Abstract:

Abstract Data mining entails discovery of beneficial patterns from an extraordinary number of large patterns. The discovered patterns are categorized by applying appropriate metrics known as the interestingness measures. Data mining is a highly effective technique that finds extensive application in various domains, including healthcare. One of its significant uses cases in healthcare is the identification of health services sequences from massive medical datasets, facilitation of decision-making, and provide early-stage treatment to patients. In the field of medical imaging, a plethora of computer-aided diagnosis (CAD) systems have been introduced to aid radiologists in their patient utilized. A head tumour is a pathological condition characterized by the anomalous proliferation of brain cells, and is considered a significant contributor to mortality rates in the population. A computerized system for detecting brain tumours enables early stage diagnosis. The present study introduces an intelligent system designed for the purpose of diagnosing and classifying brain tumour disease, as well as providing the user with a comprehensive description of the disease and offering advice for maintaining a healthy lifestyle. The methodology employed involves the utilization of data mining techniques on a health care dataset. The focus of the proposed system is primarily on the diagnosis of tumours in the brain through the utilization of Computerised Tomography (CT) images of the brain. This study provides an additional approach for neuroradiologists to readily discern neoplastic cells from cerebral images. The proposed work incorporates a crucial data mining concept, namely, the pre-processing of CT scan brain images. The proposed system is composed of four distinct phases, namely pre-processing, segmentation, feature extraction, and classification. This paper presents a proposed algorithm for image segmentation utilising shape priors, and an association rule-based classification approach using interestingness measures utilizing the image apriori algorithm. The algorithm proposed was evaluated and attained a success rate of 99%. The outcome is evaluated in comparison to other extant mining algorithms.

Keywords: