Experience

Oct 2020 - Present PhD Researcher at Metal Artifacts Reduction for Clinical Data

– Cleaned images from hospitals, selected proper images and
built an in-house dataset to start this study
– Developed an unsupervised deep framework under the CT
imaging principle using the inherent physic domain
consistency to address unavailable ground truth problems in
clinics
– Proposed to evaluate algorithm indirectly by the subsequent
segmentation to prove clinical effectiveness
– Collaborated with other researchers exploring domain
consistency from experiments and published a technical
paper
– Designed a framework specifically eliminating drawbacks of
imbalanced data to increase true positives and false
negatives (hard to trade-off in airway segmentation)
simultaneously
– Analysed tubular airway tree from a topological view and
designed/implemented differentiated distance map and
surface loss functions to improve clinically important
metrics, i.e., the continuity
– Presented my research outcomes at an international
conference
– Applying Point Clouds on Tubules in Medical Image Analysis
– Converted the volumetric leakages to point clouds
segmentation to leverage the sparsity of tubules and
converted the breakage filling to point clouds regression
of tube extension direction and length to overcome the
initial problem of fine-grained shape inconsistency in
point clouds completion
– Designed a deformable module to enhance the invariance of
point clouds in learning
– Proposed directional aggregation operation to overcome
direction information missing in point clouds and improve
the continuity learning to build a complete airway tree
– Developed a pipeline of airway segmentation to achieve good
results for better navigation in the surgery

Sept 2018 - Present Academic Tutor at University of New South Wales

– Conducted machine/deep leaning, computer vision, and data
analysis tutorials in accessible terms, adapting to diverse
student skill levels to ensure a universally satisfactory
learning outcomes
– Communicated to each and every student in the classroom to
encourage participation and to create and maintain an
inviting classroom environment at different ages

Sept 2019 - Apr 2020 M.E. Researcher at Brain Tumour Segmentation

– Designed a deep learning method addressing multi-modality
image merge to utilise data effectively
– Designed a multi-level decoder to enlarge the scale of the
network and to avoid training collapse problems
– Participated in the BraTs Challenge, achieved top-rank
performance, and published two technical papers

Apr 2020 - Nov 2020 Algorithm Engineer at SigmaStar Technology Ltd

– Deployed deep learning algorithms and debugged our
quantisation algorithm (Python) to improve robustness and
tested the performance of algorithms provided by customers
on our chip, providing reports to customers
– Developed the chip simulation (C programming) and conducted
bit-true with IC engineers to verify the correctness of
chip design

Apr 2019 - July 2019 Software Engineer at Cooperation Project from master group

– Packaged the algorithms of our group under the API
guidelines (Python) and deployed algorithms to the standard
platform
– Analysed data source, format, and size et al. in various
clinical environments to make our algorithm work