Lung CT Images - Classification

1 view (last 30 days)
tafa
tafa on 13 Aug 2019
I've prepared some algorithm in Matlab in order to detect lung nodules. I attached simple image results (to detect them I used some 'filters': area greater than some pixels, circularity and eccentricity above/below threshold.
Now I would like to classify these nodules as two classes: tumor and non-tumor. The features about I thought are from 'regionprops' and GLCM (here I am not sure if I use it correctly: for every object I calculate statistic: contrast, correlation, energy, homogeneity + I have changed the value of [1,1] for 1 since I outside of object is only black and I got result ~2k value). So based on example image I got 14 objects and for each was computed GLCM and properties of image like: Area, Circularity, MeanIntensity, MaxFeretDiameter etc.
I assume that these features I should put it one vector feature. And here I want to ask for advice. I am not sure which of machine learning methods would be sufficient for my idea (SVM, kNN, decision tree or maybe use Classification Learner App if it is possible).
I'd be grateful for your answers and any advices which help me to do that!

Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!