UXNet: Searching Multi-level Feature Aggregation for 3D Medical Image Segmentation

TL;DR

UXNet introduces a neural architecture search method that optimizes multi-level feature aggregation strategies in 3D medical image segmentation, significantly enhancing model performance and flexibility with efficient search procedures.

Contribution

It proposes a novel NAS approach that searches both scale-wise feature aggregation and block-wise operators, improving upon traditional UNet architectures for 3D medical imaging.

Findings

Outperforms state-of-the-art models in Dice scores on public benchmarks

Effectively captures boundary details and tiny tissues

Search process is computationally efficient, taking less than 1.5 days on two GPUs

Abstract

Aggregating multi-level feature representation plays a critical role in achieving robust volumetric medical image segmentation, which is important for the auxiliary diagnosis and treatment. Unlike the recent neural architecture search (NAS) methods that typically searched the optimal operators in each network layer, but missed a good strategy to search for feature aggregations, this paper proposes a novel NAS method for 3D medical image segmentation, named UXNet, which searches both the scale-wise feature aggregation strategies as well as the block-wise operators in the encoder-decoder network. UXNet has several appealing benefits. (1) It significantly improves flexibility of the classical UNet architecture, which only aggregates feature representations of encoder and decoder in equivalent resolution. (2) A continuous relaxation of UXNet is carefully designed, enabling its searching scheme performed in an efficient differentiable manner. (3) Extensive experiments demonstrate the effectiveness of UXNet compared with recent NAS methods for medical image segmentation. The architecture discovered by UXNet outperforms existing state-of-the-art models in terms of Dice on several public 3D medical image segmentation benchmarks, especially for the boundary locations and tiny tissues. The searching computational complexity of UXNet is cheap, enabling to search a network with the best performance less than 1.5 days on two TitanXP GPUs.