Alzheimer's disease (AD) is a neurodegenerative disorder that is progressive and causes loss of memory and cognitive performance, which significantly influences the quality of life of older adults. The accurate and early diagnosis of AD using MRI images provides complex brain structural changes, redundant features, and limited effectiveness of existing automated detection methods. To address this, a novel AD-HCBL is proposed for accurate and automated AD detection using MRI Images. The proposed AD-HCBL integrates a ResNet 50 + Grey-Level Co-occurrence Matrix (GLCM) for extracting the features from the MRI scan images. Frog Snake Prey Predation Relationship Optimisation (FSRO) is then used to extract the features and refine the exploration features to eliminate the redundancy, giving a good balance between exploitation and exploration. The MRI images are then correctly categorised into normal, mild and severe stages of AD, using a BiLSTM network. Two datasets are utilised in the implementation of the experimental results in the python, specifically, the AD Multiclass Dataset and the ADNI dataset. The proposed AD-HCBL achieves above 98% accuracy, precision, and recall when compared to the existing methods, respectively.