/
Medical Analysis
Hyperspectral Imaging of Skin Cancer

Detecting skin cancer with hyperspectral imaging

Hyperspectral imaging (HSI) is an emerging imaging modality that new opportunities to detect and diagnose cancer and disease.

About Opsyne

Opsyne builds and sells OEM hyperspectral imaging technology that can be integrated into complete industrial automation solutions.

Skin Cancer

Skin cancer is a very common type of cancer worldwide. While most skin cancers are not deadly, malignant melanoma is the most fatal and its early detection is vital for effective treatment and a better chance of patient survival. The survival rate for a non-metastasized melanoma is found to be 99% whereas chances of survival reduce to only 20% if it spreads to other organs. (Source: American Cancer Society, 2018). Early detection is therefore highly valuable.

Diagnosis

Pigmented skin lesions are classified according to the ABCD rule of dermatology. The respective letters in the acronym represent different features of a skin lesion:

  • A = Asymmetry of the lesion
  • B = Border irregularity
  • C = Color uniformity
  • D = Differential structures

Using the ABCD rule to detect a pigmented skin lesion, a score is assigned to each of these four features, and an overall score is computed. Higher the overall score, the greater the potential for malignancy.

However, with the addition of imaging technologies the ABCD rule of dermatology has been extended to ABCDE, where the E represents the “Evolution of the skin lesion over time.”

Detection and Classification

Optical imaging systems for the classification of skin lesions has been an active research field for several decades. While early systems used RGB images, by the early 2000s, most systems started making use of dermatologic images. The conventional and dermatologic systems appear to have reached a limit for their performance, but the performance can be improved by emerging technologies such as hyperspectral. In other words, there is a suite of non-invasive measurement technologies, but their effectiveness is limited by their ability to see beyond the visible spectrum and perform material analysis on the skin.

Hyperspectral Imaging

Hyperspectral imaging (HSI) is an emerging imaging modality that new opportunities to detect and diagnose cancer and disease.

  • The main benefit of HSI compared with conventional imaging techniques is the prospect of exploring spectral regions beyond the human eye capabilities.
  • Hyperspectral imaging has substantial potential as a non-destructive and non-ionizing imaging technique, supporting swift acquisition and analysis of diagnostic information.
  • HSI has the capability of capturing hundreds of narrow spectral bands to help distinguish the cancerous and diseased cellular structures.

Recent Advancements

A system that aims at clinical relevance must either have melanoma sensitivity close to 100% combined with reasonable specificity or help facilitate physicians to decide whether to excise the lesion in question. Until recently, hyperspectral imaging limited by available classification data and the optical sensitivity been unable to achieve either of these objectives.

Implications

Hyperspectral imaging can be used in concert with long-established diagnostic workflows to help clinicians identify disease and cancer markers and changes earlier in the process. Imaging systems can be either handheld or as a body scanner to capture image data from specified areas of interest. The combination of advanced machine learning techniques and superior spectral resolution to increase sensitivity, along with optimised signal-to-noise ratio to improve specificity are positive developments. The greater vision for non-invasive patient-friendly low-case early detection enabled by hyperspectral imaging requires a concerned and collaborative effort including researchers and clinicians to combine state-of-the-art imaging technology with the appropriate classification libraries.

Sources:

Application notes

Learn more about how hyperspectral imaging technology can be used to improve insights in specific applications.

Explore all Application notes