Sensitivity analysis for image represented by fuzzy function

Hurtík, Petr; Madrid, Nicolás; Dyba, Martin

doi:10.1007/s00500-018-3402-8

Cited by 5 publications

(1 citation statement)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, it is a mechanism of partitioning an image into various parts in such a way that each part has its own area. According to Cheng et al [23], it is a method of partitioning an image I into non-overlapping areas A I : n i=1 A I = I , and A I 1 ∩ A I 2 = ∅, I 1 = I 2 (1) Grayscale image segmentation methods are basically based on partitioning an image by detecting discontinuous gray level values in a particular region [24]. If there is a homogeneity in gray level values of a particular region, then such a partition is done using clustering, thresholding, edge detection, etc [23].…”

Section: Related Work In Image Segmentationmentioning

confidence: 99%

A Hybrid Fuzzy Clustering Approach for the Recognition and Visualization of MRI Images of Parkinson’s Disease

2020

View full text Add to dashboard Cite

Parkinson's disease (PD) is one of the serious diseases in the neurodegenerative disease group, whose early stage pre-diagnosis is still very tedious. Radiologists and medical practitioners mostly depended on the analysis of PD patients' magnetic resonance images (MRIs) to identify this disease. Due to presence of grayscale features and uncertain inherited information in MRIs, their pattern recognition and visualization were very complex. With this motivation, a new method for analyzing and visualizing patterns in MRI images was presented in this study. For this purpose, this study adopted fuzzy information gain (FIG) function and Kmeans clustering algorithm. The FIG function was used to quantify the fuzzified pixels information, whereas K-means clustering algorithm was employed to cluster those fuzzified pixels information. Finally, changes in MRIs were recognized and classified into three distinct regions, viz., the minimum changed region (MINCR), the maximum changed region (MAXCR) and the average changed region (AVGCR). Experimental results were provided by comparing PD patients' segmented MRIs with seven well-known image segmentation methods, including adaptive threshold method, watershed method, gray threshold method, fuzzy based method, K-means clustering algorithm, adaptive K-means clustering algorithm and fuzzy c-means (FCM) algorithm. The proposed method achieved an average mean squared error of 63.49, peak signal-to-noise ratio of 30.14 and Jaccard similarity coefficient of 0.92 among nine MRIs of PD. The performance showed an improvement of 20.73%-32.94%, 3.54%-6.20% and 6.98%-64.29% over the average mean squared error, peak signal-to-noise ratio and Jaccard similarity coefficient, respectively compared to other image segmentation methods.

show abstract