STA 141C Proposal
Determination of glaucoma
1. What’s the topic of your project? What question(s) will you attempt to answer or what
problems will you attempt to solve? Why and to whom are these meaningful?
Our project is about glaucoma. We will analyze some data about glaucoma to build a model that
can help us predict if a patient has glaucoma. Glaucoma is a group of eye conditions that
damage the optic nerve, the health of which is vital for good vision. Many forms of glaucoma
have no warning signs. The effect is so gradual that people may not notice a change in vision
until the condition is at an advanced stage. We hope our data analysis can help people better
understand glaucom, raise awareness of doing regular eye exams, and, hopefully, can detect
any abnormalities early.
2. What data source(s) will your team use? Briefly describe each data source and Provide
a link for each data source. This is a check to make sure that there is actually data
available for your topic. If you ultimately decide not to use some of the data sources, or
find additional data sources later, that’s okay.
We will use two data sources for our project. One dataset comes from Kaggle, and another
dataset from github.
The dataset from Kaggle contains 650 images with 4 columns: ExpCDR, Eye, Set, Glaucoma.
https://www.kaggle.com/datasets/sshikamaru/glaucoma-detection
The dataset from Github will be our secondary dataset contains 699 populations with 10
variables:
clump_thickness,unif_cell_size,unif_cell_shape,marg_adhesion,single_epith_cell_size,bare_nuc
lei,bland_chrom,norm_nucleoli,mitoses,classes.
https://github.com/Preethijazzy16/Glaucoma_Detection/blob/main/Glaucoma.csv
3. What statistical methods will your team use? Make sure the method you choose is not
too complicated and you are capable of writing the code .
We will use lasso regression to reduce the image’s noise then using some ML algorithm to
classify if this image has glaucoma or not.
4. What makes your project challenging?
For this project we are considering the image of eyes as our population sample. We
have 650 images in total. The most challenging part is that we will apply CNN and the
ResNet34 model. In addition, we will make our own lasso regression.
5. Reference
https://www.mayoclinic.org/diseases-conditions/glaucoma/symptoms-causes/syc-203728
39