For cancer detection and tissue characterization,
DCE-MRI segmentation and lesion detection is a critical image
analysis task. To segment breast MR images for lesion detection, a
hard-clustering technique with Grammatical Fireworks algorithm
(GFWA) is proposed in this paper. GFWA is a Swarm
Programming (SP) system for automatically generating computer
programs in any language. GFWA is used to create the cluster core
for clustering the breast MR images in this article. The presence of
noise and intensity inhomogeneities in MR images complicates the
segmentation process. As a result, the MR images are denoised at
the start, and strength inhomogeneities are corrected in the
preprocessing stage. The proposed GFWA-based clustering
technique is used to segment the preprocessed MR images. Finally,
from the segmented images, the lesions are removed. The
proposed approach is tested on 5 patients’ 25 DCE-MRI slices. The
proposed method’s experimental findings are compared to those of
the Grammatical Swarm (GS)-based clustering technique and the
K-means algorithm. The proposed method outperforms other
approaches in terms of both quantitative and qualitative results.
Semisupervised learning using denoising autoencoders for brain lesion detection and segmentation
