Role of image thermography in early breast cancer detection- Past, present and future

https://doi.org/10.1016/j.cmpb.2019.105074Get rights and content

Highlights

  • This systematic review aims to present the emerging techniques used to improve accuracy of thermography based diagnosis systems for breast cancer.

  • Details of feature extraction techniques by various authors have been discussed with improvements obtained.

  • Methods to overcome shortcomings of breast thermography as an adjunct tool have also been mentioned.

Abstract

One of the most prevalent cancers among women is the breast cancer. Accurate diagnosis of breast cancer at an early stage can reduce the mortality associated with this disease. Infrared Breast Thermography, which is a screening tool used to measure the temperature distribution of breast tissue, is a suitable adjunct tool to mammography. Breast thermography has many advantages as it is non-invasive, safe and painless. Thermographic image and usage of artificial neural networks have improved the accuracy of thermography in early diagnosis of breast abnormality. This paper presents survey based on the main steps of computer aided detection systems: image acquisition protocols, segmentation techniques, feature extraction and classification methods, used in the field of breast thermography over the past few decades. The detailed survey emphasizes on the improved reliability of breast thermography .This has become possible with the utilization of machine learning techniques for correct classification of breast thermograms. Numerical Simulation can be used as a supporting method to overcome high false positive rates in thermographic diagnosis. The paper also presents future recommendations to utilize recent machine learning advances in real time.

Introduction

Cancer is the condition in which cells grow irregularly and affect other parts of the body. Some of the cancers found commonly are breast, prostrate, lung, skin and pancreas. Cancer results in large number of mortalities worldwide [1], [2]. The instances of growth of breast cancer amongst women in the age group of 30- 40 years, has increased significantly over the last few decades in India [3]. The most frequently diagnosed cancer in women is breast cancer. Rates of diagnosis of breast cancer have been increasing drastically every year in nearly every region of the world. Proper identification of breast abnormality prior to the beginning of a cancerous growth is the only effective way of reducing mortality due to breast cancer. This review paper presents a chronological review of some of the papers related to breast thermography, an adjunct imaging modality for early cancer diagnosis. The main objective of the review paper is to analyze the improvement in accuracy of thermogram classification based on the selection of segmentation techniques, feature selection, feature extraction and types of classifiers used. The work also highlights the limitations of breast thermography and suggests methods of improving the sensitivity.

The paper is organized in 7 sections. Section 2 briefly explains the methodology used in the review of literature. Section 3 introduces the concept of breast thermography and the need of such imaging technique. Subsection 3 presents some of the available database of breast thermograms. Section 4 outlines the parts of intelligent diagnosing systems used in early cancer detection. This section also highlights the importance of segmentation with the recent advances. Section 5 presents a detailed progress in feature extraction and classification techniques used. Section 6 briefly describes future needs in the field. Section 7 discusses the implications of the study. Section 8 presents the main conclusions and recommendations based on them.

Section snippets

Sources

Research articles presented in the review paper are obtained from IEEEXplore, PubMed Central, Semantic Scholar, ResearchGate, ScienceDirect and other journal databases. Some of the articles were searched by going through the references of the searched papers.

Keywords

Various possible combinations of keywords such as “Infrared Thermography”, “Segmentation”, “Intelligent Cancer Classification”, “Machine Learning”, “Feature Extraction” and “Numerical Simulation” have been used for the search of articles.

Inclusion and exclusion criteria

In

Breast thermography

There are many types of cancer that can originate in any part of the breast. Breast cancer begins either in milk carrying ducts or the milk producing lobules of the breast region. The detailed structure of female breast is shown in Fig. 1.

Proper identification of breast abnormality prior to the beginning of a cancerous growth is the only effective way of reducing mortality due to breast cancer. Various imaging modalities have been developed to detect prior symptoms of breast cancer. Apart from

Machine learning based system for analysis of breast thermogram

Usage of intelligent Computer based system for analyzing the breast thermograms has emerged as an effective tool for providing the support to the radiologists. The process involves preprocessing of thermogram, extraction of region of interest or segmentation, extraction of features and classification. Fig. 4 illustrates the overall methodology involved in Cancer Diagnosis system used for thermographic images.

Breast thermograms obtained from a thermal camera may consist of some labels which are

Emerging trends in extraction of significant features and classification

The main process in Medical image processing is Feature Extraction. This involves choosing certain parameters, known as features that will be extracted from a breast thermogram, analyzing and comparing the features to obtain significant results. This will reduce complexity in classification and recognition of images. Feature extraction techniques help in overcoming some of the disadvantages of breast thermography such as (i) its inability in diagnosing small tumors, (ii) identify increase in

Future trends in breast thermography

Machine learning based system proves to be effective in classification and detection in the medical field, with necessary training skills, selection of more significant features and reduced false positives. Moreover, there is a future need to develop database with millions of thermographic images for improving the efficiency of classifiers. The future research work may also involve improving the efficiency of classifiers with the limited number of thermograms available. There is also the need

Discussion

This article attempts to present the key factors to improve the efficiency of existing thermography based diagnosis. Validation of segmentation techniques (some are enlisted in Table 1), by the physicians with ground–truth further improves the reliability of methods used. Performance of classifier is dependent on the choice of features to be extracted from a thermogram: statistical, textural, fractal or medically interpretable. Authors in the paper have used various feature selection methods to

Conclusion

The advent of computer-aided diagnostics in healthcare field has proved to be very effective in improving the role of thermography in detection of breast cancer. The study presents various segmentation, feature extraction and classification techniques used in breast thermography. High false positive and false negative values in thermography can be effectively reduced by suitable combinations of segmentation, feature extraction and classification techniques mentioned in the paper. Specific

Declaration of Competing Interest

The author(s) declare(s) that there is no conflict of interest regarding the publication of this paper.

References (54)

  • S.M. Friedewald

    Breast cancer screening using tomosynthesis in combination with digital mammography

    JAMA - J. Am. Med. Assoc.

    (2014)
  • T.D. Vreugdenburg et al.

    A systematic review of elastography, electrical impedance scanning, and digital infrared thermography for breast cancer screening and diagnosis

    Breast Cancer Res. Treat.

    (2013)
  • T.B. Borchartt et al.

    Thermal feature analysis to aid on breast disease diagnosis

  • M. Gogoi et al.

    A study and analysis of hybrid intelligent techniques for breast cancer detection using breast thermograms

  • HRT doubles breast density. [Online]....
  • Breast cancer awareness. [Online]. Available:...
  • Y. Hayashi et al.

    New approach to breast tumor detection based on fluorescence x-ray analysis

    Ger Med Sci.: GMS e-journal

    (2014)
  • C.K. Kuhl

    Mammography, breast ultrasound, and magnetic resonance imaging for surveillance of women at high familial risk for breast cancer

    J. Clin. Oncol.

    (2019)
  • S.H. Heywang-Köbrunner et al.

    Advantages and disadvantages of mammography screening

    Breast Care (Basel)

    (2011)
  • N.A. Diakides et al.

    Advances in medical infrared imaging

  • S.T. Kakileti et al.

    Advances in Breast Thermography

    In New Perspectives in Breast Imaging

    (2017)
  • H.H. Pennes

    Analysis of tissue and arterial blood temperature in the resting human forearm

    J. Appl. Physiol.

    (1998)
  • I.O.F. Surface, T.I.N. The, D. Of, and B. Cancer, Principle employed, vol. 75, pp. 309–310,...
  • E.Y.K. Ng et al.

    Statistical analysis of healthy and malignant breast thermography

    J. Med. Eng. Technol.

    (2001)
  • Image processing and image analyses applied to mastology. [Online]. Available:http://visual.ic.uff.br/en/proeng/S....
  • M.K. Bhowmik et al.

    Designing of ground-truth-annotated DBT-TU-JU breast thermogram database toward early abnormality prediction

    IEEE J. Biomed. Health Inform.

    (2018)
  • L.F. Silva

    A new database for breast research with infrared image

    J. Med. Imaging Health Inform.

    (2013)
  • Cited by (116)

    View all citing articles on Scopus
    View full text