Cytogenetic testing plays a critical role in the diagnosis and risk stratification of hematologic malignancies. However, conventional techniques are inherently constrained by technical limitations, including low resolution, labor-intensive workflows, and inter-observer variability. Recent advances in artificial intelligence, particularly deep learning-based approaches, have shown promise in addressing these limitations by enhancing image analysis, automating interpretation, and standardizing complex workflows. Many studies have demonstrated that AI-integrated platforms significantly reduce diagnostic turnaround time, detect cryptic or subclonal chromosomal aberrations, and improve interpretive concordance across laboratories. Despite these advantages, barriers such as limited model interpretability, data heterogeneity, and regulatory challenges remain. Rather than replacing human expertise, AI is emerging as a powerful adjunct that strengthens the accuracy and reproducibility of genomic assessments and promotes timely, individualized therapeutic decision-making. As the technology matures, AI is expected to become an integral component of cytogenetic diagnostics, driving a shift toward more efficient, scalable, and precision-guided clinical workflows in hematologic oncology.
Jiun Kang (Thu,) studied this question.