Advancing 3D Medical Image Analysis with Variable Dimension Transform based Supervised 3D Pre-training

TL;DR

This paper introduces a fully-supervised 3D pre-training framework leveraging large-scale 2D natural image data to improve 3D medical image analysis, reducing annotation efforts and enhancing model performance.

Contribution

It proposes a novel 3D network architecture and a supervised pre-training method that utilizes 2D natural images, addressing data scarcity and outperforming existing self-supervised approaches.

Findings

Accelerates convergence in 3D tasks

Improves accuracy across classification, segmentation, detection

Reduces annotation efforts by up to 60%

Abstract

The difficulties in both data acquisition and annotation substantially restrict the sample sizes of training datasets for 3D medical imaging applications. As a result, constructing high-performance 3D convolutional neural networks from scratch remains a difficult task in the absence of a sufficient pre-training parameter. Previous efforts on 3D pre-training have frequently relied on self-supervised approaches, which use either predictive or contrastive learning on unlabeled data to build invariant 3D representations. However, because of the unavailability of large-scale supervision information, obtaining semantically invariant and discriminative representations from these learning frameworks remains problematic. In this paper, we revisit an innovative yet simple fully-supervised 3D network pre-training framework to take advantage of semantic supervisions from large-scale 2D natural image datasets. With a redesigned 3D network architecture, reformulated natural images are used to address the problem of data scarcity and develop powerful 3D representations. Comprehensive experiments on four benchmark datasets demonstrate that the proposed pre-trained models can effectively accelerate convergence while also improving accuracy for a variety of 3D medical imaging tasks such as classification, segmentation and detection. In addition, as compared to training from scratch, it can save up to 60% of annotation efforts. On the NIH DeepLesion dataset, it likewise achieves state-of-the-art detection performance, outperforming earlier self-supervised and fully-supervised pre-training approaches, as well as methods that do training from scratch. To facilitate further development of 3D medical models, our code and pre-trained model weights are publicly available at https://github.com/urmagicsmine/CSPR.