The Influence of Multiple Loss Functions on MRI Stroke Lesion Area Segmentation

DOI

10.2991/978-94-6463-540-9_88How to use a DOI?

Keywords

Machine Learning; Loss Function; Stroke Lesion Area Segmentation

Abstract

The study solved the imbalanced Magnetic resonance imaging (MRI) dataset problem by choosing different loss functions to achieve a higher stroke lesion area segmentation accuracy. It is helpful for doctors to treat patients efficiently by segmenting the stroke areas quickly with the machine learning model. The study compared focal loss and dice loss based on the same dataset, keeping the model structure and parameters the same, using different evaluation metrics to evaluate the performance of the two loss functions. Then, the study also demonstrated some sample segmentation images to specify the segmentation in detail, helping to understand the result. The study found that the model using focal loss had a better segmentation performance on the imbalanced dataset than the model using dice loss. It was noticed that the focal loss model had clearer boundaries and more precise segmented lesion areas than the dice loss model. That means the focal loss is more suitable for doing the segmentation with small pixels than the dice loss, which would be more useful in medical images, as most of the medical images contain small positive areas and large negative areas. The study could be supportive evidence for future research by providing a strong reason for choosing focal loss as the loss function when training medical image models.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Ruihui Cao
PY  - 2024
DA  - 2024/10/16
TI  - The Influence of Multiple Loss Functions on MRI Stroke Lesion Area Segmentation
BT  - Proceedings of the 2024 2nd International Conference on Image, Algorithms and Artificial Intelligence (ICIAAI 2024)
PB  - Atlantis Press
SP  - 882
EP  - 891
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-540-9_88
DO  - 10.2991/978-94-6463-540-9_88
ID  - Cao2024
ER  -