Medical images can have extremely high resolutions which cannot be handled properly by
typical stateofart machine learning models. In this thesis, I compared the performance of two approaches of multiple instance learning models where the high resolution images are downscaled into smaller patches and low dimensional embedding are calculated using Resnet. Then low dimensional embedding are aggregated using multiple instance learning to attain class labels. The data set for this thesis consisted of high resolution histological slides of human lung which were classified to contain cancer or not.
Anotace v angličtině
Medical images can have extremely high resolutions which cannot be handled properly by
typical stateofart machine learning models. In this thesis, I compared the performance of two approaches of multiple instance learning models where the high resolution images are downscaled into smaller patches and low dimensional embedding are calculated using Resnet. Then low dimensional embedding are aggregated using multiple instance learning to attain class labels. The data set for this thesis consisted of high resolution histological slides of human lung which were classified to contain cancer or not.
Klíčová slova
CLAM, Multiple instance learning, Lung data analysis
Klíčová slova v angličtině
CLAM, Multiple instance learning, Lung data analysis
Rozsah průvodní práce
37
Jazyk
AN
Anotace
Medical images can have extremely high resolutions which cannot be handled properly by
typical stateofart machine learning models. In this thesis, I compared the performance of two approaches of multiple instance learning models where the high resolution images are downscaled into smaller patches and low dimensional embedding are calculated using Resnet. Then low dimensional embedding are aggregated using multiple instance learning to attain class labels. The data set for this thesis consisted of high resolution histological slides of human lung which were classified to contain cancer or not.
Anotace v angličtině
Medical images can have extremely high resolutions which cannot be handled properly by
typical stateofart machine learning models. In this thesis, I compared the performance of two approaches of multiple instance learning models where the high resolution images are downscaled into smaller patches and low dimensional embedding are calculated using Resnet. Then low dimensional embedding are aggregated using multiple instance learning to attain class labels. The data set for this thesis consisted of high resolution histological slides of human lung which were classified to contain cancer or not.
Klíčová slova
CLAM, Multiple instance learning, Lung data analysis
Klíčová slova v angličtině
CLAM, Multiple instance learning, Lung data analysis