Unsupervised Pre-training Across Image Domains Improves Lung Tissue Classification

By Thomas Schlegl, Joachim Ofner, Georg Langs. 2014

69 citations recorded by the source corpus.

Subfields: Pulmonary and Respiratory Medicine, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Pulmonary and Respiratory Medicine

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Variational Information Distillation for Knowledge Transfer (2019, background): “For example, in many medical image applications [24], the amount of available data is constrained by the number of patients of a particular disease.”
Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning (2016, methodology): “There are currently three major techniques that successfully employ CNNs to medical image classification: 1) training the "CNN from scratch" [13] - [17] ; 2) using "off-the-shelf CNN" features (without retraining the CNN) as complementary information channels to existing hand-crafted image features,”
Multi-modal Self-Supervision from Generalized Data Transformations (2020, methodology): “Using pretrained visual feature representations, such as those obtained by training on ImageNet LSVRC-12 and finetuning these on specific medical imaging tasks has been very successful [63, 90, 108].”
Fine-tuning Convolutional Neural Networks for Biomedical Image Analysis: Actively and Incrementally (2017, background): “[19] explored unsupervised pre-training of CNNs to inject information from sites or image classes for which no annotations were available, and showed that such across site pre-training improved classification accuracy compared to random initialization of the model parameters.”
Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning? (2016, background): “[31]–[33] wherein the entire pre-trained CNN underwent”
Deep learning in medical imaging and radiation therapy. (2018, unknown): “Schlegl et al. 205 trained a CNN in an unsupervised manner for learning more general low-level image features for images from multiple sites (as domains).”
Not‐so‐supervised: A survey of semi‐supervised, multi‐instance, and transfer learning in medical image analysis (2018, methodology): “Probably the first work to do this is Schlegl et al. [2014]. For the target task of classifying tissue types in chest CT slices, they used three different source tasks: natural images, other chest CT images, and head CT images.”
Deep Learning and Convolutional Neural Networks for Medical Image Computing (2017, methodology): “In this case, the application of these modern networks to the medical imaging domain might especially benefit from pre-training [18] (Fig.”