A clinical imaging study using a smartphone point-of-care autofluorescence device to facilitate the early detection of oral cancer in primary and community health clinical settings

Introduction

Early diagnosis of oral cancer is key to improving prognosis and patient outcomes. Suspicious oral lesions are often encountered by front-line dental and medical practitioners who must decide whether to refer the patient to a specialist for biopsy and histopathology, which is the gold standard for diagnosis. The development of objective tools for screening of oral mucosal lesions that enable clinical risk stratification is imperative for improving early detection of oral cancer, especially in resource-limited primary care settings.

Methods and Materials

The mobile Detection of Oral Cancer (mDOC) device is a smartphone autofluorescence imaging system that captures wide field white light and autofluorescence images of the oral mucosa using optics in a custom 3D printed case. A custom Android application was developed to collect patient demographic information and to guide the clinician to collect widefield images in the oral cavity. Lesion specific characteristics are also recorded in the app. This study enrolled patients presenting to the UTHealth School of Dentistry and MD Anderson Cancer Center clinics for clinical evaluation. Referral decisions were determined by expert clinicians.

Results

Preliminary results consist of white light and autofluorescence image data from a total of 223 anatomic sites in 111 patients. Of these 223 sites, 174 sites prompted referral for further evaluation and 49 sites did not prompt referral for further evaluation based on expert clinical impression. This dataset is being used to develop an automated algorithm to generate a referral decision at the point of care.

Conclusion

The mDOC device serves as a simple tool to image the oral mucosa for objective screening, with the potential to enable earlier detection of oral lesions in primary care settings. Future work involves the implementation of diagnostic algorithms to inform site-level and patient-level referral decisions to determine whether a patient requires further evaluation.