A novel and robust Bayesian approach for segmentation of psoriasis lesions and its risk stratification

https://doi.org/10.1016/j.cmpb.2017.07.011Get rights and content

Highlights

  • Automatic segmentation of psoriatic lesion using Bayesian model.

  • Multiclass psoriasis risk assessment system (pRAS).

  • Implementation of nine pRAS systems which criss-crosses three classifiers and three feature selection methods.

  • In-depth comparative analysis of performance of nine pRAS systems.

  • Comparative performance of nine pRAS systems with manually segmented data and automatic segmented data.

Abstract

Background and Objective

The need for characterization of psoriasis lesion severity is clinically valuable and vital for dermatologists since it provides a reliable and precise decision on risk assessment. The automated delineation of lesion is a prerequisite prior to characterization, which is challenging itself. Thus, this paper has two major objectives: (a) design of a segmentation system which can model by learning the lesion characteristics and this is posed as a Bayesian model; (b) develop a psoriasis risk assessment system (pRAS) by crisscrossing the blocks which drives the fundamental machine learning paradigm.

Methods

The segmentation system uses the knowledge derived by the experts along with the features reflected by the lesions to build a Bayesian framework that helps to classify each pixel of the image into lesion vs. background. Since this lesion has several stages and grades, hence the system undergoes the risk assessment to classify into five levels of severity: healthy, mild, moderate, severe and very severe. We build nine kinds of pRAS utilizing different combinations of the key blocks. These nine pRAS systems use three classifiers (Support Vector Machine (SVM), Decision Tree (DT) and Neural Network (NN)) and three feature selection techniques (Principal Component Analysis (PCA), Fisher Discriminant Ratio (FDR) and Mutual Information (MI)). The two major experiments conducted using these nine systems were: (i) selection of best system combination based on classification accuracy and (ii) understanding the reliability of the system. This leads us to computation of key system performance parameters such as: feature retaining power, aggregated feature effect and reliability index besides conventional attributes like accuracy, sensitivity, specificity.

Results

Using the database used in this study consisted of 670 psoriasis images, the combination of SVM and FDR was revealed as the optimal pRAS system and yielded a classification accuracy of 99.84% using cross-validation protocol. Further, SVM-FDR system provides the reliability of 99.99% using cross-validation protocol.

Conclusions

The study demonstrates a fully novel model of segmentation embedded with risk assessment. Among all nine systems, SVM-FDR produced best results. Further, we validated our pRAS system with automatic segmented lesions against manually segmented lesions showing comparable performance.

Introduction

Psoriasis is an incurable skin disease but controllable by persistent and vigilant medication [1]. It is characterized as a reddish lesion covered by silvery-white scales on the skin surface due to faster production of skin cells than normal [2]. The cause of this disease is unidentified yet, but genetics is believed to be the leading reason. According to statistics, it affects over 125 million individuals globally [3]. However, its dominance varies geographically such as in Europe, USA, Malaysia and India, it is about 0.6–6.5% [4], 3.15% [4], 3% [5] and 1.02% [6], respectively. Besides affecting the skin, it also affects the quality of life because of its embarrassing physical appearance [7]. Its consequence includes more risk of attempting suicide and is reported to be about 30% and is comparable to life-threatening diseases such as heart disease, diabetes and depression [8]. Psoriasis is categorized as plaque, guttate, inverse, pustular, and erythrodermic based on distinct characteristics. Since plaque psoriasis is most frequently appearing (about 80%) [9], the database considered in this study was affected with plaque psoriasis. A sample of plaque psoriasis lesions is shown in Fig. 1.

Dermatologists are mainly concerned about level of severity of psoriasis disease for prescription of better medication better medication. Currently, dermatologists follow subjective assessment by visual and haptic inspection and accuracy of which depends on the experience and gained proficiency by the dermatologist. Further, inter- and intra- observer variability issue makes the subjective assessment inefficient and unreliable [10]. Hence, we here present a psoriasis risk assessment system (pRAS) that automatically segments the lesions and stratifies the severity of psoriasis.

The machine learning protocol has been adapted in literature for stratification of different dermatology diseases such as melanoma [11], [12], Erythemato-squamous diseases [13], [14] and dermatological ulcer [15]. However, machine learning protocol has been adapted recently for stratification of psoriasis images [16], [17], [18], [19], [20], [21]. The issue with the current psoriasis risk assessment systems is the absence of automatic segmentation of psoriatic lesion. Further, since the images have fuzzy characteristics due to the nature of the disease, it thus needs a classification algorithm which can learn different classes and predict the segmentation regions. We therefore, in this paper present a Bayesian approach for an automatic psoriatic lesion segmentation followed by lesion characterization, stratification and risk severity. These segmented images are then used as inputs for pRAS system for stratification of psoriasis severity. Furthermore, a typical risk stratification system involves three key modules: namely feature extraction, feature selection and classification. Feature selection is a crucial step in the design of risk stratification system and it even becomes more important with rising number of features. Secondly, since different classifiers perform differently based on the database, selection of suitable classifier is also an essential criterion to improve the performance of risk stratification system. Thus, we have designed nine different kinds of pRAS systems using combination of these key blocks and in-depth comparative analyses of their performances have been presented. Thus, the novelties of this paper are: (i) An automatic segmentation of psoriatic lesion using Bayesian model. (ii) Multiclass pRAS system for stratification of psoriasis severity. (iii) Design and development of nine pRAS systems that crisscrosses the three different classifiers and three feature selection techniques. The nine kinds of pRAS systems are: pRAS1: SVM-PCA; pRAS2: SVM-FDR; pRAS3: SVM-MI; pRAS4: DT-PCA; pRAS5: DT-FDR; pRAS6: DT-MI; pRAS7: NN-PCA; pRAS8: NN-FDR; and pRAS9: NN-MI.

Section snippets

Methodology: multiclass framework

The fundamental building block in this novel pRAS design is the automatic lesion segmentation based on Bayesian model. The lesion characterization followed by risk assessment is used a machine learning paradigm embedded with color and grayscale transformation and inter-combination of blocks like feature extraction, selection and cross-validation for risk prediction and performance evaluation.

The proposed pRAS system is shown in Fig. 2. The dotted line divides the system in two segments: offline

Results

The result section has been split-up into two parts. First part deals with the results of proposed automatic Bayesian segmentation approach. Visual as well as quantitative evaluation of segmentation approach has been presented to show the accuracy of our segmentation technique. The second part discusses the results of nine pRAS systems.

Our system

This paper presents risk stratification system to stratify psoriasis images into five classes namely: healthy, mild, moderate, severe and very severe. Moreover, nine pRAS systems have been designed by criss-cross combination of three classifiers (SVM, DT and NN) and three feature selection techniques (PCA, FDR and MI) and their performances have been compared. The automatic psoriatic lesion segmentation approach was developed as part of the automated risk assessment system. We validated the

Conclusion

The paper presented a complete automatic system for psoriasis disease severity assessment in multi-class scenario which combines segmentation and risk stratification. The main contribution is automatic segmentation of psoriatic lesion using Bayesian approach and development of nine risk stratification system utilizing three classifiers and three feature selection techniques. A comprehensive analysis among these nine pRAS systems has been established.

The combination of SVM classifier and FDR

Declaration of conflicting interests

None declared.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this paper.

Vimal K. Shrivastava has received his BE degree in Electronics and Telecommunication engineering from the Chhattisgarh Swami Vivekanand Technical University, Bhilai, Chhattisgarh in 2009 and MTech degree in Electronics Instrumentation engineering from National Institute of Technology Warangal, Andhra Pradesh in 2011. Later, he received his PhD degree from National Institute of Technology, Raipur, India in 2016. He is currently working as Assistant Professor in School of Electronics Engineering

References (58)

  • U.R. Acharya et al.

    Atherosclerotic risk stratification strategy for carotid arteries using texture-based features

    Ultrasound Med. Biol.

    (2012)
  • WuC.M. et al.

    Statistical feature matrix for texture analysis

    CVGIP: Graph. Models Image Process.

    (1992)
  • ChuaK.C. et al.

    Application of higher order statistics/spectra in biomedical signals—A review

    Med. Eng. Phys.

    (2010)
  • U.R. Acharya et al.

    Understanding symptomatology of atherosclerotic plaque by image-based tissue characterization

    Comput. Methods Programs Biomed.

    (2013)
  • U.R. Acharya et al.

    Automated diagnosis of epileptic EEG using entropies

    Biomed. Signal Process. Control

    (2012)
  • M. Zortea et al.

    Performance of a dermoscopy-based computer vision system for the diagnosis of pigmented skin lesions compared with visual evaluation by experienced dermatologists

    Artif. Intell. Med.

    (2014)
  • P. Luukka

    Similarity classifier using similarity measure derived from Yu's norms in classification of medical data sets

    Comput. Biol. Med.

    (2007)
  • M. Karabatak et al.

    A new feature selection method based on association rules for diagnosis of erythemato-squamous diseases

    Expert Syst. Appl.

    (2009)
  • XieJ. et al.

    Using support vector machines with a novel hybrid feature selection method for diagnosis of erythemato-squamous diseases

    Expert Syst. Appl.

    (2011)
  • C. Camisa

    Handbook of Psoriasis

    (2004)
  • T. Henseler

    Genetics of psoriasis

    Arch. Dermatol. Res.

    (1998)
  • S. Dogra et al.

    Psoriasis in India: prevalence and pattern

    Indian J. Dermatol., Venereol. Leprol.

    (2010)
  • G. Krueger et al.

    The impact of psoriasis on quality of life: results of a 1998 national psoriasis foundation patient-membership

    Arch. Dermatol.

    (2001)
  • C. Olivier et al.

    The risk of depression, anxiety, and suicidality in patients with psoriasis: a population-based cohort study

    Arch. Dermatol.

    (2010)
  • T. Morrow

    Evaluating new therapies for psoriasis

    Manag. Care

    (2004)
  • N. Razmjooy et al.

    A computer-aided diagnosis system for malignant melanomas

    Neural Comput. Appl.

    (2013)
  • J. Glaister et al.

    MSIM: Multistage illumination modeling of dermatological photographs for illumination-corrected skin lesion analysis

    IEEE Trans. Biomed. Eng.

    (2013)
  • Cited by (78)

    • NeoAI 1.0: Machine learning-based paradigm for prediction of neonatal and infant risk of death

      2022, Computers in Biology and Medicine
      Citation Excerpt :

      Another weakness is that the number of risk factors or covariates present in the dataset was only 22. This was enough for preparation, predictions, and evaluation of the model; however, we could exercise the feature effect power much better with a larger number of features [16,26,93]. Even though we used the ADASYN paradigm for augmenting the imbalanced classes, various other possibilities exist to compare ADASYN models.

    View all citing articles on Scopus

    Vimal K. Shrivastava has received his BE degree in Electronics and Telecommunication engineering from the Chhattisgarh Swami Vivekanand Technical University, Bhilai, Chhattisgarh in 2009 and MTech degree in Electronics Instrumentation engineering from National Institute of Technology Warangal, Andhra Pradesh in 2011. Later, he received his PhD degree from National Institute of Technology, Raipur, India in 2016. He is currently working as Assistant Professor in School of Electronics Engineering of Kalinga Institute of Industrial Technology, Bhubaneswar, India.

    Narendra D. Londhe received his BE degree from Amravati University in 2000. Later he received his MTech and PhD degrees in the year 2004 and 2011, respectively from Indian Institute of Technology Roorkee. He is presently working as Assistant Professor in Department of Electrical Engineering of National Institute of Technology, Raipur, India.

    Rajendra S. Sonawane is graduated from National Institute of Homeopathy, Kolkata, India. Later he received his M.D. and D.I. in homeopathy from London. He has treated over 15,000 psoriasis patients personally from all over India and many countries of the world since 27 years with homeopathic medicines only. He is the ex-professor of Homeopathic Medical Colleges, Malkapur, Amravati, Dhule and Shirpur.

    Jasjit S. Suri, PhD, MBA, Fellow AIMBE is an innovator, visionary, scientist and an internationally known world leader. Dr. Suri received the Director General's Gold medal in 1980 and the Fellow of American Institute of Medical and Biological Engineering, awarded by National Academy of Sciences, Washington DC in 2004. He has published over 500 peer reviewed articles (H-index ∼ 42) and book chapters and over 100 innovations/trademarks. He is currently Chairman of Global Biomedical Technologies, Inc., Roseville, CA.

    View full text