Fast and reliable medical image enhancement using deep learning
Assoc. Prof. Alan Wang
Auckland Bioengineering Institute
Project code: ABI001
This project is about using deep learning algorithms to enhance the medical images such as dynamic enhanced MRI, arterial spin labeling MRI, and CT perfusion, etc. to reduce the imaging data acquisition time and radiation dosage, but without any loss of diagnosis value.
Ruminal microbial interactions with electro-chemically active bolus
Rumen is a dynamic environment that hosts a large number of specialized microbes that help in the fermentation of feed to generate energy and other useful substrates for the animal to survive. Modification of the feed and/or the microbial diversity has a significant impact on the host and the environment. Consequently, unravelling these mechanisms is of economic and ecological interest. Ruminal microbes use pathways such as the Embden-Meyerhof-Parnas (EMP) for the breakdown of feed. These pathways are not just well conserved across microbial species, but are amongst the most thermodynamically efficient pathways to break down complex molecules to generate ATP and other substrates that sustain life.
Rumen is an anaerobic environment where fermentation is directed by a redox chain. Electro-chemically active bolus can be introduced into the rumen to alter this redox chain—with the explicit aim of altering the cohort of fermentative pathways selected by the microbes and the resulting by-products. However, the efficacy of bolus erodes over time due to biofouling. Factors such as the shape of the bolus, surface- and material-properties determine the extent of biofouling. In this project, we will explore, in-silico, using reactive fluid dynamics methods, the impact of bolus shape and its electro-chemical properties in altering the ruminal contents and the rate of biofouling due to microbial colonization.
An ideal candidate will have a background in analytic/computational chemistry, interest in microbiology and instrumentation design.
A biomechanical assessment of the 'robust strange bird', dinornis robustus, aka the New Zealand moa
Dinornis robustus (literally meaning robust strange bird) was the largest species of New Zealand moa, stood ~2m tall and could reach foliage 3.6m above the ground. The impressive size of these flightless birds leads to some interesting biomechanical questions, which are not easily answered from the fossil record. How did these birds move and what was their body mass relative to their disproportionately wide leg bones? There appears to be profound differences between leg structures in families of moa, resulting in compromised safety factors and running performance. We will investigate these biological questions by constructing a musculoskeletal model of the moa and performing a series of simulations to predict muscle and joint function in various postures of gait. An existing set of CT scans will be used to construct the model using OpenSim modelling software.
Measurement of sarcomere length and length-tension relationship in cerebral palsy muscle
Cerebral palsy is a neuromusculoskeletal disorder that leads to weakened and impaired muscles and lifelong disability. Understanding the nature of the impairment at the muscular level is at the core of understanding the disease and working towards pharmaceutical and therapeutic treatments for the symptoms. In this summer project, a student will work alongside an ABI PhD student who will be acquiring muscle biopsies from the lower limbs of children and adolescents with cerebral palsy. The summer student will be developing and administering an experiment for obtaining the force-length relationship of individual muscle fibers from the muscle biopsies. This will be conducted using a microNewton dynamometer rig and sarcomere length assessment using light microscopy and laser diffraction.