Towards a Fully-Automated Karyotype Analysis for Detecting Chromosomal Abnormality via Intelligent Bioinformatics and Image Analysis

Professor Lam, Tak Wah   (Principal Investigator (PI))

Dr Lo Fai Man   (Co-Investigator)

Dr Luk Ho-Ming   (Co-Investigator)

Professor Luo Ruibang   (Co-Investigator)

Professor Wong Kenneth Kwan Yee   (Co-Investigator)

22

2021-09-01

2023-06-30

737000

Towards a Fully-Automated Karyotype Analysis for Detecting Chromosomal Abnormality via Intelligent Bioinformatics and Image Analysis

Analysis, Detecting Chromosomal Abnormality, Fully-Automated, Image, Intelligent Bioinformatics, Karyotype

Others - Medicine, Dentistry and Health

Engineering (E)

PiH/307/21

Research Talent Hub for ITF Projects (RTH-ITF)

2021

Completed

Karyotyping is the conventional gold standard for diagnosing chromosomal abnormalities for genetic disorders. Yet karyotype analysis is manpower-intensive; each case on average takes an experienced cytogeneticist 1.5 days. This project extends our collaboration with the Department of Health's Clinical Genetic Service (CGS) to develop the first end-to-end, intelligent karyotype analysis system. Background: Existing karyotype analysis algorithms/software have limited usage at clinical level. Current research focuses on image segmentation and classification. There is no attempt, from a bioinformatics perspective, to interpret images for structural abnormalities. Furthermore, clinical-level images (using high-banding-resolution chromosomes) create another level of challenge for segmenting intricately overlapping chromosomes, for which existing algorithms are unable to handle. R&D: We'll exploit CGS's 100,000 karyotype images and NGS bioinformatics to build a reference model for high-banding-resolution chromosomes and more importantly, the first heterozygous-abnormality model. We'll develop novel chromosome-sensitive deep- learning solutions for segmenting intricate overlaps and detecting structural abnormalities from random distortion, aiming to analyze most cases end-to-end.Impact: Karyotyping has a high demand in China. Our fully-automated system will allow optimization for individual laboratory setup and help cytogeneticists shorten the analysis time from 1.5 days to 2-3 hours. This addresses the shortage of experienced cytogeneticists and makes karyotyping more available.