Active, Continual Fine Tuning of Convolutional Neural Networks for Reducing Annotation Efforts

TL;DR

This paper introduces a combined active and transfer learning framework for CNNs that significantly reduces annotation efforts in medical imaging by selectively annotating the most informative samples and continually fine-tuning the model.

Contribution

It presents a novel integrated approach that merges active learning and transfer learning for efficient CNN fine-tuning in medical imaging tasks.

Findings

Reduces annotation efforts by at least 50% compared to random sampling.

Effective across three different medical imaging applications.

Demonstrates the practicality of combined active and transfer learning in medical domains.

Abstract

The splendid success of convolutional neural networks (CNNs) in computer vision is largely attributable to the availability of massive annotated datasets, such as ImageNet and Places. However, in medical imaging, it is challenging to create such large annotated datasets, as annotating medical images is not only tedious, laborious, and time consuming, but it also demands costly, specialty-oriented skills, which are not easily accessible. To dramatically reduce annotation cost, this paper presents a novel method to naturally integrate active learning and transfer learning (fine-tuning) into a single framework, which starts directly with a pre-trained CNN to seek "worthy" samples for annotation and gradually enhances the (fine-tuned) CNN via continual fine-tuning. We have evaluated our method using three distinct medical imaging applications, demonstrating that it can reduce annotation efforts by at least half compared with random selection.