test

0 20 40 60 80 1000

20

40

60

80

100

100-Specificity

Sens

itivi

ty

Sensitivity: 90.0 Specificity: 77.8 Criterion : >-1.335

FIELD SCREENING FOR ORAL CANCER USING OPTICAL COHERENCE TOMOGRAPHY

Emon Heidari3, Sumsum Sunny1,2, Bonney Lee James2, Ravindra DR2, Subhashini AR4, keerthi G4, Shubha G4, Uma K4 , Sandeep Kumar4, Swathi Mani4, Vikram Kekatpure1, Praveen BN4, Amritha Suresh1,2, Tracie Lam3, Zhongping Chen3, M Abraham Kuriakose1,2, Petra Wilder-Smith3 1Head and Neck Oncology, Mazumdar Shaw Medical Center, Bangalore, India. 2Integrated Head and Neck Oncology Research Program, Mazumdar Shaw Center for Translational Research, MSMF, Bangalore, India, 3University of California, Irvine, 4KLES Institute of Dental Sciences, Bangalore, India Abstract: Background: Long-term goal of this project is to develop and validate a robust, portable, low-cost tool for rapid oral cancer screening in the field. Specific aim was to evaluate the diagnostic ease of use and diagnostic accuracy of a novel prototype OCT-based device and probe in the US and in India. Materials and Methods: Patients referred for evaluation of suspect oral lesions underwent routine clinical exam, OCT imaging, photography, and biopsy when indicated by standard of care. Typically, OCT imaging had a duration of a few seconds per image and location. To date we have acquired images and biopsy data in more than 1000 subjects in India and the US. Images underwent 2 forms of evaluation: (1) visual exam by 2 pre-standardized OCT scorers and (2) automated scoring via a simple diagnostic algorithm using reflectivity and thickness ratios of superficial anatomical structures. Results and Discussion:Agreement (kappa) between visual OCT scoring and histopathology was 86%, and between scoring provided by the diagnostic algorithm vs histopathology was 89%. PPV measured 85/91%, NPV 84/87%, diagnostic sensitivity 83/87% and specificity 84/85%. Future improvements in imaging performance and diagnostic algorithm will be based on these data and should enhance diagnostic performance. Conclusion: A novel low-cost OCT-based screening approach may improve early detection of oral cancer. This research was supported by the National Institutes of Health under grant No. P41EB015890 (Laser Microbeam and Medical Program: LAMMP) and No. 1R03EB014852 (National Institute of Biomedical Imaging and Bioengineering: NIBIB), and the Beckman Foundation.

• 80% of oral cancers are preceded by OPML and Biopsy and histopathological grading of dysplasia is considered as gold-standard for diagnosis.

• Being an invasive procedure and the need for expert pathologist, biopsies and conventional histology are not a feasible tool for screening and surveillance in a large volume, high-risk population.

• OCT technology uses near infrared light to obtain real-time images of in-vivo tissue microstructure. This is proved to be effective in ocular pathology.

• The goal of the present study is to develop an OCT probe compatible to image oral cavity and to determine its effectiveness to diagnose OPML and malignant lesions.

• Re-engineer a small, robust portable diagnostic system for oral cancer on OCT • Develop and test the diagnostic algorithm for OCT

Data Collection

Work flow

Inter-observer agreement

Sensitivity and specificity

Portable OCT instrument

•Penetration depth : 1 mm to 2 mm

•2 images per second.

•Diameter of the probe is roughly 5 mm.

Schematic diagram of the proposed swept source based OCT

Study design

Field screening was performed over 1 year by field workers in India with 3rd-grade equivalent education. They were trained for 1 day, and observed continuously over the first month of this study. They were provided with a wireless-enabled cell phone for data upload, as well as direct communication link with an oral oncologist at the Mazumdar Shaw Cancer Center in Bangalore. De-identified images as well as clinical, photographic and medical history data mages were automatically transmitted every night via Drobo system to a server at the University of California, Irvine. There, images were (1) analyzed visually by 2 pre-standardized blinded scorers and (2) scored by an automated diagnostic algorithm. Gold standard was histopathology.

Alpha values

Normal -1.5 to -2.2

Dysplasia -1 to -1.6

Malignant -.8 to -1.3

Introduction

Objectives

Optical Coherence Tomography (OCT)

Methodology

Results

Conclusion Discussion

AUC 0.833 P value <0.0003

Malignant

Our novel device demonstrated the potential for detecting oral dysplasia and malignancy earlier and improving oncologic outcomes in low-resource, remote, non-specialist settings. It is ideal for non-invasive, point-of-care community-based screening of high-risk subjects enabling earlier detection of premalignant and malignant lesions in patients.

A simple, non-surgical imaging-based approach may greatly improve screening and detection of oral pre-cancer and cancer.