Machine learning, unsupervised learning and stain normalization in digital nephropathology
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3085855Utgivelsesdato
2023Metadata
Vis full innførselSamlinger
- Studentoppgaver (TN-IDE) [823]
Sammendrag
Chronic kidney disease is a serious health challenge and still, the field ofstudy lacks awareness and funding. Improving the efficiency of diagnosingchronic disease is important. Machine learning can be used for various tasksin order to make CKD diagnosis more efficient. If the disease is discoveredquickly it can be possible to reverse changes. In this project, we exploretechniques that can improve clustering of glomeruli images.The current thesis evaluates the effects of applying stain normalization tonephropathological data in order to improve unsupervised learning cluster-ing. A unsupervised learning pipeline was implemented in order to evaluatethe effects of using stain normalization techniques with different referenceimages. The stain normalization techniques that were implemented are:Reinhard stain normalization, Macenko stain normalization and Structurepreserving color normalization. The evaluation of these methods was doneby measuring clustering results from the unsupervised learning pipeline,using the Adjusted Rand Index metric. The results indicate that usingthese techniques will increase the cluster agreement between results andtrue labels for the data. Six reference images were used for each stain nor-malization technique. The average Adjusted Rand Index score for all ref-erence images was increased using all three stain normalization techniques.The best performing method overall was the Reinhard stain normalizationtechnique. This method gave both the highest single experiment and aver-age score. The other normalization methods both have one score close tozero (unsuccessful clustering), and structure preserving color normalizationwould outperform the Reinhard method if this single clustering was moresuccessful. Chronic kidney disease is a serious health challenge and still, the field ofstudy lacks awareness and funding. Improving the efficiency of diagnosingchronic disease is important. Machine learning can be used for various tasksin order to make CKD diagnosis more efficient. If the disease is discoveredquickly it can be possible to reverse changes. In this project, we exploretechniques that can improve clustering of glomeruli images.The current thesis evaluates the effects of applying stain normalization tonephropathological data in order to improve unsupervised learning cluster-ing. A unsupervised learning pipeline was implemented in order to evaluatethe effects of using stain normalization techniques with different referenceimages. The stain normalization techniques that were implemented are:Reinhard stain normalization, Macenko stain normalization and Structurepreserving color normalization. The evaluation of these methods was doneby measuring clustering results from the unsupervised learning pipeline,using the Adjusted Rand Index metric. The results indicate that usingthese techniques will increase the cluster agreement between results andtrue labels for the data. Six reference images were used for each stain nor-malization technique. The average Adjusted Rand Index score for all ref-erence images was increased using all three stain normalization techniques.The best performing method overall was the Reinhard stain normalizationtechnique. This method gave both the highest single experiment and aver-age score. The other normalization methods both have one score close tozero (unsuccessful clustering), and structure preserving color normalizationwould outperform the Reinhard method if this single clustering was moresuccessful.