1. Radiomics
1) predicting binomial prognosis of glioblastoma with MRI image
MRI T1, T2 image -> use 1 axial image penetrating the center of tumor in each patient image
data: 70 patient data consisting of clinical information and DICOM image
-> segmentation of tumor using ImageJ
-> define and calculate numercial features using radiomics package in R
-> train and test ml model (Ex: Random Forest) - 0/1=f (68 features)
* we can select principal components as an input of ml model! (use pca)
** process images which are took with different protocol? -> can have different information
** fixed standard for segmentation or feature definement?
2) ml model for tumor segmentation
CNN model, encoder-decoder model => image -> 0/1 values in 2*2 image matrix
3) CNN+RNN image captioning model
CXR image -> analyze with CNN -> build description with RNN -> caption
2. Reinforcement Learning
1) use SMART system to model personal treatment plan
train model with several treatment cases of patients -> (patient state, longitudinal treatment plan, prognosis)
-> how should we make treatment plan to maximize his personal future prognosis?
-> use SMART system for decision making in each timestep