Available project#1 for Master thesis

10-month master thesis project

I have an available 10-month project for master thesis. You will be working with me, Gennady and Joyce. Please contact me if you are interested in it or you have any questions.

Research question:
Methodology-wise: to develop a novel algorithm for analyzing associations between two high-dimensional (e.g., image) domains
Application-wise: to explore the association between hand skeleton (e.g., length of the finger) and facial morphology (e.g., width of the nose), and link it with genetics

Data sets:
Sevearl thousands hand skeleton X-ray images (at 9 and 13 years), 3D facial mesh images (at 9 and 13 years) and whole-genome SNP-based genotype data for each subject, from the Generation R Study.
Sevearl thousands hand skeleton X-ray images (age > 50), 3D facial mesh images (age > 50) and whole-genome SNP-based genotype data for each subject, from the Rotterdam Study.

Datailed Tasks:
This project is divided into 3 phase:
Phase I: To explore the connection between the hand skeleton (e.g., length of the finger) and facial morphology

  • Develop a pipeline to automatically derive measurements (e.g, length of the finger) from x-ray hand skeleton image
  • Derive measurements from the facial mesh image. This part was done already.
  • Perform analysis to calculate the connection between hand measurements and facial measurements.

Phase II: To further explore the grow patterns of the connection, in 9 to 13 years old children data

  • To be explored…

Phase III: If there exist some connections, try to explain the underlying mechanism (e.g., genetic determinants)

  • To be explored…

Please also note that Phase I is enough for a master thesis.

Requirements:
Preferred background in statistics, genetics, epidemiology, computer science, biomedical engineering. Experience with machine/deep learning and image processing is a plus.

Reference:

  1. Genome-wide association of phenotypes based on clustering patterns of hand osteoarthritis identify WNT9A as novel osteoarthritis gene. ANNALS OF THE RHEUMATIC DISEASES
  2. Genome-wide association study identifies nine novel loci for 2D:4D finger ratio, a putative retrospective biomarker of testosterone exposure in utero Hum Mol Genet.
  3. Novel genetic loci affecting facial shape variation in humans elife
  4. Association between prenatal alcohol exposure and children’s facial shape. A prospective population-based cohort study. Human reproduction
  5. Automated Bone Age Assessment based on DXA scans for a variety of ethnicities using Deep Transfer Learning. Master thesis
Have fun!