This dissertation reports on an investigation into relationships between human embryo, oocyte and follicular features and the developmental potential of the associated IVF embryos on transfer. An improved understanding could yield insights into biological factors influencing the development process. Besides, identifying the features associated with a good developmental potential could assist embryologists in an informed selection of the 'best' embryos for transfer. Previous studies with the same broad goal have suffered from a number of limitations. The present study attempts to overcome some of these by using the technique of 'decision tree' analysis developed in Machine Learning in Computer Science. The technique analyses data related to a sufficient number of specific examples in a group and discovers the general combinations of features characterising the group, presenting results in an easily comprehensible form. The feasibility of using the technique was first explored in a pilot study. The patterns of features characterising the 'good' embryos that resulted in triplet pregnancies on 3 embryo transfer or high hCG secretion in vitro and the 'bad' embryos that did not progress to the blastocyst stage on in vitro culture were identified. Given the promising nature of the results, a larger study, which forms the core of the dissertation, was conducted. Data included 53 features relating to the follicle, oocyte and embryo for batches of 3 embryos transferred that resulted in either a Take home baby (TH) outcome or a negative pregnancy test (NTH). The relationship between the 53 features and the 'TH' or 'NTH' developmental potential was investigated. The results identify, amongst others, two features related to the follicular stage of development as being important determinants of the TH potential, thereby confirming that developmental potential of an embryo is determined as early as the stage of the unfertilised oocyte. The embryo grading used by embryologists in the Oxford IVF Unit was identified as important, providing for the first time an unbiased corroboration of the selection process used in the Unit. The results refine the selection criteria and suggest two additional features, namely, follicular fluid volume and rate of cleavage as being worthy of further investigation, with the latter feature being indicated as the main component of the embryo grade characterising batches that resulted in a TH outcome. The relative unimportance of a number of features recorded in the patient notes but showing no relationship to development too is identified. The method of analysis used here could have wide-ranging applicability to a number of areas within IVF. A demonstration of this is provided in this dissertation in an area of growing importance, namely, the identification of blastocysts of different quality. The main study reported here can be extended to incorporate maternal and sperm features. Results obtained could form the basis for assisting embryologists in the selection of embryos with the 'best' developmental potential, with the possibility of an increase in IVF success rates.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:339285 |
| Date | January 1996 |
| Creators | Saith, Ruhi |
| Publisher | University of Oxford |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://ora.ox.ac.uk/objects/uuid:3c970960-6f8e-4c03-9e42-9282ba00f189 |
Page generated in 0.0017 seconds