In Vitro Fertilization to In Virtual Fertilization Notes
In Vitro Fertilisation (IVF)
Overview of the IVF process:
Ovarian hyperstimulation.
Transvaginal oocyte retrieval.
Sperm preparation.
Sperm and egg incubation.
Embryo culture.
Embryo transfer.
Pregnancy.
Repeat step 6 if necessary.
Embryology Lab
Insemination Techniques:
ICSI (Intracytoplasmic Sperm Injection): The best sperm cell is selected by the embryologist and injected into the egg using a needle.
IVF: Sperm cells are added to the egg, and one sperm enters the egg on its own.
Embryo Development
Day 1: Fertilisation status is checked.
Days 2-5: The embryo grows and divides.
Day 4: The embryo compacts and contains 16-32 cells.
Day 5/6/7: Blastocyst stage.
Not every embryo reaches the blastocyst stage.
Aneuploidies can prevent growth (insufficient genetic information).
Preimplantation Genetic Testing and Trophectoderm Biopsy
Embryo Grading - The Gardner Grading
Performed at the blastocyst stage.
ICM (inner cell mass) and TE (trophectoderm) are graded based on size and compaction.
Grading scale:
6AA being the best.
1CC being poor.
Categories include: 6AA, 6AB, 6BB, 6BC, 6CC, 5AA, 5AB, 5BB, 5BC, 4AA, 4CC, 3AA, 2AA, 2AB, 2BB, 2BC, 2CC, 1AA, 1AB, 1BB, 1BC, 1CC
Embryos can be classified as good, best, poor or discard based on the grading.
'In Silico' Fertilisation and Embryo Grading
Can technology improve outcomes compared to humans?
Predict blastocyst formation.
Predict euploid embryos.
Predict implantation/live birth rate.
Time-Lapse Imaging
New annotation system tracks:
Cleavage pattern.
Number of cells after each division.
Blastocyst formation.
Examples of tracked events:
Fertilisation.
PB2 extrusion (2:40).
PN appearance (7:50).
2PN (12:15).
2PB Clear (12:15).
Syngamy (20:40).
1st division (22:40).
2 cell (27:30).
2nd division (35:15).
4 cell (38:05).
3rd division (47:10).
8 cell (62:40
Artificial Intelligence (AI) in IVF
Definitions:
AI: A program that can sense, reason, act, and adapt.
Machine Learning: Algorithms that improve performance with more data exposure.
Deep Learning: A subset of machine learning using multilayered neural networks.
Machine Learning
Training process:
Input raw data.
Algorithm processes data.
Model training occurs.
Output is generated by the trained model.
KID Score
Example of a traditional machine learning algorithm for embryo grading.
Based on:
Number of pronuclei.
Predicts implantation potential.
Deep Learning Models
CHLOE EQ as an example.
Data-Driven Approaches to IVF
Turning uncertainty into data-driven approaches.
Avenues: Integrating Digitalisation + AI at every step of the patient journey.
End-to-End Digitalisation
Components:
Pathology/Genetics labs, Electronic data capture (microscopy, witnessing, sperm/egg images).
Embryologist remote access, Clinical app, Patient app with payment & consent.
AI for oocyte & sperm assessment.
Fully integrated platform, Drugs dispensing, AI robotics, Cryostorage management.
Ultrasound machine (follicular scans, embryo transfer).
KPI (Key Performance Indicator) tracking.
AI-Driven Diagnostics
CASA sperm assessment.
2D & 3D pelvic scan.
Hormone + genetic profiles.
Questionnaire + wearables.
Issues with AI-Based Embryo Selection Tools
Explainable AI (XAI)
Opening the 'black box' of AI to understand how decisions are made.
Provides annotations and explanations for AI outputs.
Sex Bias in Embryo Selection
Conflicting evidence regarding sex differences in developmental timings.
Some studies suggest XX embryos develop slower due to X chromosome inactivation.
Other studies find no differences.
References: Alfarawati et al., 2011, Serdarogullari et al., 2014, Carrasco et al., 2022
Methods for Investigating Sex Differences
N=1411 embryos with time-lapse and known sex information through PGT-A at blastocyst stage.
Three different embryo grading tools were used to compare outcomes between XX and XY embryos:
Manual morphological grading of the blastocyst
Traditional machine learning approaches (KID Score)
Modern deep learning approaches (Chloe EQ)
Results of Sex Differences in Embryo Selection Algorithms
XY embryos received higher grades using manual grading and the KID score, but not using CHLOE EQ.
XY embryos (2.809 ± 0.628, N=669) received higher TE grades than XX embryos (2.656 ± 0.623, N=616) (U=179584, p<0.00001).
No difference observed in aneuploidy rates overall between XX (73/642) and XY (75/692) embryos (=0.0958, p=0.757).
KID Score differences in aneuploid XY vs XX.
Virtual Reality (VR) in Fertilisation
Immersion into a new virtual world using special headsets creating endless surroundings (Moon, island, virtual lecture auditorium, virtual lab).
Getting Started with VR
Create your avatar.
Access the virtual space through a device.
VR in Medicine and Surgery
Process:
Image Acquisition (CT/MRI)
Image Segmentation
Importing the 3D model into the VR environment
Applications of VR in Medicine
Pre-op and post-op visualizations.
Double-Outlet Right Ventricle Repair.
Stent Implantation.
Siamese Twins Separation.
Hosting Events or Live Training in VR
Virtual Clinic
Patients can meet doctors/embryologists remotely in real-time.
Access patient records and embryology images in 3D.
Environment can replicate an actual IVF clinic.
Virtual Reality (VR) Software for ICSI & Embryo Biopsy
Technical training tool.
Mixed Reality (XR)
Physical reality and digital content are combined.
Digital 3D objects are superimposed on top of the ‘real’ reality.
Applications of Mixed Reality
Lab checklists.
Training and SOPs (Standard Operating Procedures).
Patient education.
The Virtual Hospital
Examples: AIMEDIS
Benefits of Virtualization
Virtual clinic.
Virtual patient interaction.
24-h clinic.
Moving fertility care beyond clinic walls.
Saving costs.
Increasing patient satisfaction.