The purpose of Dementia identification methods-based literature review is to gain a detailed understanding of the methodologies and experiments of existing research works, which helps us to avoid redundancies in the proposed research. This review contains two major objectives: first, the content covering of existing research works, theories and evidences of a specified title; second, the evaluation and their discussion of content. It establishes the familiarity and understanding of existing methodologies. It helps to refine proposed research. It analyzes the advantages and limitations of various research works in Dementia detection. The existing limitations of Dementia identification methods are motivated to propose deep learning based automated pattern identification system for early detection of Dementia from the symptoms of affected patients. Accuracy levels of 93.14% and 94% are achieved via k-means and fuzzy segmentation, respectively. However, the model has a classification accuracy of 81% and 86%, respectively, when using k-means and fuzzy segmentation to classify Alzheimer’s disease, Parkinson’s disease, and frontotemporal dementia in comparison to healthy brains. This leads to the conception of model 2, which is intended to extract brain characteristics for improved categorization.