e following classification groups w

e following classification groups were used for compari�sons between the classifier trained on physiological parame�ters and that based on features extracted using PCA sincethey are most pertinent for NMSC diagnosis: [1] (BCC,SCC) vs AK and [2] (BCC, SCC, AK) vs visually-normaltissue. These are the same tasks reported in Lim et al [7].Classification group (1) simulates the clinical scenario ofa clinician examining a patient and finding a lesion that isconcerning for skin cancer. The provider would use the non�invasive device to decide whether they should pursue a sur�gical treatment of cancer or just treat the lesion as an AKwith cryotherapy or prescription creams (less invasive). Theclinician uses this device to avoid a biopsy when possible.The clinical scenario for classification group (2) is wherea patient has a known skin cancer and is undergoing surgery.The clinician can use noninvasive imaging to quickly assess4 of 8 ZHANG ET AL.the margins of that tumor and decide if the edges are “clearof cancer” or if they need to go back and remove more tissue. The clinician would use this device to know when to“stop” and avoid cutting normal skin