If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To develop an automated sperm injection robot (ICSIA).
Design
The ICSIA robot automated the sperm injection procedure, including injection pipette advancement, zona pellucida and oolemma penetration with piezo pulses, and pipette removal after sperm release. The robot was first tested in mouse, hamster and rabbit oocytes, and subsequently using discarded human oocytes that were injected with microbeads. Finally, we performed a small clinical pilot trial with donor oocytes to study the feasibility of the robot in a clinical setting. The ICSIA robot was controlled by engineers with no micromanipulation experience and results compared to those obtained with manual ICSI performed by experienced embryologists.
Results
The ICSIA robot demonstrated similar results as the manual procedure in the different animal models tested as well as in the pre-clinical validations performed in discarded human oocytes. In the clinical validation, 13/14 oocytes injected with ICSIA (93%) fertilized correctly and 8 (61.5%) developed into good quality blastocysts, and 4 (50%) were diagnosed as chromosomally normal. In the manual control, 16/18 (88.8%) oocytes injected fertilized correctly, 12 (75%) reached the blastocyst stage and 10 (83%) were euploid. Three euploid blastocysts from the ICSIA robot group have been transferred into two recipients, which resulted in two babies born.
Conclusions
The ICSIA robot showed high proficiency in injecting animal and human oocytes when operated by inexperienced personnel. The preliminary results obtained in this first clinical pilot trial are within key performance indicators. This is the first time that babies are born using an automated sperm injection system.
We developed an ICSI (ICSIA) robot to automate the sperm injection procedure. ICSIA was first tested in mouse, hamster and rabbit oocytes, and subsequently using discarded human oocytes. Next, we conducted a clinical pilot trial with donor oocytes resulting in the first baby born from automated ICSI.
INTRODUCTION
Intracytoplasmic sperm injection (ICSI) is a clinical procedure that is currently performed worldwide in most in vitro fertilization (IVF) centers. The technique was introduced in human assisted reproductive technology (ART) about thirty years ago to overcome severe male factor infertility (Palermo et al., 1992) . However, about 70% of the total IVF cycles performed currently use ICSI as the preferred method for oocyte insemination, regardless of the sperm parameters (Esteves, 2021). Since its implementation, the procedure has been conducted manually by highly skilled embryologists and, although it has shown to be reproducible in most laboratories, its global success rates can vary significantly depending on the experience and technical expertise of the operator.
Recently, the Vienna consensus supported by the European Society of Human Reproduction and Embryology (ESHRE) and Alpha Scientists in Reproductive Medicine interest groups proposed definitions and recommended values for indicators used in IVF laboratories, including ICSI. Minimum performance-level ('competency') and aspirational ('benchmark') values for fertilization rates (no. oocytes with 2PN and 2PB/ no. MII oocytes injected × 100) were established at ≥65% and ≥80%, respectively (The Vienna consensus, 2017) . These figures were based on the use of the conventional-ICSI technique, where bevelled, spiked, glass micropipettes are used to penetrate the oolemma at the same time as the cytoplasm is aspirated into the micropipette and the sperm is injected in the oocyte (Palermo et al., 1992; Vanderzwalmen et al., 1996) . In the procedure, the force executed to cross the oolemma and the volume of cytoplasm aspirated into the micropipette can vary with every oocyte and operator, which may cause an impact in the fertilization and further development of the oocyte (Vanderzwalmen et al., 1996). In addition, oocytes are subject to potential stress or damage during the procedure, such as spindle dislocation due to large deformation or aspiration of cytoplasm, which could also lead to decreased developmental potential (Cohen, 2004; Montag et al., 2006; Tomari et al., 2018; Vanderzwalmen et al., 1996) or result in the complete degeneration or lysis of the oocyte that typically accounts for approximately 5-10% of the injected oocytes(The Vienna consensus 2017). An alternative method involving Piezo-ICSI has been proposed (Kimura and Yanagimachi, 1995; Yanagida et al., 1999) and reported significantly higher fertilization rates and lower percentages of oocyte degeneration in several studies when compared to conventional-ICSI (Fujii et al., 2020; Furuhashi et al., 2019; Hiraoka et al., 2022, 2018; Hiraoka and Kitamura, 2015; Takeuchi et al., 2001; Yanagida et al., 1999; Zander-Fox et al., 2021) . One of the advantages of this method is that there is no need to aspirate the cytoplasm of the oocyte into the micropipette for the breakage the oolemma, which makes the technique less invasive and more consistent (Hiraoka et al., 2018; Takeuchi et al., 2001; Yanagida et al., 1999) . However, the procedure requires expensive equipment and has a longer learning curve compared to conventional-ICSI, which makes it more difficult to be implemented globally.
The use of less-invasive techniques that can reduce oocyte damage resulting from ICSI procedures combined with the development of automated systems to simplify the control of the micropipettes and microinjectors holds potential to help standardize results, allow less experienced embryologists to perform the procedure, and democratize access to infertility treatment by reducing labor costs.
The past decade witnessed significant engineering efforts to automate cell injection using automation and robotic approaches. These systems all borrowed the architecture directly from manual operation, but automated only a few steps of the process (Abdullah et al., 2022; Leung et al., 2011; Lu et al., 2011; Tan et al., 2009; Yagoub et al., 2022a, 2022b; Zhang et al., 2019) and they have not been tested clinically yet.
We developed a robotic system that allows the performance of automated intracytoplasmic sperm injection (ICSIA) with minimal human involvement and no need for micromanipulation skills. The robot was engineered to integrate AI algorithms, optics, cell microinjectors, mechatronics and a piezo-drill actuator. First, AI algorithms were developed and trained to identify morphological structures of MII oocytes in different animal models and human oocytes. Afterwards, the performance of robotic injection was evaluated in proof-of-concept studies carried out in mouse, hamster and rabbit oocytes. Subsequently, the ICSIA robot was tested in pre-clinical experiments using discarded human oocytes that were matured in vitro and injected with latex microspheres. Finally, a clinical pilot study was conducted using human donor oocytes that were randomized either to ICSIA or manual injection with patient's sperm. In the different models tested, the ICSIA device was controlled by engineers with no micromanipulation experience and results compared to those obtained with manual ICSI performed by experienced embryologists.
METHODS
Proof of concept studies in animals
Animal care and procedures were conducted according to protocols approved by the Ethics Committee on Animal Research of the Science Park of Barcelona (PCB), Spain. All animals were purchased from Janvier Labs. Upon arrival, they were quarantined and acclimated to the PCB Animal facility (PRAAL) for a minimum of one week prior to use and housed in a room with a 12-hr light/dark cycle (lights on at 7:00 A.M.) and ad libitum access to food and water.
Pre-clinical and clinical trial in human oocytes
The project received approval from two independent Institutional Review Boards (IRBs) to work with human oocytes. The first IRB was approved to conduct a pre-clinical validation using discarded human oocytes from ICSI cycles, as these are immature at the time of denudation (IRB: Comité de Ética de la Investigación con medicamentos del Parc de Salut Mar, Barcelona, Spain; Protocol # 2021/9933/I). The second IRB was obtained to conduct a clinical pilot study to evaluate the feasibility of the developed robot in a few oocyte-donation cycles (WCG IRB, USA, study # 1318771). Specific consent was obtained from donors and patients participating in the two phases of the project.
The ICSIA device
The Automated Intracytoplasmic Sperm Injection (“ICSIA”) robot is a prototype device that automates sperm injection into oocytes. The robot was developed to automate critical steps of the ICSI procedure, including ICSI pipette motion control, zona pellucida and oolemma perforation with piezo-pulses, and oocyte analysis using Artificial Intelligence (AI) and Computer Vision (CV) algorithms. Manual intervention is only required for immobilization and capture of sperm cells using the ICSI pipette and to release the targeted sperm cell in the oocyte cytoplasm, without the need for micromanipulation skills. The time taken by ICSIA system to perform the sperm injection once the sperm is manually captured and loaded in the ICSI needle is comparable to that needed by an experienced embryologist to perform the technique manually, approximately 1 min/oocyte, regardless of the oocyte species used to validate the system. All the components used in the ICSIA robot are commercial off-the-self (COTS) products and most are common to the set-up used in the conventional manual ICSI technique (Figure 1). Briefly, the ICSIA robot is a multi-axis machine integrated in an anti-vibration platform, able to control the movement of pipettes and petri dishes, which uses AI and CV algorithms with the images originating from a digital microscope. The system integrates a Piezo drill unit that is controlled and activated by CV and uses a blunt-end pipette (internal diameter 6 µm and angle 25°) and a non-toxic dumping fluid (Perfluorooctane). The piezo-Unit has been integrated in the ICSIA robot to reduce the strain on the zona pellucida and oolemma, which may vary from oocyte to oocyte. By using piezo pulses to penetrate these structures, the injection of the sperm cells can be performed more consistently and with minimal cytoplasm displacement than when the procedure is performed manually with the conventional technique, resulting eventually also less traumatic to the oocyte. The ICSIA device is patent pending (US Provisional 63342793 Priority Date May 17, 2022).
Figure 1(a) CAD model of the automated ICSI station with protective cover panels. (b) The ICSIA station with the piezo-drill unit.
An AI pipeline was created to be responsible for detecting the position of the different elements visible under the microscope, so that it could communicate to the ROS2 system to direct the ICSIA motors on each axis in the appropriate direction and distance. The first step of the AI pipeline is to align the injection pipette with the oocyte's edge at the same height plane to ensure precise puncture. Once the pipette and the oocyte are aligned, a You Only Look Once (YOLO) detection model is used to determine the precise position of the pipettes. An image segmentation algorithm is responsible for classifying each pixel of the oocyte image into one of four parts: the zona pellucida, perivitelline space, polar body, and oolema. The segmentation model used is a U-Net with few modifications in the last layers to facilitate the understanding of the oocyte's structure by reducing the complexity of the model.
Training of detection AI algorithms
The detector algorithms were trained using images captured by the robot's camera and labeled according to the criteria established by senior embryologists. High performances (over 0.99 IoU score) were achieved with the pipettes detection algorithms after using Data Augmentation approaches, as the nature of the problem was simple, and the models benefit with the high contrast of the pipettes in comparison to the background of the image.
Training of segmentation AI algorithms
The segmentation algorithms have been trained with over 600 MII oocytes images (different species) from the robot's imaging system and labelled by senior embryologists. A model has been fine-tuned for every animal (or human) species to increase the correctness of the algorithms. To compensate for the limited amount of data available, the models were augmented with data and simplified to ensure a robust algorithm that is adapted to the problem.
Collection of mouse oocytes, sperm, manipulation and embryo culture
For oocyte collection, mouse hybrid B6CBAF1 females were induced to super ovulate by intraperitoneal injection of 5 IU of pregnant mare serum gonadotropin (PMSG) followed 48 h later by 5 IU of human chorionic gonadotropin (hCG). Cumulus–oocyte complexes were released from the oviducts by 14–15 h after hCG administration and treated with hyaluronidase until cumulus cells dispersed. Denuded oocytes were then washed several times and kept in culture medium under oil at 37°C, in an atmosphere with 7% CO2 and O2 in air, until use. Fresh mouse spermatozoa were collected from the cauda epididymis taken from an adult male in a microdroplet of culture medium and cultured for 15 min at 37°C, in an atmosphere with 7% CO2 and O2 in air. After incubation, 3 μL of the concentrated sperm solution were further diluted into a 150 μL droplet of culture medium. Sibling oocytes were randomly assigned to be injected with sperm heads either with ICSIA robot or manually with a piezo-device (PiezoXpert, Eppendorf). After injection, all oocytes were washed thoroughly and cultured uninterruptedly in single step medium (KSOM, Millipore) supplemented with 5% HSA in a benchtop incubator under optimal conditions up to 96 h post-ICSI.
Cell counts in mouse blastocysts
Mouse blastocysts were fixed in a fixative solution for 10 min at 37°C. Afterwards, cell nuclei were stained by incubation of the blastocysts in a 10 μg/ml bisbenzimide solution (Hoechst 33342) for 10 min at room temperature. The stained blastocysts were then washed and mounted on a glass slide in a 1:1 phosphate-buffered-saline–glycerol droplet and flattened with a cover slip. On observation under UV-2A filter fluorescence microscopy (Olympus BX43F), the total cell number was assessed manually with an image processing software (ImageJ; NIH, USA).
Vitrification and warming of mouse blastocysts
Mouse blastocysts obtained by ICSI in the manual control and test group were vitrified using the standard Cryotop method following the protocol provided by the manufacturer (Kitazato BioPharma Co., Ltd, Japan). Briefly, samples were exposed to equilibration solution (ES) at room temperature for 13 to 15 min and transferred to vitrification solution (VS) for a maximum of 1 min. Afterwards, blastocysts were loaded onto the surface strip of a Cryotop device and directly plunged into LN2. For the warming procedure, the Cryotop device containing the blastocysts was transferred from the LN2 into the first warming solution (WS, 4ml, at 37°C) for 1 min. Then, embryos were gradually moved to dilution solution (DS) for 3 min, to washing solution (WS) for 5 min, and finally to a new droplet of WS for additional 1 min (at room temperature). After warming, blastocysts were extensively washed and kept in culture medium until being transferred into synchronized recipients.
Mouse embryo transfers
Outbred CD-1 female mice (8–12 weeks of age) were mated with vasectomized males to induce pseudopregnancy. Successful mating was confirmed by the presence of a vaginal plug on the following morning. Blastocysts were transferred surgically into the uteri of the pseudo-pregnant synchronized females 2.5 days after mating. On each embryo transfer procedure, a maximum of 10 blastocysts were transferred with a glass polished pipette into the lumen of one uterine horn of the recipients.
Collection of hamster oocytes
Mature golden hamster females (2–3 months old) were induced to super ovulate by intraperitoneal injection of 30 IU PMSG on the day of the post-estrus discharge followed by an intraperitoneal injection of 30 IU hCG 56 h later. Mature MII oocytes were collected from oviducts approximately 15 h after hCG injection. Afterwards, they were freed from the cumulus cells by a 1-min treatment hyaluronidase until cumulus cells dispersed. Denuded oocytes were then washed several times and kept in culture medium under oil at 37°C, in an atmosphere with 7%CO2 and O2 in air, until use. Hamster oocytes were used to train the AI algorithms and randomly assigned to be injected with a human sperm either with the ICSIA robot or manually with conventional-ICSI. This human-sperm-hamster-oocyte model has been previously proposed and used routinely for training and assessing the technical competency of operators with the conventional manual ICSI technique in human IVF laboratories (Gvakharia et al., 2000). After injection, oocytes were washed thoroughly and cultured in KSOM medium supplemented with 5% HSA in a benchtop incubator at 37°C, in an atmosphere with 7%CO2 and O2 in air. Survival rates were assessed in an inverted microscope by 2 h post-ICSI and after overnight culture.
Collection of rabbit oocytes, sperm, manipulation and embryo culture
Rabbit oocytes were collected as previously described (Jiménez-Trigos et al., 2013). Briefly, New Zealand White female rabbits (2-3 months old) were injected with 25 I.U of hCG (Coriogan, Lab. Ovejero, Spain) and oocytes recovered from oviducts at 14 h after endovenous hCG treatment by flushing in Dulbecco's phosphate-buffered saline supplemented with 20% (v/v) serum. The collected oocytes were then denuded by briefly exposure to hyaluronidase and gently pipetted through a small-bore pipette. After treatment, oocytes were selected morphologically according to the integrity of the first polar body (PB1) and kept in culture medium under oil at 38.5°C, in an atmosphere with 7% CO2 and O2 in air, until use. Selected oocytes were vitrified with the Cryotop method following the protocol recommended by the manufacturer (Kitazato BioPharma Co., Ltd, Japan). Two-hours before use, oocytes were warmed following the manufacturer's recommended warming protocol (Kitazato BioPharma Co., Ltd, Japan). Semen was collected, cryopreserved and thawed, as previously described (Jiménez-Trigos et al., 2013). Rabbit oocytes were used to train the AI algorithms and randomly assigned to be injected with rabbit sperms either with the ICSIA robot or manually with conventional-ICSI. After injection, oocytes were washed thoroughly and cultured uninterruptedly in KSOM medium supplemented with 10% FBS up to 144 h post-ICSI in a benchtop incubator at 38.5°C, in an atmosphere with 7% CO2 and O2 in air.
Pre-clinical validations in discarded human oocytes
The human discarded oocytes were in vitro matured in a homemade medium (TCM-199 supplemented with serum, hormones, growth factors and antioxidants) at 37°C and a humidified atmosphere with 6% CO2. Oocyte maturation progress was monitored until a maximum of 48h. Oocytes showing a first PB were considered mature and were used, firstly, to train the AI algorithms in identifying oocyte morphological structures (zona pellucida, perivitelline space, oolemma, cytoplasmic inclusions and PB) and then to evaluate the injection efficiency and survival rates of the ICSIA robot. In this case, injections were performed with latex microspheres similar in size to the head of a human sperm (approximately 5 µm) using vitrified-warmed or fresh oocytes that were randomly assigned to be injected with the ICSIA robot or manually.
After injection, oocytes were washed thoroughly and cultured in KSOM medium supplemented with 5% HSA in a benchtop incubator at 37°C, in an atmosphere with 7% CO2 and O2 in air. Survival rates were assessed in an inverted microscope by 2 h post-ICSI and after overnight culture.
Clinical pilot trial in human donor oocytes
Donors were selected and screened under FDA regulations. Donors were downregulated with LUPRON/LEUPROLIDE (Sun Pharmaceutical, India) in luteal phase for 10-14 days. Ovarian Stimulation was performed using FSH injection 225 units over the course of 12-14 days. When the lead follicle reached 22mm or above, 30 units of LUPRON/LEUPROLIDE (Sun Pharmaceutical, India) were administered. Oocyte Retrieval was performed 34-36 hours later. Sibling oocytes were randomly assigned and inseminated using either manual ICSI or ICSIA, as described above. A video of the injection procedure performed by the ICSI-A robot is shown in the supplementary video. At blastocyst stage all embryos were biopsied and vitrified. Preimplantation genetic testing for aneuploidy (PGT-A) screening was performed by using PGTai 2.0 (CooperSurgical, Inc.) which utilizes sequencing on a NextSeq™ (Illumina, Inc.) with a paired-end 2 × 36 base pair strategy targeting 4 M raw reads. PGTai 2.0 interrogates drift in heterozygous SNP ratios as an aggregate signature across CNV regions called by the base algorithms. Genome-wide heterozygous SNP ratio drift was also used to identify haploid and 69,XXX triploid biopsies. For transfer of euploid embryos, the donor oocyte recipients underwent hormone replacement therapy (HRT), starting with Estrace/Estradiol 4mg daily for 12-18 days. At the time that endometrial lining reached 7-9mm, Progesterone 800mg was initiated vaginally daily, and embryo transfer was scheduled 6 ½ days after start of Progesterone medication. Transvaginal guided embryo transfer was performed using Kitazato catheter (Manufacturer: Kitazato Corporation, Japan). A pregnancy test was performed 7 days after blastocyst transfer. Controlled positive pregnancy results were monitored weekly with blood test analysis (estradiol, progesterone, quantitative HCG) and by transvaginal ultrasounds 4 weeks after embryo transfer.
Statistical analysis
The results were analysed using Student's t test for comparison of averages and a chi-squared test for comparison of proportions. P<.05 was considered statistically significant. Statistical analyses were performed using the Statistical Package for the Social Sciences 22.0 (SPSS Inc.).
RESULTS
Proof of concept in animals
Robot injection performance was first evaluated in proof-of-concept studies carried out in mouse (Figure 2), hamster (Figure 3) and rabbit (Figure 4) oocytes.
Figure 2Representative images of the automated ICSI process in the mouse model. (a) A MII mouse oocyte held in the holding pipette before starting the analysis. (b) A segmented image of the oocyte after AI analysis showing the identification of its morphological structures in different colors, including the zona pellucida (light blue), perivitelline space (green), first polar body (red), and cytoplasm (dark blue). The horizontal line displayed in green represents the trajectory to be followed by the injection pipette, while the two red lines indicate the area to be avoided during injection. (c) The points of interest (black crosses) showed in the oocyte are references to instruct the activation of the piezo pulses for the perforation of the zona pellucida and oolema. (d-f) sequential images of the sperm injection procedure. (g) Mouse blastocysts produced by automated sperm injection. (h) A blastocyst processed for total cell count analysis with nuclei stained. (i) First pups produced with the automated ICSI device two days after birth.
Figure 3Representative images of the automated ICSI process in the human sperm-hamster oocyte model. (a) A MII hamster oocyte held in the holding pipette before starting the analysis. (b) A segmented image of the oocyte after AI analysis showing the identification of its morphological structures in different colors, including, the zona pellucida (light blue), perivitelline space (green), first polar body (red), and cytoplasm (dark blue). The horizontal line displayed in green represents the trajectory to be followed by the injection pipette, while the two red lines indicate the area to be avoided during injection. (c) The points of interest (black crosses) showed in the oocyte are references to instruct the activation of the piezo pulses for the perforation of the zona pellucida and oolema. (d-f) sequential images of the sperm injection procedure. (g) A hamster oocyte with normal morphology 2 h after being injected with a human sperm with the automated device. (h) A hamster oocyte injected with a human sperm using the ICSIA device showing pronuclei (white arrow) after overnight culture. (i) A hamster oocyte injected with the ICSIA device that lysed after the procedure.
Figure 4Representative images of the automated ICSI process in rabbit oocytes. (a) A MII rabbit oocyte held in the holding pipette before starting the analysis. (b) A segmented image of the oocyte after AI analysis identifying the zona pellucida (light blue), perivitelline space (green) and cytoplasm (dark blue). The horizontal line displayed in green represents the trajectory to be followed by the injection pipette. (c) The points of interest (black crosses) showed in the oocyte are references to instruct the activation of the piezo pulses for the perforation of the zona pellucida and oolema. (d-f) Sequential images of the sperm injection procedure. (g) Rabbit oocytes showing normal morphology after being injected with a rabbit sperm with the automated device. (h) Rabbit blastocysts produced with the ICSIA. (i) An expanded rabbit blastocyst showed at higher magnification (40x).
In the mouse model, 84.1% (n=201) of the oocytes assigned to the automated ICSIA device survived the injection procedure, of which, 89.3% (n=151) developed to two-cells and 73.9% (n=125) to the expanded blastocyst stage. No statistical differences were found when these efficiencies were compared with manual controls performed with a Piezo-actuator (n=206, 88.8% (n=183), 90.7% (n=166) and 81.9% (n=150), respectively) (Figure 2,Table I). To ascertain the quality of the blastocysts produced, some of those obtained in the ICSIA group were processed for total cell counts (average +/- SD, 120.3 +/- 27.4; n=26), showing an average number of cells statistically equivalent to those produced in the control group (125.7 +/- 22.2; n= 20). Nine pups were born as result of the transfer of mouse blastocysts produced by ICSIA (19.1%). This efficiency contrasts with the 43.1% full-term development obtained in the control group. However, parent–offspring cannibalism was detected in two cages from the ICSIA group (Table I).
Table I– Results from the validation performed in the mouse model.
Using the hamster oocyte-human sperm model (Figure 3,Table II), a survival rate of 93.1% (n=43) was achieved with the ICSIA robot when oocytes were assessed by 2 h after injection, which was statistically similar to the results obtained in the controls injected manually (92%, n=125) at the same time-point. After overnight culture, two oocytes degenerated in each group, resulting in a 90.4% survival rate in the control and 88.1% in the ICSIA group.
Table II– Results from the validation performed in the hamster model.
When rabbit oocytes were used and injected with rabbit sperm, a total of 30 out of 31 oocytes (96.7%) survived the injection procedure in the ICSIA group, out of which 10 (33.3%) cleaved and 8 developed into good quality blastocysts (26.7%) (Figure 4,Table III). The results obtained from the ICSIA robot that was controlled by an engineer with no previous micromanipulation experience, are statistically equivalent to those obtained in the control group performed manually with conventional ICSI by experienced embryologists (n=25, 100% survival, 40% cleavage and 24% blastocysts) (Figure 4,Table III).
Table III– Results from the validation performed in the rabbit model.
In a pre-clinical validation with human oocytes, a total of 124 immature discarded oocytes were matured in vitro, out of which, 84 showed a first PB (66.1% maturation rate) within 48 h after culture. These oocytes were considered to have reached the MII stage and thus were scanned by the robot and the resultant images used to train the AI algorithms. The AI algorithms were successful in detecting the morphological structures of the human oocytes, as well as, in analyzing the injection pipette, and in choosing the best focal plane for injection. Thirty-seven out of 40 MII oocytes injected with the ICSIA robot (92.5%) survived the procedure, while 95.6% (n=46) survival rates were obtained in the control group performed manually with conventional-ICSI. The results were statistically similar, and similar to the expected benchmark (Special et al., 2017) . All oocytes that survived the procedure showed a microsphere inside the cytoplasm by 2 h post-ICSI and after overnight culture (Figure 5,Table IV).
Figure 5Representative images of the pre-clinical validation experiments performed in human oocytes. (a-c) The in vitro maturation progress of a human oocyte at germinal vesicle (a), MI (b) and MII (c). (d-f) Images of the human oocytes injected with beads. (d) A human oocyte with normal morphology 2 h after being injected with a bead (white circle) with the automated device. (e) A human oocyte injected with a bead (white circle) after overnight culture. (f) A human oocyte injected with the ICSIA device that lysed after the procedure.
Thirty-two metaphase II oocytes were obtained from 3 donors which were randomly assigned to be injected either with the ICSIA robot controlled by an engineer or manually with conventional ICSI by an experienced embryologist. In total, 14 oocytes were injected with ICSIA system and 18 with manual ICSI. Out of the 14 oocytes injected with ICSIA, 13 (93%) fertilized, 8 (61.5%) developed into good quality blastocysts, and 50% of the blastocysts were diagnosed by PGT-A as chromosomally normal (Table V). A video of the injection procedure performed by the ICSI-A robot is shown in the supplementary video. Of the oocytes manually injected, 16 (88.8%) fertilized, 12 (75%) reached the blastocyst stage, and 10 (83%) blastocysts were classified as euploid. Two transfers of euploid blastocysts (one SET and one DET) obtained in the ICSIA group were transferred and resulted in the birth of two healthy babies () (Figure 6,Table VI).
Figure 6Representative images of the clinical trial performed in human oocytes. (a) A MII human oocyte held in the holding pipette before starting the analysis. (b) A segmented image of the oocyte after analysis by AI showing the identification of its morphological structures in different colors, including the zona pellucida (light blue), perivitelline space (green), first polar body (red), and cytoplasm (dark blue). The horizontal line displayed in green represents the trajectory to be followed by the injection pipette, while the two red lines indicate the area to be avoided during injection. (c) The points of interest (black crosses) showed in the oocyte are references to instruct the activation of the piezo pulses for the perforation of the zona pellucida and oolema. (d-f) Sequential images of the sperm injection procedure performed by the automated device. (g-i) Embryo development sequences of the oocyte injected with the ICSIA device that resulted in the first clinical pregnancy.
The manual nature of IVF processes challenges the reproducibility and efficiency of ART. Many existing steps of the IVF process are labor- and time-consuming, as well as subject to high inter- and intra-operator variability (Abdullah et al., 2022). This is reflected in the high variability in clinical outcomes reported by IVF clinics (i.e., SART, 2019). The best comparison is between oocyte donor results since they are all fertile and chosen with similar criteria. In that category, birth rates in oocyte donor cycles range from 14% to 74% (SART, 2019). Therefore, a significant increase in IVF success rates may require optimization and consistency of protocols, which may only be possible with the automation of critical steps of the IVF process. Aside from time-lapse incubators (Armstrong et al., 2014), biochips to sort sperm (Ahmadkhani et al., 2022; de Wagenaar et al., 2016), and a vitrification machine (Gatimel et al., 2021; Roy et al., 2017, 2014) no other devices are commercially available that allow to automate laboratory procedures.
Automation in IVF is desired also because techniques like oocyte/embryo vitrification, ICSI or embryo biopsy, all require high skill levels and currently there is a scarcity worldwide of highly trained embryologists, while the demand for IVF treatments is increasing faster.
In particular, for ICSI, there have been some prototypes described that attempted to its automation (Leung et al., 2011; Tan et al., 2009; Yagoub et al., 2022a), but none has reached clinical stage yet. The ICSIA device developed in our project demonstrated a high degree of consistency and operator skill independence in three different animal models, such as mouse, hamster, and rabbit, while allowing human supervision and minimal manual intervention. After validating that the ICSIA robot performs consistently in the different animal models, we decided that it was safe to test it in human oocytes and to move to a clinical pilot study. The ICSIA robot showed high proficiency in injecting human oocytes and results obtained were within the key performance indicators described for this technique as a reference for embryologists and IVF laboratories (The Vienna consensus, 2017). However, euploidy rates were lower than expected compared to the results obtained in the manual group. We attribute this to potential technical aspects needing further improvement, such as speed and temperature control, and/or to the fact that sperm selection in the ICSIA arm was carried out by an engineer with no embryology experience, while in the manual group it was carried out by an experienced embryologist. Nevertheless, to our knowledge, this is the first time that babies resultant from an automated ICSI system are reported.
Next steps will involve improving engineering processes to speed up the sperm injection and to integrate automated sperm immobilization and capture as part of the steps carried out by the robot automatically. Additional validations will be performed to elucidate this and enlarge the sample size of injected human oocytes in both arms.
The combination of multidisciplinary teams allows the development of automated processes that can reduce variability in certain IVF procedures, while supervised and assisted by experienced embryologists. It is expected that other laboratory procedures can be automated in the field of assisted reproductive treatments in a near future.
CONTRIBUTORS
NC-B, LM, GC and SM conceptualized the study. LM, EA, GG, SM engineered the ICSIA robot. LM and SM developed AI and CV algorithms. NC-B, EM, MA, QM, CC, performed IVF methods in animal models. LM, MR, CM, IV, AP, DM, CC, NC-B performed IVF methods involving discarded human oocytes. ZL, CC, JZ recruited donors, patients and performed medical treatments involved in the clinical pilot trial. EA and GG operated the ICSIA robot. NC-B, EM, MA, QM, GC, SM performed formal analysis. NC-B performed statistical analyses and prepared figures. NC-B, GC, SM, JZ supervised the study. SM and JZ acquired funding. NC-B, SM, JZ, MS-S wrote the original draft and revised the manuscript with input from all authors. All authors had final responsibility for the decision to submit for publication and assume responsibility for the accuracy and completeness of the analyses and the fidelity of this manuscript.
REFERENCES
Abdullah KAL, Atazhanova T, Chavez-Badiola A, Shivhare SB. Automation in ART: Paving the Way for the Future of Infertility Treatment. Reproductive Sciences 2022; published online Aug 3. DOI:10.1007/s43032-022-00941-y.
Ahmadkhani N, Hosseini M, Saadatmand M, Abbaspourrad A. The influence of the female reproductive tract and sperm features on the design of microfluidic sperm-sorting devices. J Assist Reprod Genet 2022; 39: 19–36.
Armstrong S, Arroll N, Cree LM, Farquhar C. Time-lapse systems for embryo incubation in assisted reproduction. In: Armstrong S, ed. Cochrane Database of Systematic Reviews. Chichester, UK: John Wiley & Sons, Ltd, 2014. DOI:10.1002/14651858.CD011320.
Cohen Y. Spindle imaging: a new marker for optimal timing of ICSI? Human Reproduction 2004; 19: 649–54.
de Wagenaar B, Dekker S, de Boer HL, et al. Towards microfluidic sperm refinement: impedance-based analysis and sorting of sperm cells. Lab Chip 2016; 16: 1514–22.
Esteves SC. Intracytoplasmic sperm injection versus conventional IVF. The Lancet 2021; 397: 1521–3.
Fujii Y, Endo Y, Mitsuhata S, Hayashi M, Motoyama H. Evaluation of the effect of piezo‐intracytoplasmic sperm injection on the laboratory, clinical, and neonatal outcomes. Reprod Med Biol 2020; 19: 198–205.
Furuhashi K, Saeki Y, Enatsu N, et al. Piezo‐assisted ICSI improves fertilization and blastocyst development rates compared with conventional ICSI in women aged more than 35 years. Reprod Med Biol 2019; 18: 357–61.
Gatimel N, Moreau J, Bettiol C, Parinaud J, Léandri RD. Semi-automated versus manual embryo vitrification: inter-operator variability, time-saving, and clinical outcomes. J Assist Reprod Genet 2021; 38: 3213–22.
Gvakharia MO, Lipshultz LI, Lamb DJ. Human sperm microinjection into hamster oocytes: a new tool for training and evaluation of the technical proficiency of intracytoplasmic sperm injection. Fertil Steril 2000; 73: 395–401.
Hiraoka K, Kitamura S, Otsuka Y, Kawai K, Harada T, Ishikawa T. Effects of sperm direction in Piezo-ICSI on oocyte survival, fertilization, embryo development and implantation ability in humans: A preliminary study. Journal of Obstetrics and Gynaecology Research 2018; published online July 11. DOI:10.1111/jog.13727.
Hiraoka K, Kitamura S. Clinical efficiency of Piezo-ICSI using micropipettes with a wall thickness of 0.625 μm. J Assist Reprod Genet 2015; 32: 1827–33.
Hiraoka K, Uchida N, Ishikawa T, Kawai K. Value of Meiotic Spindle Imaging on Fertilization and Embryo Development Following Human Oocyte Piezo-ICSI. Fertility & Reproduction 2022; 04: 26–31.
Jiménez-Trigos E, Vicente JS, Marco-Jiménez F. Live Birth from Slow-Frozen Rabbit Oocytes after In Vivo Fertilisation. PLoS One 2013; 8: e83399.
Kimura Y, Yanagimachi R. Intracytoplasmic Sperm Injection in the Mouse1. Biol Reprod 1995; 52: 709–20.
Leung C, Zhe Lu, Esfandiari N, Casper RF, Yu Sun. Automated Sperm Immobilization for Intracytoplasmic Sperm Injection. IEEE Trans Biomed Eng 2011; 58: 935–42.
Lu Z, Zhang X, Leung C, Esfandiari N, Casper RF, Sun Y. Robotic ICSI (Intracytoplasmic Sperm Injection). IEEE Trans Biomed Eng 2011; 58: 2102–8.
Montag M, Schimming T, van der Ven H. Spindle imaging in human oocytes: the impact of the meiotic cell cycle. Reprod Biomed Online 2006; 12: 442–6.
Palermo G. Pregnancies after intracytoplasmic injection of single spermatozoon into an oocyte. The Lancet 1992; 340: 17–8.
Roy TK, Brandi S, Tappe NM, et al. Embryo vitrification using a novel semi-automated closed system yields in vitro outcomes equivalent to the manual Cryotop method. Human Reproduction 2014; 29: 2431–8.
SART. National Summary Report. 2019.
Special E, Group I, Scientists A. The Vienna consensus: report of an expert meeting on the development of ART laboratory performance indicators. Reprod Biomed Online 2017; 35: 494–510.
Takeuchi S, Minoura H, Shibahara T, Shen X, Futamura N, Toyoda N. Comparison of Piezo-Assisted Micromanipulation with Conventional Micromanipulation for Intracytoplasmic Sperm Injection into Human Oocytes. Gynecol Obstet Invest 2001; 52: 158–62.
Tan KK, Huang S, Tang KZ. Robust computer-controlled system for intracytoplasmic sperm injection and subsequent cell electro-activation. The International Journal of Medical Robotics and Computer Assisted Surgery 2009; 5: 85–98.
The Vienna consensus: report of an expert meeting on the development of art laboratory performance indicators†‡. Hum Reprod Open 2017; 2017. DOI:10.1093/hropen/hox011.
Tomari H, Honjo K, Kunitake K, et al. Meiotic spindle size is a strong indicator of human oocyte quality. Reprod Med Biol 2018; 17: 268–74.
Vanderzwalmen P, Bertin G, Lejeune B, Nijs M, Vandamme B, Schoysman R. Two essential steps for a successful intracytoplasmic sperm injection: injection of immobilized spermatozoa after rupture of the oolema. Human Reproduction 1996; 11: 540–7.
Yagoub SH, Thompson JG, Orth A, Dholakia K, Gibson BC, Dunning KR. Fabrication on the microscale: a two-photon polymerized device for oocyte microinjection. J Assist Reprod Genet 2022; 39: 1503–13.
Yagoub SH, Thompson JG, Orth A, Dholakia K, Gibson BC, Dunning KR. Fabrication on the microscale: a two-photon polymerized device for oocyte microinjection. J Assist Reprod Genet 2022. DOI:10.1007/s10815-022-02485-1.
Yanagida K, Katayose H, Yazawa H, Kimura Y, Konnai K, Sato A. The usefulness of a piezo-micromanipulator in intracytoplasmic sperm injection in humans. Human Reproduction 1999; 14: 448–53.
Zander-Fox D, Lam K, Pacella-Ince L, et al. PIEZO-ICSI increases fertilization rates compared with standard ICSI: a prospective cohort study. Reprod Biomed Online 2021; 43: 404–12.
Zhang Z, Dai C, Huang J, et al. Robotic immobilization of motile sperm for clinical intracytoplasmic sperm injection. IEEE Trans Biomed Eng 2019; 66: 444–52.
Data availability
Data will be made available on request.
Declaration of Competing Interest
SM is a shareholder and officer of Overture Life. NC-B and GC are shareholders of Embryotools. JZ is shareholder of New Hope Fertility Center.
ROLE OF THE FUNDING SOURSE
The engineering project was funded by Overture Life, the PGT test by Cooper Surgical, and the human clinical pilot by New Hope Fertility Center. All authors had access to primary data and accept responsibility for accuracy and completeness of data. The corresponding authors had access to all data and final responsibility for the decision to submit for publication.
Santiago Munné, Ph.D. Chief Innovation Officer at Overture Life, with >260 peer-reviewed publications and multiple prize papers. He developed the first assays of Preimplantation Genetic Testing for Aneuploidy (PGT-A) and obtained the first pregnancies. Cofounded several companies in reproductive medicine such as Reprogenetics, Recombine, Phosphorus, MedAnswers, Overture Life and HoMu Health ventures.
Article info
Publication history
Accepted:
May 17,
2023
Received in revised form:
May 14,
2023
Received:
March 17,
2023
Publication stage
In Press Journal Pre-Proof
Footnotes
Funding: This study was privately funded by Overture Life and New Hope Fertility Center.
RBMO has just published a paper by Nuno Costa-Borges and colleagues in which the authors announce the birth of the first babies following clinical application of automated intracytoplasmic sperm injection (ICSIA). This is an important achievement and a milestone. It is also welcome news to those who see the letter 'T' in 'ART' (assisted reproductive technology) as fundamental to progress in the field.
Parent–offspring cannibalism detected in two cages from the ICSIA group.
00335-8&id=gr1.jpg){kind=link}
00335-8&id=gr2.jpg){kind=link}
00335-8&id=gr3.jpg){kind=link}
00335-8&id=gr4.jpg){kind=link}
00335-8&id=gr5.jpg){kind=link}
00335-8&id=gr6.jpg){kind=link}