Chromosome Abnormality Detection Using Visual Geometric Transformer and Mantis Search Optimization

Chromosomes, which carry vital genetic material, have a distinctive thread-like appearance located within the cell nucleus. The process of examining these structures known as karyotyping is fundamental for identifying genetic abnormalities. Although several techniques have been developed for this purpose, many existing methods are limited by inefficiencies, particularly in terms of processing time and accurate feature extraction. To overcome these issues, this study introduces a novel algorithm called Visual Geometric Transformer-based Mantis Search (VGT-MS) for effective detection of chromosomal anomalies. Given that chromosome images often include irrelevant background elements, a preprocessing step is applied to eliminate these artifacts. Feature extraction is performed using the VGG-16 network, followed by classification using the Vision Transformer to pinpoint abnormalities. To further enhance the model's effectiveness, its parameters are optimized using the Mantis Search Algorithm. The performance of the proposed framework is assessed using evaluation metrics including accuracy, F1-score, recall, precision, and ROC. The experimental results indicate that the proposed model excels in all key metrics, achieving an accuracy of 98.0%, precision of 97.2%, recall of 96.2%, and an F1-score of 96.7%, all while reducing computational overhead. Overall, the VGT-MS framework proves to be a powerful and efficient solution for chromosome abnormality detection, successfully addressing the drawbacks of conventional methods.