Low-cost and label-free blue light cystoscopy through digital staining of white light cystoscopy videos

Abstract

Bladder cancer is the 10th most common malignancy and carries the highest treatment cost among all cancers. The elevated cost stems from its high recurrence rate, which necessitates frequent surveillance. White light cystoscopy (WLC), the standard of care surveillance tool to examine the bladder for lesions, has limited sensitivity for early-stage bladder cancer. Blue light cystoscopy (BLC) utilizes a fluorescent dye to induce contrast in cancerous regions, improving the sensitivity of detection by 43%. Nevertheless, the added equipment cost and lengthy dwell time of the dye limits the availability of BLC. Here, we report the first demonstration of digital staining as a promising strategy to convert WLC images collected with standard-of-care clinical equipment into accurate BLC-like images, providing enhanced sensitivity for WLC without the associated labor or equipment cost. By introducing key pre-processing steps to circumvent color and brightness variations in clinical datasets needed for successful model performance, the results achieve a staining accuracy of 80.58% and show excellent qualitative and quantitative agreement of the digitally stained WLC (dsWLC) images with ground truth BLC images, including color consistency. In short, dsWLC can affordably provide the fluorescent contrast needed to improve the detection sensitivity of bladder cancer, thereby increasing the accessibility of BLC contrast for bladder cancer surveillance. The broader implications of this work suggest digital staining is a cost-effective alternative to contrast-based endoscopy for other clinical scenarios outside of urology that can democratize access to better healthcare. Bladder cancer is one of the most common and costly cancers to treat. Traditional white light imaging of the bladder is not very effective at detecting early-stage cancer. Blue light imaging is better able to detect these cancers but requires administration of a dye. In this study, we use a computational process to transform white light bladder images into fluorescent, blue light versions, which improves detection of early-stage cancers. Our approach may be applicable to other clinical uses and could potentially be used to improve diagnosis of cancer. Chang et al. convert white light cystoscopy (WLC) images collected with standard-of-care clinical equipment into accurate blue light cystoscopy (BLC)-like images. By introducing key pre-processing steps to circumvent color and brightness variations in clinical datasets, they provide enhanced sensitivity without labor or equipment cost.