A standardised protocol for texture feature analysis of endoscopic images in gynaecological cancer

Department of Computer Science, University of Cyprus (UCY), Nicosia, Cyprus.
BioMedical Engineering OnLine (Impact Factor: 1.43). 02/2007; 6(1):44. DOI: 10.1186/1475-925X-6-44
Source: DOAJ


In the development of tissue classification methods, classifiers rely on significant differences between texture features extracted from normal and abnormal regions. Yet, significant differences can arise due to variations in the image acquisition method. For endoscopic imaging of the endometrium, we propose a standardized image acquisition protocol to eliminate significant statistical differences due to variations in: (i) the distance from the tissue (panoramic vs close up), (ii) difference in viewing angles and (iii) color correction.
We investigate texture feature variability for a variety of targets encountered in clinical endoscopy. All images were captured at clinically optimum illumination and focus using 720 x 576 pixels and 24 bits color for: (i) a variety of testing targets from a color palette with a known color distribution, (ii) different viewing angles, (iv) two different distances from a calf endometrial and from a chicken cavity. Also, human images from the endometrium were captured and analysed. For texture feature analysis, three different sets were considered: (i) Statistical Features (SF), (ii) Spatial Gray Level Dependence Matrices (SGLDM), and (iii) Gray Level Difference Statistics (GLDS). All images were gamma corrected and the extracted texture feature values were compared against the texture feature values extracted from the uncorrected images. Statistical tests were applied to compare images from different viewing conditions so as to determine any significant differences.
For the proposed acquisition procedure, results indicate that there is no significant difference in texture features between the panoramic and close up views and between angles. For a calibrated target image, gamma correction provided an acquired image that was a significantly better approximation to the original target image. In turn, this implies that the texture features extracted from the corrected images provided for better approximations to the original images. Within the proposed protocol, for human ROIs, we have found that there is a large number of texture features that showed significant differences between normal and abnormal endometrium.
This study provides a standardized protocol for avoiding any significant texture feature differences that may arise due to variability in the acquisition procedure or the lack of color correction. After applying the protocol, we have found that significant differences in texture features will only be due to the fact that the features were extracted from different types of tissue (normal vs abnormal).

Download full-text


Available from: Marios S Pattichis
  • Source
    • "A snapshot of the CAD system is demonstrated in Fig. 1. Moreover, a standardized protocol to support CAD in hysteroscopy (as well as in endoscopy and laparoscopy) image analysis is given in Fig. 2. The original protocol described in [5] has been extended from covering RGB to include HSV and YCrCb. Furthermore, the new method includes ROI classification and reporting/diagnosis. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The paper presents the development of a Computer Aided Diagnostic (CAD) system for the early detection of endometrial cancer. The proposed CAD system supports reproducibility through texture feature standardization, standardized multi-feature selection, and provides physicians with comparative distributions of the extracted texture features. The CAD system was validated using 516 Regions of Interest (ROIs) extracted from 52 subjects. The ROIs were equally distributed among normal and abnormal cases. To support reproducibility, the RGB images were first gamma corrected and then converted into HSV and YCrCb. From each channel of the gamma-corrected YCrCb, HSV, and RGB color systems, we extracted the following texture features: (i) Statistical Features (SF), (ii) Spatial Gray Level Dependence Matrices (SGLDM), and (iii) Gray Level Difference Statistics (GLDS). The texture features were then uses as inputs with Support Vector Machines (SVM) and the Probabilistic Neural Network (PNN) classifiers. After accounting for multiple comparisons, texture features extracted from abnormal ROIs were found to be significantly different than texture features extracted from normal ROIs. Compared to texture features extracted from normal ROIs, abnormal ROIs were characterized by lower image intensity, while variance, entropy and contrast gave higher values. In terms of ROI classification, the best results were achieved by using SF and GLDS features with an SVM classifier. For this combination, the proposed CAD system achieved an 81% correct classification rate.
    Full-text · Article · Jun 2014 · IEEE Journal of Biomedical and Health Informatics
  • Source
    • "The most popular methods according to [10] are based on the polynomial based correction method [11] and neural networks transformation [12]. However, a very limited number of studies used color correction for hysteroscopy or labaroscopy imaging [6]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this study was to investigate the reliability of two hysteroscopy cameras, the Circon IP 4 and Karl Storz HD, in relation to white balance, camera response over time, and color correction. Experimental results, show that for both the Circon and Karl Storz cameras: (i) for white balancing, either guaze, or white sheet, or the white color cheker can be used, (ii) white balance can be carried out at any distance between 0.5 to 3.0 cm, (iii) white balance can be carried out at 1cm and at any angle between 0 to 45 degrees, (iv) there was no camera color variation over both short (60 min) and long (4 weeks) time intervalls, and (v) color correction algorithm 2 gave better results. Most imortantantly, there was no significant difference between the MSE of the Circon and Storz cameras. The above results will be incorporated in a standardized protocol for texture feature analysis of endoscopic imaging for gyneacological cancer developed by our group, and will enable multi-center quantitative analysis (constrained by the use of the two cameras investigated). KeywordsEndoscopy imaging-laparoscopy imaging-hysteroscopy imaging-white balance-gamma correction algorithm-color systems-CCD medical camera
    Full-text · Conference Paper · Jan 2010
  • Source
    • "ROIs of 64x64 pixels were manually cropped and classified into two categories: (i) normal ROIs (N=202) and (ii) abnormal ROIs (N=202) based on the physician's subjective criteria and the histopathological examination (see Fig. 1). All images were color corrected for camera sensor variations based on an extended gamma algorithm [2]. Furthermore, all ROIs were transformed to grey scale, HSV, and YCrCb. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The objective of this study was to develop a CAD system for the classification of hysteroscopy images of the endometrium (with suspicious areas of cancer), based on two data mining procedures, the C4.5 and the Hybrid Decision Tree (HDT) algorithms. Twenty-six texture features were extracted from three texture features algorithms: (i) Statistical Features (SF), (ii) Spatial Gray Level Dependence Matrices (SGLDM), and (iii) Gray level difference statistics (GLDS). A total of 404 ROIs of the endometrium in RGB system format were recorded (202 normal and 202 abnormal) from 40 subjects. Images were gamma corrected and converted to grey scale, and the HSV and YCrCb systems. Results show that abnormal ROIs had lower grey scale median and homogeneity values, and higher entropy and contrast values when compared to the normal ROIs. The maximum average correct classifications score was 72,2% and was achieved using the HDT algorithm using 26 texture features, for the Y channel. Similar performance was achieved with both the HDT and the C4.5 algorithms when trained with the YCrCb texture features. Although similar performance to these models was also achieved when using the SVM and PNN models, the decision tree algorithms investigated, facilitated also the rule extraction, and their use for classification. These models can help the physician especially in the assessment of difficult cases of gynaecological cancer. However, more cases have to be collected and analysed before the proposed CAD system can be exploited in clinical practise.
    Full-text · Conference Paper · Nov 2008
Show more