Automated System for Chromosome Karyotyping

Chromosome karyotyping, crucial for identifying genetic abnormalities, remains largely manual and time-intensive. This paper introduces a novel approach that automates chromosome segmentation and classification using deep learning techniques. Initially, various image-processing techniques for pre-processing, segmentation, and feature extraction were explored. The study aims to compare the segmentation results obtained from traditional image processing methods with those from the deep learning model, specifically focusing on the YOLOv8 framework applied to the annotated metaphase images. Our findings demonstrate significant improvements in the segmentation of single, touching, and overlapping chromosomes, indicating that the deep learning approach outperforms conventional techniques in both accuracy and efficiency. This research lays the groundwork for a more streamlined karyotyping process, enhancing the capabilities of cytogeneticists in genetic analysis and diagnostics.