Extraction and Evaluation of Corpus Callosum from 2D Brain MRI Slice: A Study with Cuckoo Search Algorithm

Prior research has shown that structural differences in the corpus callosum may correlate with autism spectrum disorder (ASD). However, no prior work had resolved the challenge of accurately segmenting the corpus callosum from 2D brain MRI slices using automated methods. Established knowledge includes the use of thresholding and contouring techniques in image segmentation. This gap motivated the development of a new computer-based diagnosis method. Current approaches often struggle with noise and variability in MRI data. No prior work had combined chaotic cuckoo search with active contour methods for this task. The need for precise segmentation remains unmet in clinical settings. This paper introduces a novel approach to address these limitations.

The study aims to develop a computer-based diagnosis method (CBDM) for delineating and evaluating the corpus callosum from 2D brain MRI slices. The specific problem is the difficulty of accurately segmenting the corpus callosum in noisy and variable MRI data. The motivation stems from the need for reliable diagnostic tools in autism spectrum disorder. The proposed method integrates preprocessing and postprocessing stages. The goal is to improve segmentation accuracy compared to existing techniques. The study also seeks to validate the method using benchmark datasets. The CBDM is designed to extract key anatomical parameters for classification. This approach is intended to support automated diagnosis in clinical applications.

The proposed CBDM includes two main stages: preprocessing and postprocessing. Preprocessing uses a multithreshold technique with chaotic cuckoo search (CCS) algorithm. A preferred threshold procedure is also applied in preprocessing. Postprocessing involves a delineation process to extract the corpus callosum section. The method incorporates Shannon entropy and active contour techniques. The study uses benchmark datasets such as ABIDE and MIDAS for validation. The CBDM is compared against FCM + LS and MLS techniques. Experimental results are analyzed to assess segmentation accuracy and classification performance.

The proposed CBDM achieved higher accuracy in corpus callosum segmentation compared to existing methods. The ABIDE dataset results showed improved performance over FCM + LS and MLS techniques. The method successfully extracted total brain area (TBA) and corpus callosum area (CCA) parameters. The CC section was delineated with greater precision than alternative approaches. The CBDM demonstrated robustness in handling noisy and variable MRI data. Validation on the MIDAS dataset confirmed the method's reliability. Clinical dataset testing further supported the method's effectiveness. The results suggest that the CBDM is a promising tool for automated diagnosis in ASD.

The authors propose that the CBDM offers a reliable approach for corpus callosum segmentation from 2D MRI slices. The method may improve classification accuracy in autism spectrum disorder diagnostics. The CBDM may outperform existing segmentation techniques in terms of precision. The study suggests that the chaotic cuckoo search algorithm enhances segmentation performance. The proposed method may support automated diagnosis in clinical settings. The results suggest that the CBDM is a viable alternative to traditional methods. The authors propose that the method may be useful in future studies on brain structure analysis. The study concludes that the CBDM is a promising tool for medical imaging applications.