In advanced countries throughout the world, the population of Alzheimer's Disease(AD) patients has been gradually increasing with the aging of the society. As a result, it has become an important research topic how to diagnose AD early and give necessary treatment and training to AD patients, especially those with mild cognitive impairment(MCI), whose executive functions such as response inhibition, cognitive flexibility, attention switching and planning may display evident disorder and impairment. Unlike traditional paper tests and subjective assessments by the patient's relatives, this study adopts virtual reality(VR) technology to develop a novel diagnosis & assessment system, which uses head mounted display(HMD), game technology and sensors to generate an interactive and panoramic scenario - a virtual convenience store - for assessment of executive functions and memory. A variety of tasks of multi-layered difficulty-level hierarchy, such as memorizing a shopping list, looking for certain goods, and checking out, has been designed for customized and adaptive assessment, training, and treatment of MD. In the meantime, the study also records test-takers' performance data (including path and central-vision movement) in the process of all tasks for the development of a novel diagnosis & assessment method. Moreover, test-takers' technology acceptance is measured for assessing the elderly's subjective perception of new technology and discussing the topic of human-machine interaction. In the study, tests on 2 healthy adults have been completed, the system's functionality has been preliminarily verified, and test-takers' subjective perception of the system has been investigated.