Topology and Intersection-Union Constrained Loss Function for Multi-Region Anatomical Segmentation in Ocular Images

TL;DR

This work proposes a new topology and intersection-union constrained loss function (TIU loss) that improves performance using small training datasets and is accurate, and the code along with the trained model is publicly available.

Abstract

Ocular Myasthenia Gravis (OMG) is a rare and challenging disease to detect in its early stages, but symptoms often first appear in the eye muscles, such as drooping eyelids and double vision. Ocular images can be used for early diagnosis by segmenting different regions, such as the sclera, iris, and pupil, which allows for the calculation of area ratios to support accurate medical assessments. However, no publicly available dataset and tools currently exist for this purpose. To address this, we propose a new topology and intersection-union constrained loss function (TIU loss) that improves performance using small training datasets. We conducted experiments on a public dataset consisting of 55 subjects and 2,197 images. Our proposed method outperformed two widely used loss functions across three deep learning networks, achieving a mean Dice score of 83.12% [82.47%, 83.81%] with a 95% bootstrap confidence interval. In a low-percentage training scenario (10% of the training data), our approach showed an 8.32% improvement in Dice score compared to the baseline. Additionally, we evaluated the method in a clinical setting with 47 subjects and 501 images, achieving a Dice score of 64.44% [63.22%, 65.62%]. We did observe some bias when applying the model in clinical settings. These results demonstrate that the proposed method is accurate, and our code along with the trained model is publicly available.

Figures (4)

• Figure 1: Processing flow of the proposed method. This diagram describes the process that the pixels in the predicted image are subject to topological constraints.
• Figure 2: Examples of the ocular images from the public dataset (UBIRIS.v2) for healthy subjects (Latin or Caucasian) and the clinical private dataset for subjects diagnosed with OMG disease (Chinese, Asian).
• Figure 3: Qualitative results of different models trained with 100% of training data, compared with different loss combinations on UBIRIS.v2.
• Figure 4: Qualitative results of UNet trained with 100% of the UBIRIS.v2 dataset using Dice+TIU on Clinical data of the OMG and healthy eye.