Scientists from Gdańsk Tech Use AI to Detect Chromosome Abnormalities
Scientists from Gdańsk Tech and Chengdu University of Information Technology have developed the CS-Net platform, which uses AI to detect abnormalities in chromosome structures. In the future, the solution may support karyotyping and clinical diagnostics.
Scientists from Gdańsk Tech and Chengdu University of Information Technology have developed the CS-Net platform, which uses AI to detect abnormalities in chromosome structures. In the future, the solution may support karyotyping and clinical diagnostics.
Karyotyping allows for the assessment of the number and structure of chromosomes in a patient's cells. This enables the detection of changes that may indicate genetic abnormalities.
The CS-Net project was developed by a team led by Prof. Edward Szczerbicki from the Faculty of Management and Economics at Gdańsk Tech and Prof. Haoxi Zhang from Chengdu University of Information Technology in China. Prof. Zhang has been engaged in the use of AI in DNA research for several years.
– We want CS-Net to be able to detect an increasing number of abnormalities in genetic data obtained from real patients. One of the next steps will also be to incorporate large language models. Artificial intelligence has great potential in this field, and this is the direction we want to continue developing – says Prof. Edward Szczerbicki.
AI in Chromosome Analysis
CS-Net uses machine learning and computer vision methods to identify changes in chromosome structure. It is designed to support karyotype analysis and help detect abnormalities that may be relevant for diagnosis.
According to the authors, the solution could be used not only in further research on karyotyping, but also more broadly in biomedical research and healthcare.
The results have been presented in the article “CS-Net: Bridging Local and Temporal Features for Chromosomal Structural Abnormality Detection”. The paper will be published in IEEE Journal of Biomedical and Health Informatics and is already available in Early Access.
The article is available on the IEEE Xplore platform.