Segmentation of Tiny Intracranial Hemorrhage via Learning-to-Rank Local Feature Enhancement

Shizhan Gong, Yuan Zhong, Yuqi Gong, Nga Yan Chan, Wenao Ma, Calvin Hoi-Kwan Mak, Jill Abrigo, Qi Dou

Abstract:

Intracranial hemorrhage (ICH) is a common head disease that
can result in significant disability or mortality. Segmentation
of ICH is an important yet challenging step for medical image
interpretation. While numerous methods have been proposed
for automatic ICH segmentation, they still suffer from poor
performance when dealing with small-sized hemorrhages. In
this paper, we propose an effective framework to improve the
segmentation accuracy of cases involving tiny hemorrhages.
Our approach integrates a designed learning-to-rank module
within the segmentation network, which can enhance the dis-
crimination capability of local features. This enhancement as-
sists the network in distinguishing patches w/ and w/o hemor-
rhages. Furthermore, we incorporate a hard negative mining
strategy during the pairing process, ensuring the network fo-
cuses on the most ambiguous and challenging cases. We have
conducted experiments on a labelled in-house CT dataset con-
sisting of 267 head CT scans. The results demonstrate that our
method can achieve state-of-the-art performance, particularly
for tiny hemorrhages with a volume size smaller than 1 ml.
Our method has outperformed nnU-Net by a large margin.