Design of a multispectral digital colposcope

Abstract

Measurement quality assurance plans for optical devices should be a mandatory part of grant funding submissions and should explicitly affect scoring during review. These should include calibration strategy, standards selection strategy, performance verification plan, performance validation plan and thorough preclinical performance validation. A multispectral digital colposcope (MDC) has been designed to collect image data from patients as part of an NIH sponsored clinical trial, based on a technology assessment model. Calibration strategy, standards selection and performance verification methods are presented that may be used as a template for smaller groups or more limited studies. With the MDC, red green and blue fluorescence images are captured under ultraviolet light excitation and red and green images are captured under blue light excitation. Red, green and blue reflectance images are captured under broadband white light illumination from a metal halide lamp in three modes - ordinary reflectance, and with polarized illumination in combination with parallel and cross-polarized filtered imaging. The highly automated system was designed to collect images of the cervix prior to and following the application of acetic acid. Three systems have been built and will be operated in clinics in Vancouver, Canada, Houston, Texas and other locations in the developed and developing world including Nigeria. The system is designed with a comprehensive set of calibration and performance verification standards, based on our experience with large scale multi-center spectroscopy clinical trials and measurements are made frequently prior to and following patient measurements. Automated performance verification procedures are being designed based on measurements made during pilot studies to facilitate larger clinical trials.