I was first introduced to medical data analysis in 2016, when I was taking a course on
Artificial Intelligence and Medicine, through which I learned about machine learning and its
application to medical data processing. Our final project was to classify patients based on
training data collected from their physiological features. First, I built a model using a support
vector machine (SVM). To find the optimal parameter combination, I used a dual loop to
convert the Sigma of Gauss Kernal Function and the Boxconstraint into a more organized
order. Then I drew a 3D map showing the changes in performance, from which I could
analyze the trends to determine the best combination of parameters. Finally, I used ten-fold
cross validation and obtained 84% accuracy in my results, making my final project top in the
Through this course, I also learned that computers are not only capable of performing
complicated medical procedures, but also are extremely accurate—even more so than
doctors—in making sophisticated clinical decisions with the help of machine learning
algorithms. Having witnessed the great potential of applying computational methods to
medicine and healthcare and with my strong aspiration to improve public health, I am
determined to devote my future career to research in this interdisciplinary field.
I was admitted to Shanghai Jiaotong University four years ago after ranking in the top 5 out of
80000 in the national higher-education entrance examination. During my study in the
department of Biomedical Engineering, I took a wide range of courses, such as Physiology,
Medical Image Processing, and Digital Signal Processing, from which I built a solid
background in both biomedical and engineering fields. In addition, I attended the IBM Data
Science Bootcamp in 2017 and won third prize in the final project presentation, further
developing my proficiency in coding and data analysis. I am also a quick learner. Using
online sources, I learned to build a convolutional neural network (CNN) using Caffe in just a
few days. My experiences thus far have put me on track to further develop my passions by
pursuing graduate study in the field of medical image analysis.
Not only did I study in classrooms, but I also participated in several research projects
throughout my academic career. I began working in Dr. Qian Wang’s lab, where I focused on
applying the Deep Embedding Convolutional Neural Network to be able to generate CT
images from MRI images. My project was to understand how the neural network architecture
worked by analyzing feature maps generated from each convolutional layer. Figuring out a
pattern among these feature maps was a difficult challenge, but I managed to overcome it by
testing many methods, such as the Isomap method and measuring the Euclidian distance
between feature maps. Next, I calculated the structural mutual information between each
feature map and the input/output images. I then was able to successfully draw a curve
showing the transformation of feature maps from MRI to CT. This curve further verified that
our deep embedding method can significantly improve the synthesis performance. We have
since submitted our results to the journal of Medical Image Analysis. Currently, my project in
the lab is focused on combining Dense-net and U-net, another challenge that I hope to
overcome as successfully as the first.  This past summer, I visited the department of Biomedical Engineering at University of
Minnesota and performed a research project under Professor David Odde’s supervision. It
was my first time working as a full-time researcher and I enjoyed it very much. My first goal
was to estimate glioma cell volume and shape from image stacks collected by a fluorescence
microscope. This time, the challenge was that the top and bottom layers of the image stack
acquired by the microscope was severely defocused, making it difficult to segment. After
reviewing some literature on the topic, I measured the point spread function (PSF) of the
microscope using fluorescent beads and implemented deconvolution on the image stack
before attempting segmentation. In addition, I developed a software that enabled experts to
segment the image in real time. Finally, I validated my method by measuring the volumeknown
beads and found the result to be remarkably accurate.
Upon finishing the first task, I became curious about blood vessel deformation rates in mouse
brain, since knowing this parameter would allow me to calculate traction force given the
stiffness of a blood vessel. To approach this experiment, I generated kymographs from short
dynamic movies recorded by the microscope. I then wrote a 2D cross-correlation algorithm to
analyze the kymographs and calculated the deformation rate from the results. In the end, I was
extremely proud that my work was able to provide an important reference for future related
research, since I was the first to have made these measurements.
Through my extensive studies and research experiences, I have built a strong background in
medical data analysis and have also developed proficiency in numerical simulation, as well as
experimental studies. I have also learned that there currently exists a plethora of different
magnificent imaging devices, yet there is still great potential for development in the field of
image understanding systems, especially in the medical-imaging area. I strongly believe that
the knowledge I have acquired over the years will lead me to excel in future research relating
to this topic.
Based on my experiences so far, I have developed a clear goal and plan for my future: to
become a researcher in this interdisciplinary field and develop techniques for biomedical
image understanding to tangibly improve clinical practice and overall well-being in the world.
I believe the Ph.D. program in Biomedical Engineering of Yale is an ideal platform for me to
take steps to achieve this goal and expand my horizon. The research on medical image
analysis conducted by Professor James Duncan largely overlaps with my long-lasting
interests and previous research experiences. I sincerely hope that I will be given the
invaluable opportunity to pursue my graduate studies at Yale’s Department of Biomedical
Engineering. Given my background, I believe I am well-positioned to make valuable
contributions to the field.

Post Author: admin


I'm Irvin!

Would you like to get a custom essay? How about receiving a customized one?

Check it out