article · Journal of Imaging
Automating brain tumour segmentation is challenging due to the significant variations found in tumours. A new automated framework combines a deep capsule network, known as CapsNet, with a latent-dynamic conditional random field, or LDCRF. The overall pipeline operates across three stages: pre-processing, segmentation, and post-processing. Pre-processing uses N4ITK bias field correction on magnetic resonance images before normalising signal intensity. For segmentation, CapsNet is first trained on image patches, followed by training the combined LDCRF-CapsNet model on axial view image slices once initial parameters are established. Finally, the post-processing phase uses thresholding to refine pixel labelling and eliminate small three-dimensional connected regions. Evaluated on the BRATS 2015 and BRATS 2021 benchmark datasets, the combined technique demonstrates competitive performance and outperforms existing state-of-the-art approaches under comparable conditions.
Brain tumours vary substantially in shape, size, and appearance, making accurate computerised identification difficult. By improving automated segmentation through advanced neural network architectures, medical imaging tools can more reliably isolate tumour regions. This helps provide consistent, objective measurements from scans, supporting clinicians who analyse complex magnetic resonance imaging datasets.
This technology offers an algorithmic pipeline relevant to developers of diagnostic medical imaging software and clinical decision-support systems. The abstract demonstrates validation on standard academic benchmark datasets, specifically BRATS 2015 and BRATS 2021. Because testing is limited to retrospective benchmark datasets without clinical workflow integration or prospective trials mentioned, the approach remains at an applied research stage rather than ready for immediate clinical deployment.
AI-generated from the published abstract. Always read the original work before citing.
Because of the large variabilities in brain tumors, automating segmentation remains a difficult task. We propose an automated method to segment brain tumors by integrating the deep capsule network (CapsNet) and the latent-dynamic condition random field (LDCRF). The method consists of three main processes to segment the brain tumor-pre-processing, segmentation, and post-processing. In pre-processing, the N4ITK process involves correcting each MR image's bias field before normalizing the intensity. After that, image patches are used to train CapsNet during the segmentation process. Then, with the CapsNet parameters determined, we employ image slices from an axial view to learn the LDCRF-CapsNet. Finally, we use a simple thresholding method to correct the labels of some pixels and remove small 3D-connected regions from the segmentation outcomes. On the BRATS 2015 and BRATS 2021 datasets, we trained and evaluated our method and discovered that it outperforms and can compete with state-of-the-art methods in comparable conditions.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.3390/jimaging8070190
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
