The Utility of Optical Coherence Tomography for Diagnosis of Basal Cell Carcinoma

A Quantitative Review

N. Reddy; B.T. Nguyen

Disclosures

The British Journal of Dermatology. 2019;180(3):475-483. 

In This Article

Abstract and Introduction

Abstract

Background: Optical coherence tomography (OCT) is a noninvasive near-infrared light imaging technology that can be utilized to diagnose basal cell carcinomas (BCCs) based on specific morphological features.

Objectives: To conduct a quantitative review using tumour-level data from published studies to assess: (i) the in vivo diagnostic accuracy of different OCT systems; (ii) correlation between OCT features and histopathological diagnosis; and (iii) factors that impact the accuracy of tumour depth estimation.

Methods: Primary tumour-level data were extracted from published studies on the use of time-domain (TD-OCT), frequency-domain (FD-OCT) and high-definition (HD-OCT) systems for diagnosis of BCCs. Quality assessment was performed using the Newcastle–Ottawa Scale and the Cochrane Risk of Bias Tool. Sensitivity and specificity for diagnosis of BCC, prevalence of morphological features and correlation of tumour depth between OCT and histopathology were analysed.

Results: In total, 901 BCCs from 31 studies were included. The sensitivity and specificity were 89·3% and 60·3% overall, and were highest for FD-OCT (93·7% and 61·4%, respectively). The most prevalent morphological features were lobular pattern (80·2%, 315 of 393 tumours) and hyper-reflective peritumoral stroma (51·7%, 203 of 393). Concordance between OCT and histopathological tumour depth categories was moderate (Pearson coefficient 0·48); it was highest for tumours < 1 mm and those on the extremities. The overall bias was 0·075 mm with an agreement range from −0·88 to 1·03 mm. HD-OCT and FD-OCT were superior to TD-OCT at identifying morphological features, but not at tumour depth estimation.

Conclusions: OCT is a viable tool for in vivo diagnosis of BCCs. FD-OCT and HD-OCT outperformed TD-OCT in diagnostic accuracy and detection of morphological features, but not tumour depth estimation.

Introduction

Basal cell carcinoma (BCC) is the most common type of skin cancer. Its incidence in Europe has increased at an alarming rate of 5·5% per year over the past four decades.[1] In the U.K., the incidence of BCC has risen at a rate six times higher than in the rest of Europe.[1] A similar trend is observed in the U.S.A. Studies have estimated an increase ranging from 2% to 11% per year in the incidence of cases of BCC, with approximately 200–400 new cases per 100 000 persons each year.[1,2] Diagnosis of BCC typically requires a skin biopsy and histopathological exam. More recently, noninvasive imaging modalities such as dermoscopy, confocal microscopy and optical coherence tomography (OCT)[3] have been used to diagnose BCC.

OCT is an imaging technology analogous to ultrasonography, except it utilizes near-infrared light rather than acoustic waves to generate real-time, noninvasive cross-sectional images of biological tissues.[3,4] OCT was introduced into the field of dermatology in 1997 when Welzel et al. used the technology to image human skin.[5] It has since shown promising results in several studies as a diagnostic tool.[3,6–15] By measuring the reflected or backscattered light from a sample of skin tissue through a process known as interferometry, it is possible to visualize structures such as the epidermis, dermoepidermal junction (DEJ), dermis, hair follicles, sweat glands and blood vessels.[5,16,17] It is also possible to visualize morphological features of pathological structures within the skin such as a BCC lesion (Figure 1).

Figure 1.

An optical coherence tomography image of a BCC lesion with several visible morphological features: lobular structures of the tumour (asterisks); a dark ring-like pattern surrounding the tumour representing peripheral rimming (red arrows); and hyper-reflective stroma below the lesion (blue arrow).

Early OCT machines, known as time-domain (TD)-OCT, provide penetration depths of up to 2 mm at resolutions of approximately 10–15 μm.[18] At greater depths, resolution is sacrificed, resulting in poorer differentiation of microstructures and cellular details.[17] Conversely, improving the resolution of cellular details will result in a more limited penetration depth.[3,17] Recent advances in OCT technology have made it possible to overcome some of the limitations of TD-OCT. High-definition (HD)-OCT offers significantly improved resolutions of 1–3 μm while sacrificing only some penetration depth (570–750 μm).[19,20] Frequency- or Fourier-domain (FD)-OCT systems seek to combine the best aspects of TD-OCT and HD-OCT, offering a higher resolution (7·5 μm) than TD-OCT and higher penetration depths (1·2–1·8 mm) than HD-OCT.[20] The newer types of OCT machines enable better visualization of skin structures and more accurate characterization of pathological features at greater depths and higher resolutions.[3,20,21]

Prior literature reviews have highlighted the technological principles of OCT and its utility in diagnosing BCCs.[22–25] However, these reviews provide mainly qualitative summaries of published results. There has been no quantitative analysis using tumour data to determine the overall sensitivity and specificity of OCT for diagnosing BCCs or to analyse tumour and machine factors that may influence the diagnostic accuracy of OCT. Therefore, we conducted a quantitative review of primary data to assess the following: (i) the sensitivity and specificity of different OCT systems for in vivo diagnosis of BCC; (ii) correlation between OCT features and histopathological findings; and (iii) the impact of tumour and machine factors on tumour depth estimation by OCT.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....