2023 ACMG Annual Clinical Genetics Meeting Digital Edition: Poster Gallery-AI-Assisted Karyotyping Improves Efficiencies at Scale in the Cytogenetics Laboratory

Pdf Summary

This document discusses the use of AI-assisted karyotyping in cytogenetics laboratories to improve efficiency and accuracy. The researchers present the initial validation results of an AI-assisted digital metaphase chromosome analysis workflow in an academic reference laboratory setting.<br /><br />The study used AI algorithms to streamline the analysis process and save time and costs. The results showed that AI-assisted chromosome classification and orientation had a higher accuracy rate of 100% compared to the baseline accuracy of 21.7%. The AI system also provided confidence indicators for accurate classification. The AI-assisted segmentation of metaphases resulted in a reduction of 17 processing steps and 57s saved per metaphase. Bone marrow specimens showed the greatest reduction in processing steps by an average of 43%.<br /><br />The study also evaluated the recognition of whole arm translocations as a quality control measure, confirming the effectiveness of the AI system in identifying joined whole chromosomes. Specimen types evaluated included bone marrow, constitutional blood, amniotic fluid, and products of conception.<br /><br />Overall, AI-assisted segmentation of chromosomes across various sample types led to significant reductions in processing steps and time. The use of AI-based algorithms in digital analysis workflows improved efficiencies in karyotype preparation, allowing for scaled operations with quality retention. Future directions include exploring larger datasets, optimizing the system based on in-house image capture, and providing training for novice staff.<br /><br />The validation of the AI-assisted analysis workflow was performed using the manufacturer's Deep Neural Networks (DNN) AI-based algorithms. The study compared the efficiency and accuracy of DNN segmentation and classification by evaluating images without AI assistance. The processes were performed using software and image capture systems specifically designed for metaphase detection and classification.<br /><br />In summary, AI-assisted karyotyping has shown promising results in improving efficiency and accuracy in cytogenetics laboratories. The use of AI algorithms streamlines analysis processes, reduces processing steps and time, and enables scalability while maintaining high-quality services. Further research and optimization are needed to explore its full potential and provide training for laboratory staff.

Asset Subtitle

Presenting Author - R. Brian Fedderson, CG(ASCP) CM; Co-Author - Brandye Tambunga, CG(ASCP) CM; Co-Author - Gabriel Vitier, CG(ASCP) CM; Co-Author - Chantry J. Clark, C(ASCP)CM; Co-Author - Jian M. Zhao, PhD; Co-Author - Erica F. Andersen, PhD, FACMG; Co-Author - Bo Hong, MD, FACMG;

Meta Tag

Cancer Cytogenetics

Chromosomal Abnormalities

Clinical Cytogenetics

Cytogenetics

Methodology

Co-Author Brandye Tambunga, CG(ASCP) CM

Co-Author Gabriel Vitier, CG(ASCP) CM

Co-Author Chantry J. Clark, C(ASCP)CM

Co-Author Jian M. Zhao, PhD

Co-Author Erica F. Andersen, PhD, FACMG

Co-Author Bo Hong, MD, FACMG

Presenting Author R. Brian Fedderson, CG(ASCP) CM

Keywords

AI-assisted karyotyping

cytogenetics laboratories

efficiency

accuracy

chromosome analysis

metaphase segmentation

processing steps reduction

bone marrow specimens

whole arm translocations

high-quality services