An invasive biopsy followed by histological staining is the benchmark for pathological diagnosis of skin tumors. The process is cumbersome and time-consuming, often leading to unnecessary biopsies and scars. Emerging noninvasive optical technologies such as reflectance confocal microscopy (RCM) can provide label-free, cellular-level resolution, in vivo images of skin without performing a biopsy. Although RCM is a useful diagnostic tool, it requires specialized training because the acquired images are grayscale, lack nuclear features, and are difficult to correlate with tissue pathology. Here, we present a deep learning-based framework that uses a convolutional neural network to rapidly transform in vivo RCM images of unstained skin into virtually-stained hematoxylin and eosin-like images with microscopic resolution, enabling visualization of the epidermis, dermal-epidermal junction, and superficial dermis layers. The network was trained under an adversarial learning scheme, which takes ex vivo RCM images of excised unstained/label-free tissue as inputs and uses the microscopic images of the same tissue labeled with acetic acid nuclear contrast staining as the ground truth. We show that this trained neural network can be used to rapidly perform virtual histology of in vivo, label-free RCM images of normal skin structure, basal cell carcinoma, and melanocytic nevi with pigmented melanocytes, demonstrating similar histological features to traditional histology from the same excised tissue. This application of deep learning-based virtual staining to noninvasive imaging technologies may permit more rapid diagnoses of malignant skin neoplasms and reduce invasive skin biopsies.
Background Cutaneous metastases (CM) diagnosis is clinically challenging, requiring an invasive biopsy for confirmation. A novel, RCM‐OCT device combines the advantage of horizontal high‐resolution reflectance confocal microscopy (RCM) images and vertical deeper optical coherence tomography (OCT) images to aid in non‐invasive diagnosis of CM from breast cancers. Objective Characterize CM from breast cancers using RCM‐OCT device. Methods Seven patients suffering from breast cancers with suspicious CM were consented and imaged with RCM‐OCT device. CM features were defined by comparing with histopathology. Tumour depths were measured on OCT and on H&E‐images and correlated using statistical analysis Pearson test. 3D‐OCT images were reconstructed to enhance tumour visualization. Results 6/7 lesions were CM from breast cancers, and one was vascular ectasia, on histopathology. CM appeared as greyish‐darkish oval to round structures within the dermis on RCM and OCT‐images. On RCM, individual tumour cells were seen, enabling identification of even small tumour foci; while, on OCT deeper tumours were detected. Inflammatory cells, dilated vessels and coarse collagen were identified in the dermis. Pearson correlation had an r2 of 0.38 and a significant P‐value <0.004 for depth measurements. CM from breast cancers could be differentiated from ecstatic vessels on 3D‐reconstructed OCT image. Limitation Small sample size and lack of clinical mimickers. Conclusion RCM‐OCT can detect CM and has potential in aiding non‐invasive diagnosis and management.
BACKGROUND Reflectance confocal microscopy (RCM) is a noninvasive tool that is used to diagnose skin cancers. However, RCM requires an expert consultation, which is often performed via store-and-forward (SAF) teledermatology. Unfortunately, SAF does not mimic bedside diagnosis, nor permits interaction between the remote expert reader, physician, and patient. Recently, a live interactive method (LIM)–tele-RCM approach was shown to diagnose basal cell carcinoma (BCC) from a remote location, demonstrating advantages over SAF by providing a bedside diagnosis during consultation. OBJECTIVE The aim of this study is to validate the LIM-tele-RCM approach to diagnose BCC in a real-world setting. METHODS In this pilot study, 4 patients with 6 clinically suspicious BCC lesions were enrolled and imaged with RCM at a Los Angeles dermatology clinic. A Health Insurance Portability and Accountability Act–compliant teleconferencing application was used to livestream RCM images to an expert RCM reader in New York. The expert reader had remote control of the software, direct audio communication with the clinic, and the patient’s clinical history with dermoscopy. During imaging, RCM features were noted, and a diagnosis was made at the bedside. After imaging, patients completed a short questionnaire (on a 5-point scale, with 5 being the highest score) about satisfaction, comfort, and communication during the session. RESULTS RCM diagnosed 4/6 (67%) lesions correctly as BCC and 2/6 (33%) were false-positive diagnoses. The true-positive lesions had “tumor islands with palisading and clefting” and were directly managed with Mohs surgery. The false-positive lesions had “dark silhouettes” (a common false-positive feature for BCC) and underwent a shave biopsy for confirmation. The entire session ranged from 15 to 20 minutes (an average of 17.7 minutes), comparable to the reported RCM procedure time. On the questionnaire, all patients responded with the highest rating (5/5) for each question. CONCLUSIONS LIM-tele-RCM demonstrates potential advantages over the SAF method, enabling bedside diagnosis with similar diagnostic accuracy as reported in the literature and proper management. Additionally, the remote reader had access to patients’ clinical backgrounds and could engage with patients. It may also be useful for training novice RCM users and beneficial in settings where remote diagnostics are desired, such as during the COVID-19 pandemic. However, technical challenges such as image quality degradation during video streaming, poor internet bandwidth, and end user latency may impact diagnosis. Larger, multicenter studies are needed to assess the accuracy of LIM-tele-RCM for the diagnosis of BCC and other neoplastic and inflammatory lesions, and to quantify technical limitations.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.