Histogram-based gravitational optimization algorithm on single MR modality for automatic brain lesion detection and segmentation

•We propose a new fully automatic method to detect and segment brain lesions.•The method is based on enhanced gravitational optimization algorithm.•Ischemic stroke lesions are segmented with 91.5% accuracy and 91.1% sensitivity.•The algorithm is independent of multi-spectral MRI, and local or global registration.•It is practically applicable due to computational efficiency and handling noisy data.

Magnetic resonance imaging (MRI) is a very effective medical imaging technique for the clinical diagnosis and monitoring of neurological disorders. Because of intensity similarities between brain lesions and normal tissues, multispectral MRI modalities are usually applied for brain lesion detection. However, the time and cost restrictions for collecting multi-spectral MRI, and the issue of possible errors from registering multiple MR images necessitate developing an automatic lesion detection approach that can detect lesions using a single anatomical MRI modality. In this paper, an automatic algorithm for brain stroke and tumor lesion detection and segmentation using single-spectral MRI is presented. The proposed algorithm, called histogram-based gravitational optimization algorithm (HGOA), is a novel intensity-based segmentation technique, which applies enhanced gravitational optimization algorithm on histogram analysis results. The mathematical descriptions as well as the convergence criteria of the developed optimization algorithm are presented in detail. Using this algorithm, brain is segmented into different number of regions, which will be labeled as lesion or healthy. Here, the ischemic stroke lesions and tumor lesions are segmented with 91.5% and 88.1% accuracy, respectively.