TY - CHAP
T1 - MixUp-MIL: Novel Data Augmentation for Multiple Instance Learning and a Study on Thyroid Cancer Diagnosis
AU - Gadermayr, Michael
AU - Koller, Lukas
AU - Tschuchnig, Maximilian
AU - Stangassigner, Lea Maria
AU - Erhardt-Kreutzer, Christina
AU - Couillard-Després, Sébastien
AU - Oostingh, Gertie Janneke
AU - Hittmair, Anton
N1 - Lehr-KH Kardinal Schwarzenberg KLinikum
Kreutzer, Couillard-Despres: Spinal Cord Injury and Tissue Regeneration Center Salzburg, Research Institute of Experimental Neuroregeneration, Salzburg, Austria
PY - 2023
Y1 - 2023
N2 - Multiple instance learning is a powerful approach for whole slide image-based diagnosis in the absence of pixel- or patch-level annotations. In spite of the huge size of whole slide images, the number of individual slides is often rather small, leading to a small number of labeled samples. To improve training, we propose and investigate novel data augmentation strategies for multiple instance learning based on the idea of linear and multilinear interpolation of feature vectors within and between individual whole slide images. Based on state-of-the-art multiple instance learning architectures and two thyroid cancer data sets, an exhaustive study was conducted considering a range of common data augmentation strategies. Whereas a strategy based on to the original MixUp approach showed decreases in accuracy, a novel multilinear intra-slide interpolation method led to consistent increases in accuracy.
AB - Multiple instance learning is a powerful approach for whole slide image-based diagnosis in the absence of pixel- or patch-level annotations. In spite of the huge size of whole slide images, the number of individual slides is often rather small, leading to a small number of labeled samples. To improve training, we propose and investigate novel data augmentation strategies for multiple instance learning based on the idea of linear and multilinear interpolation of feature vectors within and between individual whole slide images. Based on state-of-the-art multiple instance learning architectures and two thyroid cancer data sets, an exhaustive study was conducted considering a range of common data augmentation strategies. Whereas a strategy based on to the original MixUp approach showed decreases in accuracy, a novel multilinear intra-slide interpolation method led to consistent increases in accuracy.
U2 - 10.1007/978-3-031-43987-2_46
DO - 10.1007/978-3-031-43987-2_46
M3 - Chapter
SN - 978-3-031-43986-5
T3 - Lecture Notes in Computer Science (LNCS)
SP - 477
EP - 486
BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2023
T2 - 26th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI)
Y2 - 8 October 2023 through 12 October 2023
ER -