A 3D adaptive template matching algorithm for brain tumor detection

作者：Wang Xiao-Fei; Gong Jing; Bu Rui-Rui; Nie Sheng-Dong

来源：Communications in Computer and Information Science, 2014, 461: 50-61.

• Adaptive
• Adaptive template matching
• Brain tumors
• Receiver operating characteristics
• Region of interest
• Similarity coefficients
• Template marching
• Template matching method

This paper presents a three-dimensional adaptive template matching algorithm to detect brain tumors from Magnetic Resonance Images quickly. First, skull and other non-brain tissues were removed by the improved BET algorithm. Then we extracted the structures that contain all small tumors as ROIs (Region of Interest). After that, we screened all the ROIs by the circular degree and other features. Then a three-dimensional template was created conformed to tumor characteristics for each ROI. Finally, the three-dimensional templates were marched with the original images to calculate the similarity coefficient. Then the threshold was determined according to the matching characteristic. After that, the three-dimensional ROI with the similarity coefficient which was higher than the threshold value was marked as the tumor region. To evaluate the performance of the algorithm, 23 clinical cases which contain 124 tumors (3mm-15mm) in different size was used to test the system, using ROC (Receiver Operating Characteristic) curve to analysis the test results. According to the ROC curve, the positive rate reach 88.7097% and the false position rate is 16.03%. Compared to other template matching methods, the algorithm provided has been significantly improved.

• 出版日期2014