UNETR++: Delving into Efficient and Accurate 3D Medical Image Segmentation

TL;DR

UNETR++ introduces an efficient 3D medical image segmentation model utilizing a novel paired attention mechanism, achieving state-of-the-art accuracy while significantly reducing computational costs across multiple benchmarks.

Contribution

The paper proposes UNETR++, a novel transformer-based segmentation model with a new efficient paired attention block that reduces complexity and parameters while maintaining high accuracy.

Findings

Sets new state-of-the-art on Synapse with 87.2% Dice score.

Reduces parameters and FLOPs by over 71% compared to previous methods.

Demonstrates high efficiency and accuracy across five benchmark datasets.

Abstract

Owing to the success of transformer models, recent works study their applicability in 3D medical segmentation tasks. Within the transformer models, the self-attention mechanism is one of the main building blocks that strives to capture long-range dependencies. However, the self-attention operation has quadratic complexity which proves to be a computational bottleneck, especially in volumetric medical imaging, where the inputs are 3D with numerous slices. In this paper, we propose a 3D medical image segmentation approach, named UNETR++, that offers both high-quality segmentation masks as well as efficiency in terms of parameters, compute cost, and inference speed. The core of our design is the introduction of a novel efficient paired attention (EPA) block that efficiently learns spatial and channel-wise discriminative features using a pair of inter-dependent branches based on spatial and channel attention. Our spatial attention formulation is efficient having linear complexity with respect to the input sequence length. To enable communication between spatial and channel-focused branches, we share the weights of query and key mapping functions that provide a complimentary benefit (paired attention), while also reducing the overall network parameters. Our extensive evaluations on five benchmarks, Synapse, BTCV, ACDC, BRaTs, and Decathlon-Lung, reveal the effectiveness of our contributions in terms of both efficiency and accuracy. On Synapse, our UNETR++ sets a new state-of-the-art with a Dice Score of 87.2%, while being significantly efficient with a reduction of over 71% in terms of both parameters and FLOPs, compared to the best method in the literature. Code: https://github.com/Amshaker/unetr_plus_plus.