LORCK: Learnable Object-Resembling Convolution Kernels

TL;DR

This paper introduces learnable hollow convolution kernels that mimic organ contours to improve segmentation accuracy of complex hollow organs like the bladder in medical imaging, outperforming existing models.

Contribution

The paper proposes a novel class of hollow kernels that learn to replicate organ shapes, enhancing segmentation performance in medical imaging tasks.

Findings

Hollow kernels outperform traditional spatial models in segmentation accuracy.

Spatio-temporal models with hollow kernels set new benchmarks for bladder segmentation.

Achieved high dice scores for bladder wall and tumor regions.

Abstract

Segmentation of certain hollow organs, such as the bladder, is especially hard to automate due to their complex geometry, vague intensity gradients in the soft tissues, and a tedious manual process of the data annotation routine. Yet, accurate localization of the walls and the cancer regions in the radiologic images of such organs is an essential step in oncology. To address this issue, we propose a new class of hollow kernels that learn to 'mimic' the contours of the segmented organ, effectively replicating its shape and structural complexity. We train a series of the U-Net-like neural networks using the proposed kernels and demonstrate the superiority of the idea in various spatio-temporal convolution scenarios. Specifically, the dilated hollow-kernel architecture outperforms state-of-the-art spatial segmentation models, whereas the addition of temporal blocks with, e.g., Bi-LSTM, establishes a new multi-class baseline for the bladder segmentation challenge. Our spatio-temporal model based on the hollow kernels reaches the mean dice scores of 0.936, 0.736, and 0.712 for the bladder's inner wall, the outer wall, and the tumor regions, respectively. The results pave the way towards other domain-specific deep learning applications where the shape of the segmented object could be used to form a proper convolution kernel for boosting the segmentation outcome.