Digital-Medicine-Case-presentation-2-YM_1

Environment: Matlab 2020a

Step1: Preprocessing

The dicompreprocess.m contant following steps including

1. Image Sharpening (histeq)
2. Gaussian high-pass filter (HPF.m)
3. Iterative graph binarization (Itersplit.m)
4. Marker Connected Area (masker.m)
5. Masking
6. Image resizing
7. remember to import 'BD2.csv' in this step

Step2: Train the CNN model

Three types of models can be trained respectively by resnet18.m、resnet50.m、resnet101.m

Step3: Random forest

The prediction results of the three models are saved as a csv file, imported into matlab, and the Midium tree is generated using the built-in Classification Learner.

Step4: Generate the validation Data

Use 3 types of resnet to generate prediction results,

1. remember to import 'BD.csv' in dicompreprocess in this step
2. Use the code yfit = predict(MTree.ClassificationTree,input); to get the Final Answer