Identification of Cancerous Skin Lesions Using Vibrational Optical Coherence Tomography (VOCT): Use of VOCT in Conjunction with Machine Learning to Diagnose Skin Cancer Remotely Using Telemedicine

SIMPLE SUMMARY: Skin cancer detection is an important problem since it is the most common form of cancer in the US and the number of cases is increasing in the US and worldwide. Early detection of these cancers can limit their ability to: (1) disseminate throughout the body, (2) cause illnesses or (3) even cause premature death. We have used audible sound from a speaker at different frequencies to vibrate the skin and then used low intensity red light to determine how the different sound waves displaced the skin. The amount of displacement is related to the stiffness of each tissue component. Cancerous tissues are found to be stiffer than normal skin and the degree of stiffness can be used to differentiate between normal skin and skin cancers. The results of these studies indicate that skin cancers can be detected remotely using a device to vibrate and measure the displacement of tissues at different frequencies. Use of computer techniques and remote testing can facilitate identification of skin cancers in areas underserved by Dermatologists. Since this technology can detect skin cancers as small as 0.1 mm early detection will limit the undesirable effects of skin cancer. ABSTRACT: In this pilot study, we used vibrational optical tomography (VOCT), along with machine learning, to evaluate the specificity and sensitivity of using light and audible sound to differentiate between normal skin and skin cancers. The results reported indicate that the use of machine learning, and the height and location of the VOCT mechanovibrational peaks, have potential for being used to noninvasively differentiate between normal skin and different cancerous lesions. VOCT data, along with machine learning, is shown to predict the differences between normal skin and different skin cancers with a sensitivity and specificity at rates between 78 and 90%. The sensitivity and specificity will be improved using a larger database and by using other AI techniques. Ultimately, VOCT data, visual inspection, and dermoscopy, in conjunction with machine learning, will be useful in telemedicine to noninvasively identify potentially malignant skin cancers in remote areas of the country where dermatologists are not readily available.