[Paper Review] An Improved Image Mining Technique For Brain Tumour Classification Using Efficient Classifier

P. Rajendran, M. Madheswaran|arXiv (Cornell University)|Jan 12, 2010

Image Retrieval and Classification TechniquesComputer Science26 references22 citations

TL;DR

This paper proposes an improved image mining technique for brain tumor classification using pruned association rule mining with the MARI algorithm, integrating low-level image features and expert knowledge to enhance diagnostic accuracy. The method achieves 96% sensitivity and 93% accuracy in classifying CT brain scans into normal, benign, or malignant categories.

ABSTRACT

An improved image mining technique for brain tumor classification using pruned association rule with MARI algorithm is presented in this paper. The method proposed makes use of association rule mining technique to classify the CT scan brain images into three categories namely normal, benign and malign. It combines the low level features extracted from images and high level knowledge from specialists. The developed algorithm can assist the physicians for efficient classification with multiple keywords per image to improve the accuracy. The experimental result on prediagnosed database of brain images showed 96 percent and 93 percent sensitivity and accuracy respectively.

Motivation & Objective

• To improve the accuracy of brain tumor classification from CT scan images using data mining techniques.
• To integrate low-level image features with high-level expert knowledge for enhanced diagnostic support.
• To develop a system capable of multi-keyword classification per image to improve classification precision.
• To reduce false positives and improve early detection of brain tumors through automated mining.
• To validate the method on a prediagnosed brain image database for real-world applicability.

Proposed method

• The method employs association rule mining with pruning to extract meaningful patterns from CT scan images.
• Low-level features such as texture, intensity, and shape are extracted from brain MRI/CT images.
• The MARI (Mining Association Rules in Image) algorithm is used to generate and refine association rules from the feature set.
• Expert knowledge from radiologists is encoded as high-level rules to guide and validate the mining process.
• The system applies multiple keywords per image to improve classification granularity and reduce ambiguity.
• A hybrid approach combines image features and expert rules to train an efficient classifier for three-class classification: normal, benign, and malignant.

Experimental results

Research questions

• RQ1 Can association rule mining with pruning improve the classification accuracy of brain tumors in CT scans?
• RQ2 How effectively can low-level image features combined with expert knowledge enhance diagnostic precision?
• RQ3 To what extent does multi-keyword classification per image improve sensitivity and specificity?
• RQ4 Can the MARI algorithm efficiently handle complex image data for medical diagnosis?
• RQ5 What is the performance of the proposed system in terms of sensitivity and accuracy on a real prediagnosed dataset?

Key findings

• The proposed method achieved 96% sensitivity in detecting brain tumors from CT scans.
• The system demonstrated 93% overall accuracy in classifying brain images into normal, benign, or malignant categories.
• The integration of low-level image features and expert-derived high-level rules significantly improved classification reliability.
• The use of multiple keywords per image enhanced the system's ability to distinguish subtle tumor characteristics.
• The pruned association rule mining approach reduced irrelevant rules and improved computational efficiency.
• The results were validated on a prediagnosed database, confirming the method's potential for clinical deployment.