Iraqi Journal of Intelligent Computing and Informatics

Digital communication has witnessed remarkable growth since the advent of the internet. However, security concerns can be attributed to the inherent openness of the existing internet environment. Steganography is an effective technique for hiding sensitive information within a digital medium. The proposed research presents an enhanced spatial domain image steganography approach developed using the least significant bit (LSB) technique. This approach is further enhanced using two meta-heuristics: the genetic algorithm (GA) and Tabu search (TS). However, no existing approach has simultaneously employed GA and TS optimisers. Therefore, the proposed approach is a unique contribution to the field of image steganography. In this approach, GA is utilised as an optimiser to facilitate the selection of the most suitable image pixels for data embedding. The optimisation process helps minimise image distortion. The method is further improved using the TS optimiser, thus achieving the least image distortion. Results indicate the effectiveness of the proposed approach in image steganography. Compared with the conventional LSB steganography technique, the proposed approach achieves a 55.120% improvement in peak signal-to-noise ratio and a 0.26% improvement in mean squared error. Thus, the proposed approach represents a remarkable contribution to digital communication, demonstrating its effectiveness in protecting confidential information. Overall, the proposed approach can be further enhanced and explored for applications in other fields.

Globally, brain tumours are a leading cause of death. Depending on whether they are malignant or benign, tumours can interfere with neurological and body functions as well as cause severe health problems. Carrying out the diagnosis of brain tumours through the conventional methods of test, such as magnetic resonance imaging (MRI) scan and clinical examination, would be cumbersome and time consuming in most situations. With the latest machine learning hype, deep learning emerged as the medical image analysis game-changer. Early, precise identification of brain tumours has been made possible due to the remarkable performance of convolutional neural networks (CNNs) in extracting fine details from medical images. Support vector machines (SVMs) and CNNs can be combined to improve classification accuracy. The goal of this study is to achieve an improved model for classifying brain tumours using MRI images. This model extracts complex features using a CNN whilst improving classification with an SVM. The resulting model can enhance and speed up medical diagnosis. With a 99.2% training set success rate and a 96.1% test set success rate, the hybrid method is extremely accurate.