Transfer Learning for Image Segmentation by Combining Image Weighting and Kernel Learning

Abstract

Many medical image segmentation methods are based on supervised classification of voxels. Such methods generally perform well when provided with a training set that is representative of the test images to segment. However, problems may arise when training and test data follow different distributions, for example due to differences in scanners, scanning protocols, or patient groups. Under such conditions, weighting training images according to distribution similarity has been shown to greatly improve performance. However, this assumes that part of the training data is representative of the test data; it does not make unrepresentative data more similar. We therefore investigate kernel learning as a way to reduce differences between training and test data and explore the added value of kernel learning for image weighting. We also propose a new image weighting method that minimizes maximum mean discrepancy (MMD) between training and test data, which enables the joint optimization of image weights and kernel. Experiments on brain tissue, white matter lesion, and hippocampus segmentation show that both kernel learning and image weighting, when used separately, greatly improve performance on heterogeneous data. Here, MMD weighting obtains similar performance to previously proposed image weighting methods. Combining image weighting and kernel learning, optimized either individually or jointly, can give a small additional improvement in performance.

DMCNN: A Deep Multiscale Convolutional Neural Network Model for Medical Image Segmentation

“…We also conduct comparison experiment with state-ofthe-art segment methods including TLWK [31], MNF [32], and SUSAN [33] on our medical data. e results are given in Table 5.…”

Point Cloud Denoising with Principal Component Analysis and a Novel Bilateral Filter

“…The normal vector is a fundamental attribute of point cloud. The accuracy of the normal vector directly affects the application effect of point cloud data in reverse engineering, and the rendering and processing of point cloud in many other areas, such as denoising, segmentation, data reduction, and surface reconstruction [11][12][13][14]. For surfaces with sharp features, the details of the surfaces are easily lost in point cloud processing, if the normal vector of the feature region, i.e.…”

Segmentasi Warna Citra Bunga Daisy dengan Algoritma K-Means pada Ruang Warna Lab

“…Segmentasi citra itu sendiri dapat dihasilkan dengan menggunakan kaidah statistik seperti korelasi [9] dan distribusi [10]. Selain itu, segmentasi dapat dilakukan dengan memanfaatkan struktur morpologi citra [11], pengelompokan hirarkie [12] ataupun menggunakan deep learning [13]. Sementara itu, segmentasi citra memiliki beberapa kendala, dimulai dari kesulitan dalam mendapatkan data ground truth dan menghasilkan segmentasi yang akurat seperti ganguan derau [14].…”