Predictive ability of hyperspectral signatures from single immobilized spermatozoa as a novel approach to non-invasive sperm selection during ICSI

STUDY QUESTION: Can hyperspectral imaging of single immobilized spermatozoa prior to ICSI be used to characterize individual spermatozoa quality and develop predictive models for fertilization, blastulation, and blastocyst usability? SUMMARY ANSWER: Hyperspectral imaging characterization of individual immobilized spermatozoa was successfully implemented prior to ICSI and enabled the development of predictive models for key reproductive outcomes. WHAT IS KNOWN ALREADY: Sperm selection during ICSI is mainly based on morphology and motility, which provide limited information about functional competence. Most techniques capable of assessing molecular sperm quality are destructive and cannot be applied to the injected spermatozoon. Hyperspectral imaging (HS) is a non-invasive technique capable of generating molecular spectral fingerprints of individual cells. STUDY DESIGN, SIZE, DURATION: This prospective two-phase study was conducted between March 2020 and August 2024 and included 93 couples undergoing ICSI cycles at a single IVF centre. PARTICIPANTS/MATERIALS, SETTING, METHODS: Single immobilized spermatozoa were imaged prior to ICSI using a hyperspectral camera coupled to a micromanipulation microscope. Phase I evaluated the feasibility and short-term technical repeatability of hyperspectral acquisition, as well as preliminary classification consistency across repeated acquisitions. Phase II included a safety evaluation using a sibling donor-oocyte design and the development of predictive models using hyperspectral spectra and clinical data. Machine learning models based on XGBoost decision trees were trained to predict fertilization, blastulation, and usable blastocyst formation. MAIN RESULTS AND THE ROLE OF CHANCE: Hyperspectral imaging was successfully implemented prior to ICSI. Phase I supported high short-term technical repeatability under stable acquisition conditions, with a global intraclass correlation coefficient of 0.999. Exploratory spectral classification across repeated technical acquisitions achieved a sensitivity of 93.84% and a specificity of 96.76%. In the safety evaluation, fertilization, blastulation, and usable blastocyst rates were comparable between HS-ICSI and standard ICSI groups. Predictive models based on hyperspectral spectra and clinical data were developed, with chained models integrating both data types showing the most favourable balance between predictive precision and uncertainty among the evaluated model configurations. LIMITATIONS, REASONS FOR CAUTION: Model performance may be influenced by limited training data and class imbalance; therefore, further validation in larger datasets is required. This approach may enable non-invasive functional characterization of individual spermatozoa and could support the future development of more informed sperm selection strategies during ICSI. WIDER IMPLICATIONS OF THE FINDINGS: This study demonstrates the feasibility of implementing hyperspectral imaging of single spermatozoa within the routine workflow of an IVF laboratory. This approach could enable non-invasive functional characterization, predict reproductive chances of individual spermatozoa, and support the future development of more informed sperm selection strategies during ICSI. FUNDING: This work was supported by the Centro para el Desarrollo Tecnológico Industrial (CDTI, Spain; Project ID: IDI-20210432). DISCLOSURES: The authors declare no competing interests. TRIAL REGISTRATION NUMBER: N/A.

Authors

Institutions

Publication Details

Journal
Human Reproduction
Published
2026-11-18
DOI
https://doi.org/10.1093/humrep/deag176
Primary Topic
Sperm and Testicular Function
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Predictive ability of hyperspectral signatures from single immobilized spermatozoa as a novel approach to non-invasive sperm selection during ICSI

Pilar Gámiz, Rocío Rivera‐Egea, Nicolás Garrido, María José de los Santos et al.
1 links
Human Reproduction
Sperm and Testicular Function
article

Predictive ability of hyperspectral signatures from single immobilized spermatozoa as a novel approach to non-invasive sperm selection during ICSI

Pilar Gámiz, Rocío Rivera‐Egea, Nicolás Garrido, María José de los Santos, Diana Beltrán, Laura Escrich, J.M. De los Santos, María Gil Juliá, Thamara Viloria, Rosa María Pacheco-Rendón, D. Muñoz, Rocío Trinidad
article en

Abstract

STUDY QUESTION: Can hyperspectral imaging of single immobilized spermatozoa prior to ICSI be used to characterize individual spermatozoa quality and develop predictive models for fertilization, blastulation, and blastocyst usability? SUMMARY ANSWER: Hyperspectral imaging characterization of individual immobilized spermatozoa was successfully implemented prior to ICSI and enabled the development of predictive models for key reproductive outcomes. WHAT IS KNOWN ALREADY: Sperm selection during ICSI is mainly based on morphology and motility, which provide limited information about functional competence. Most techniques capable of assessing molecular sperm quality are destructive and cannot be applied to the injected spermatozoon. Hyperspectral imaging (HS) is a non-invasive technique capable of generating molecular spectral fingerprints of individual cells. STUDY DESIGN, SIZE, DURATION: This prospective two-phase study was conducted between March 2020 and August 2024 and included 93 couples undergoing ICSI cycles at a single IVF centre. PARTICIPANTS/MATERIALS, SETTING, METHODS: Single immobilized spermatozoa were imaged prior to ICSI using a hyperspectral camera coupled to a micromanipulation microscope. Phase I evaluated the feasibility and short-term technical repeatability of hyperspectral acquisition, as well as preliminary classification consistency across repeated acquisitions. Phase II included a safety evaluation using a sibling donor-oocyte design and the development of predictive models using hyperspectral spectra and clinical data. Machine learning models based on XGBoost decision trees were trained to predict fertilization, blastulation, and usable blastocyst formation. MAIN RESULTS AND THE ROLE OF CHANCE: Hyperspectral imaging was successfully implemented prior to ICSI. Phase I supported high short-term technical repeatability under stable acquisition conditions, with a global intraclass correlation coefficient of 0.999. Exploratory spectral classification across repeated technical acquisitions achieved a sensitivity of 93.84% and a specificity of 96.76%. In the safety evaluation, fertilization, blastulation, and usable blastocyst rates were comparable between HS-ICSI and standard ICSI groups. Predictive models based on hyperspectral spectra and clinical data were developed, with chained models integrating both data types showing the most favourable balance between predictive precision and uncertainty among the evaluated model configurations. LIMITATIONS, REASONS FOR CAUTION: Model performance may be influenced by limited training data and class imbalance; therefore, further validation in larger datasets is required. This approach may enable non-invasive functional characterization of individual spermatozoa and could support the future development of more informed sperm selection strategies during ICSI. WIDER IMPLICATIONS OF THE FINDINGS: This study demonstrates the feasibility of implementing hyperspectral imaging of single spermatozoa within the routine workflow of an IVF laboratory. This approach could enable non-invasive functional characterization, predict reproductive chances of individual spermatozoa, and support the future development of more informed sperm selection strategies during ICSI. FUNDING: This work was supported by the Centro para el Desarrollo Tecnológico Industrial (CDTI, Spain; Project ID: IDI-20210432). DISCLOSURES: The authors declare no competing interests. TRIAL REGISTRATION NUMBER: N/A.

Human Reproduction
Universitat de València (ES), Valencian Infertility Institute (ES), Instituto de Investigación Sanitaria La Fe (ES), Universidad Europea de Valencia (ES), IVIRMA Global (ES), Fundación IVI (ES)
Centro para el Desarrollo Tecnológico Industrial
Good health and well-being
Sperm and Testicular Function
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon