Cutting our losses – how point-of-care tissue-scanning technology can reduce unnecessary biopsies

Skin cancer may initially present with visible changes, such as changes to pigmentation, size, border, or texture. Typically, this leads to a referral to a dermatological specialist, who inspects the lesion, and if considered necessary, it will be biopsied. While biopsy and histopathological analysis can tell a clinician a great deal about the nature of the lesion that cannot be achieved by the clinical features and examination alone, they do not come without risks. Biopsy increases the risk of infection and scarring, as well as causing pain and distress to the patient. Additionally, there are a great deal of benign lesions biopsied for every case of skin cancer that is detected, and many samples go without assessment due to the high-demand and the time it takes to process and assess the samples. There are significant moral questions around the issue of unnecessarily giving medical interventions that carry risk to the patient, cause both physical discomfort and emotional distress, and any equally effective alternative that allows for a quicker turnaround time, less pain, injury, and less distress, should be strongly considered.

“Approximately 15–30 benign lesions are biopsied to diagnose one case of skin cancer”

Detection of skin cancer by harnessing computer-aided diagnostic (CAD) systems has been developed and may be used in the clinical setting. These utilise different approaches, but one of note uses convolutional neural networks which have been educated to identify and classify skin lesions. To date, a major issue with CNN pattern recognition in the field of diagnostics is poor specificity, although sensitivity tends to be high, running the risk of a disproportionate number of false positive results. As such, a clinician’s opinion is currently comparable to one of these systems.
Researchers from Stevens Institute of Technology, New Jersey, have developed non-invasive, hand-held technology, using mullimeter-wave imaging (MMWI), for use in the diagnosis of skin cancer. It does this by measuring whether biochemical and structural changes, such as tissue shape, size, density, and altered microstructures, associated with malignancy have occurred within tissue. It was also shown to be capable of distinguishing between the major types of skin cancer, including melanoma, basal cell carcinoma (BCC), and squamous cell carcinoma (SCC), as well as pre-cancerous changes. To assess this, a number of lesions were assessed using MMWI, and the outcome was compared with histopathological analysis and clinical assessment to establish whether these aligned and gave comparable diagnoses. Inclusion criteria for this study included lesions with concerning features that might indicate malignancy or pre-malignant changes that will later require treatment, and benign dermatological conditions that may be mistaken for a skin cancer. In total, 146 skin lesions were assessed.
This technology was shown to differentiate between pre-cancerous and cancerous lesions, and healthy tissue, with a sensitivity and specificity measure of 97% and 98%, respectively. Such innovation has the potential to improve detection of skin cancers, reduce the number of unnecessary biopsies carried out on patients, and speed up necessary referrals in those patients deemed eligible for further interventions. Additionally, the possibility of differentiating between different skin cancers will help inform treatment options from an earlier stage in the patient pathway. Finally, the prospect of using MMWI in screening technology in high-risk individuals is exciting, as it may facilitate early detection, reduce disease burden, and innovate prophylactic approaches in those individuals where pre-cancerous changes have been seen, much like the cervical screening programme currently employed in the UK.
Furthermore, the implications on its quick-processing and non-invasive methodology are that it allows a more rapid assessment of patients in clinic, provides more data to be considered by the clinician as part of the diagnostic process, and can be incorporated into a more compact, hand-held format, useful for point-of-care diagnosis, which is cheaper, and allows more widespread access to improve diagnostic technology for a greater number of patients.

https://www.nature.com/articles/s41598-022-09047-6