DTC: A Deformable Transposed Convolution Module for Medical Image Segmentation

TL;DR

This paper introduces DTC, a deformable transposed convolution module that adaptively learns sampling positions to improve upsampling in medical image segmentation, enhancing detail and structural preservation.

Contribution

It proposes a novel deformable transposed convolution method that learns dynamic sampling locations, improving upsampling quality in 2D and 3D medical image segmentation models.

Findings

DTC improves segmentation accuracy on BTCV15, ISIC18, and BUSI datasets.

DTC enhances feature reconstruction and detail recovery in segmentation models.

The method is effectively integrated into existing architectures, showing consistent performance gains.

Abstract

In medical image segmentation, particularly in UNet-like architectures, upsampling is primarily used to transform smaller feature maps into larger ones, enabling feature fusion between encoder and decoder features and supporting multi-scale prediction. Conventional upsampling methods, such as transposed convolution and linear interpolation, operate on fixed positions: transposed convolution applies kernel elements to predetermined pixel or voxel locations, while linear interpolation assigns values based on fixed coordinates in the original feature map. These fixed-position approaches may fail to capture structural information beyond predefined sampling positions and can lead to artifacts or loss of detail. Inspired by deformable convolutions, we propose a novel upsampling method, Deformable Transposed Convolution (DTC), which learns dynamic coordinates (i.e., sampling positions) to generate high-resolution feature maps for both 2D and 3D medical image segmentation tasks. Experiments on 3D (e.g., BTCV15) and 2D datasets (e.g., ISIC18, BUSI) demonstrate that DTC can be effectively integrated into existing medical image segmentation models, consistently improving the decoder's feature reconstruction and detail recovery capability.