Mean Maximum Raw Thresholding Algorithm for Detecting Suspected Regions of Interest of Tumors in Breast Magnetic Resonance Imaging

In this study, the regions of interest of the suspected tumor in the breast magnetic resonance imaging (MRI) are detected using a new automatic global thresholding approach. As a key pre-processing stage alongside the tumor segmentation process, image thresholding is one of the most significant image processing techniques. In order to prepare the image for more precise tumor detection in the following stage of segmentation, it is crucial to identify the areas of any suspected tumor in the breast MRI. The proposed Mean Maximum Raw Thresholding approach (MMRT), which uses the sub-window frames concept to automatically seek the best threshold value that separates the image into two classes-suspected tumor locations and the background-considers the tumor intensity features. On 40 test images from the RIDER breast MRI dataset, the approach is implemented and tested. The results are then analyzed and recorded in relation to the dataset's Ground Truths using two assessment strategies: pixel-based assessment (Jaccard = 0.643 and Dice = 0.683 measures), and quality assessment (PSNR = 69.97 and MSE = 0.011 measures). The evaluation findings demonstrate that the suggested approach outperforms five common thresholding methods that were tried on the same dataset.