Home
Research Team
Projects
Publications
Contact Us
 

Skin Lesion Segmentation

  1. Eye-gaze Driven Interactive Image Segmentation

  2. Skin Lesion Segmentation using the Random Walker Method

     

 1)Eye-gaze Driven Interactive Image Segmentation

 We developed a hands-free interactive image segmentation using an eye-gaze tracking system.

Abstract: This paper explores a novel approach to interactive user-guided image segmentation, using eyegaze information as an input. The method includes three steps: 1) eyegaze tracking for providing user input, such as setting object and background seed pixel selection; 2) an optimization method for image labeling that is constrained or affected by user input; and 3) linking the two previous steps via a graphical user interface for displaying the images and other controls to the user and for providing real-time visual feedback of eyegaze and seed locations, thus enabling the interactive segmentation procedure. We developed a new graphical user interface supported by an eyegaze tracking monitor to capture the user's eyegaze movement and fixations (as opposed to traditional mouse moving and clicking). The user simply looks at different parts of the screen to select which image to segment, to perform foreground and background seed placement and to set optional segmentation parameters.

There is an eyegaze-controlled "zoom" feature for difficult images containing objects with narrow parts, holes or weak boundaries. The image is then segmented using the random walker image segmentation method. We performed a pilot study with 7 subjects who segmented synthetic, natural and real medical images. Our results show that getting used the new interface takes about only 5 minutes. Compared with traditional mouse-based control, the new eyegaze approach provided a 18.6% speed improvement for more than 90% of images with high object-background contrast. However, for low contrast and more difficult images it took longer to place seeds using the eyegaze-based "zoom" to relax the required eyegaze accuracy of seed placement.

Random Walker Segmentation :

  • The user species seed points and the seeds are labeled as either object or background;
  • The image is modeled as a graph where image pixels are represented by graph nodes;
  •  Graph edges connect neighboring pixels;
  • The weight of an edge is set as a function of the intensity difference between a pixel and its neighbor. This function or mapping is controlled by a parameter beta;
  • The probability that a random walker starting from a particular pixel reaching any of the labeled pixels (seeds) is computed for every pixel by solving a linear system of equations;
  • The maximum probability label is assigned to each pixel, which constitutes the image segmentation.

         From left to right respectively: Original image, obj and bckG seeds, probability map, and segmented lesion

The user-interface used to segment the lesion

Result of the segmentation on a more difficult image

Publication (PDF): Maryam Sadeghi, Geoff Tien, Ghassan Hamarneh, and Stella Atkins. Hands-free Interactive Image Segmentation Using Eyegaze. In SPIE Medical Imaging, 2009.

 

2)Skin Lesion Segmentation using the Random Walker Method

 Abstract:

We present a method for automatically segmenting skin lesions by initializing the random walker algorithm with seed points whose properties, such as colour and texture, have been learnt via a training set. We leverage the speed and robustness of the random walker algorithm and augment it into a fully automatic method by using supervised statistical pattern recognition techniques. We validate our results by comparing the resulting segmentations to the manual segmentations of an expert over 120 cases, including 100 cases which are categorized as difficult (i.e.: low contrast, heavily occluded, etc.). We achieve an F-measure of 0.95 when segmenting easy cases, and an F-measure of 0.85 when segmenting difficult cases.

Introduction

  • We have developed a fully automatic skin lesion segmentation method by leveraging texture metrics, a supervised learning, and the Random Walker segmentation algorithm.

  • We validate our method using a challenging set of images where:
    1) Contrast between skin and lesion is low
    2) Lesion border is not clearly defined (Fuzzy Border)
    3) The entire border is not visible in the lesion
    4) There is considerable occlusion (hair or oil)
    5) There are many different colours present

Method


 

  • Texture features are created by convolving the images with a filterbank consisting of Gaussian and Laplacian of Gaussian filters.

  • Using expertly labeled ground truth and Linear Discriminant Analysis (LDA) the optimal linear combination of texture features which separates the groups of pixels (into lesion/background) is determined. The probability that each pixel belongs to the lesion is computed.

  • A histogram analysis of these ‘probability images’ determines candidate seed points. We fit a Gaussian Mixture Model to the histogram and extract the dominant Gaussians that represent the certain skin and lesion boundaries.

  • Seed points are placed and the random walker algorithm [1] is used to segment the lesion.

  • For the uncertain values Random Walker method labels the pixels.

Results

  • A set of challenging images including low contrast, multi color, fuzzy border, and partial lesions. The  green border is ground truth and the black color represents our automatic segmentation.

Publication (PDF) (Poster): P. Wighton, M. Sadeghi, T.K. Lee and M.S. Atkins. “A fully automatic random walker segmentation for skin lesions in a supervised setting.”  Medical Image Computing and Computer Assisted Interventions – MICCAI 2009, Springer-Verlag Lecture Notes in Computer Science, Sept. 2009, London, UK.

 

Home Research Team Projects Publications Contacts
Last updated: May 4, 2010