R. Sowmya Devi1, Dr. L.Padma Suresh2, Dr.K.L.Shunmuganathan

Intelligent Fussy System Based Dermoscopic ImageSegmentation for Melanoma Detection

R. Sowmya Devi, Dr. L. Padma Suresh, Dr. K. L. Shunmuganathan.

Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System

(SEISCON 2011)

INTRODUCTION• Malignant melanoma is the most frequent type of skin

cancer and its incidence has been rapidly increasing over the last few decades.

• Nevertheless, it is also the most treatable kind of skin cancer, if diagnosed at an early stage.

• The clinical diagnosis of melanoma is commonly based on the ABCD rule , an analysis of four parameters (asymmetry, border irregularity, color, and dimension), or the 7-points checklist which is a scoring method for a set of different characteristics depending on color, shape and texture.

• Dermoscopic images have great potential in the early diagnosis of malignant melanoma, but their interpretation is time consuming and subjective, even for trained dermatologists.

• So, we prefer automatic dermoscopic image analysis.• The standard approach in automatic dermoscopic image

analysis has usually three stages: 1) image segmentation; 2) feature extraction and feature selection; and 3)lesion classification.

• To address this problem, several algorithms have been proposed.

• In this paper we propose and evaluate several Fuzzy based clustering techniques: Fuzzy C Means Algorithm (FCM), Possibilistic C Means Algorithm.

Intelligent Fuzzy Clustering Techniques

• The fuzzy intelligent system is a branch of computer science concerned with making computers behave like humans .

• Cluster analysis is a technique for classifying data, i.e., to divide a given dataset into a set of classes or clusters.

• Clustering is the process of identifying natural groupings or clusters within unlabelled data based on some similarity measure.

How could we know what constitutes “different” clusters?

• Green Apple and Banana Example. • Two features: shape and color.

Clusters Example

What is Intelligent Fuzzy Cluster analysis?

• In classical cluster analysis each datum must be assigned to exactly one cluster.

• The intelligent Fuzzy cluster analysis relaxes this requirement by allowing gradual memberships.

• The general philosophy of clustering is to divide the initial set into homogeneous groups and to reduce the data.

Clustering methods

• Clustering methods can be of two types: Crisp and Fuzzy clustering.

• Crisp clustering assigns each data to a single cluster.

• In fuzzy the membership function measures the degree of belonging of each feature in a cluster.

Degrees of membership

• The degrees of membership to which a given data point belongs to the different clusters are computed from the distances of the data point to the cluster centers.

• The closer a data point lies to the center of a cluster (i.e. size and shape), the higher is its degree of membership to this cluster.

Distance functions

Euclidean distance function :

Mahalanobis distance function:

Fuzzy C means algorithm

• Fuzzy c-means (FCM) is a method of clustering which allows one piece of data to belong to two or more clusters.

• The FCM algorithm receives the data or sample space in matrix format.

• The number of clusters C, the assumption partition matrix U, convergence value E all must be given to the algorithm.

Flow of algorithm• The first step is to calculate the cluster centers• The cluster centroid Vi for each cluster.

Where Vi = ith cluster center Uij = membership of jth data point to ith cluster center Xj = jth data point

• The second step is to calculate the distance matrix d. The distance matrix constitutes the Euclidean distance between every pixel and every cluster center.

• The distance is simply given as the difference of the magnitudes of the data point and the cluster center.

• The third step is to find the final partition matrix by using memberships assigned to each data point.

• Memberships are assigned in such a way that the lower distant ponts are assighned with higher membership and vice-versa

Fuzzy C means without noise

• After obtaining the final partition matrix the each data point in that partition matrix is

compared with the threshold value.• So the final matrix with elements of 0 and 1 is

obtained.• The final matrix is taken and used to

reconstruct the image.

Fuzzy C means with noise

Possibillistic C means Algorithm

• The normalization of memberships in FCM, can lead to undesired effects in the presence of noise and outliers.

• In PCM one tries to achieve a more intuitive assignment of degrees of membership by dropping the probability constraint of FCM, which is responsible for the undesirable effect.

Possibillistic C-Means Clustering • Here we assume

&

• Constraint 1 guarantees that no cluster is empty

• The [0,1] interpreted as the degree of representativity 𝑢𝑖𝑗∈or typicality of the datum to cluster г .𝑥𝑗 𝑖

• However, this leads to the mathematical problem that the objective function is now minimized by assigning uij = 0 for all i € {1,……c} and j € {1,…..,n}.

• In order to avoid this trivial solution, a penalty term is introduced, which forces the membership degrees away from zero.

• That is, the objective function J is modified as

dij is the distance between the jth data and the ith cluster,μij is the degree of belonging of the jth data to the ith cluster,m is the degree of fuzziness ,ηi is a suitable positive number andc is the number of clusters andN is the number of datas

• μij can be obtained as

• The value of ηi determines the distance at which membership value of a point in a cluster becomes 0.5. It is obtained as

• The value of ηi can be fixed or changed in

each iteration by varying dij and μij.• This method is more Robust in the

presence of Noise in finding valid clusters and giving robust estimation of the centers.

• There is a more difference between FCM and PCM in terms of clustering.

Variation of PCM with ‘m’

FCM & PCM differences

The interpretation of m is different in the FCM and the PCM. In the FCM, increasing values of m represent increased sharing of points among all clusters, whereas in the PCM, it represent increased possibility of all points in the data set completely belonging to a given cluster. Thus, the value of m that gives us satisfactory performance is different in the two algorithms

Results

Results