characteristics and clinical embryology features; on the other hand, morphologically abnormal embryos could still
result in healthy live births [7]. For these reasons, developing reliable methodologies that integrate clinical characteristics as well as clinical embryology features to assist the clinical decisions of clinicians as well as scientists through Artificial Intelligence (AI) could revolutionise the management of patients undergoing assisted reproductive technologies (ART).

Clinical Decision Support Systems (CDSS) are becoming an increasingly common tool for helping clinicians make better care decisions for their patients in many areas of medicine. At present, significant development has seen value in radiology and pathology disciplines, with some groups developing CDSS for physician problems as well as AI technology becoming more advanced. Indeed, it is anticipated that future medical practitioners will be working alongside these systems in every specialty [8]. However, in the field of IVF that future is under development and heavily reliant only on digital information from images and time-lapse videos of embryos [9] without integrating clinical information that identifies real treatment planning and prognosis that can potentially improve the chances of achieving pregnancy.

This project aims to use the latest developments in AI, machine learning and deep learning in the field of IVF by testing and implementing a more achievable technology that is cost and time-efficient. The proposed system will calculate the probability of pregnancy for each embryo after analysing the image of the embryo whilst considering confounding clinical factors.

    Student cohort

    Double Semester

    Aim/outline

    • Help the team in creating a database of cycles of IVF patients including their clinical features and digital images
    • Design and test different neural network architectures based on (convolutional neural network, multi-layer perception and/or transformers) to input and fuse embryo images with their associated clinical data to reliably estimate the probability of pregnancy for each sample. 
    • Comprehensive experiments by evaluating and comparing the designed models on the collected datasets

    URLs/references

    References

    1. https://en.wikipedia.org/wiki/In_vitro_fertilisation
    2. Horta F, Vollenhoven B, Healey M, Busija L, Catt S, Temple-Smith P. Male ageing is negatively associated with the chance of live birth in IVF/ICSI cycles for idiopathic infertility. Human Reproduction. 2019;34(12):2523-32.
    3. Wittemer C, Ohl J, Bailly M, Bettahar-Lebugle K, Nisand I. Does body mass index of infertile women have an impact on IVF procedure and outcome? Journal of assisted reproduction and genetics. 2000;17(10):547-52.
    4. Fauque P, Léandri R, Merlet F, Juillard J-C, Epelboin S, Guibert J, et al. Pregnancy outcome and live birth after IVF and ICSI according to embryo quality. Journal of assisted reproduction and genetics. 2007;24(5):159-65.
    5. Gardner DK, Lane M, Stevens J, Schlenker T, Schoolcraft WB. Blastocyst score affects implantation and pregnancy outcome: towards a single blastocyst transfer. Fertility and sterility. 2000;73(6):1155-8.
    6. Sanchez T, Venturas M, Aghvami SA, Yang X, Fraden S, Sakkas D, et al. Combined noninvasive metabolic and spindle imaging as potential tools for embryo and oocyte assessment. Human Reproduction. 2019;34(12):2349-61.
    7. Sanchez T, Zhang M, Needleman D, Seli E. Metabolic imaging via fluorescence lifetime imaging microscopy for egg and embryo assessment. Fertility and sterility. 2019;111(2):212-8.
    8. Briganti G, Le Moine O. Artificial Intelligence in Medicine: Today and Tomorrow. Frontiers in Medicine. 2020;7(27):1-6.
    9. Khosravi P, Kazemi E, Zhan Q, Malmsten JE, Toschi M, Zisimopoulos P, et al. Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization. NPJ digital medicine. 2019;2(1):1-9.

    Required knowledge

    1. Good coding skills in a variety of coding languages
    2. Previous experience working with deep learning models for different tasks
    3. ​​​​​Proficient programming skills in Python and one of the main deep learning libraries (e.g., TensorFlow, PyTorch, Keras)