Bladder image stitching algorithm for navigation and referencing using a standard cystoscope.

Area of Science:

• Urology
• Medical Imaging
• Computational Biology

Background:

• Bladder carcinoma diagnosis and monitoring require accurate anatomical referencing.
• Standard cystoscopy videos lack a systematic method for creating referenceable bladder maps.
• Specialized equipment is often needed for detailed bladder imaging.

Purpose of the Study:

• To develop and validate a vision-based algorithm for generating 2D bladder maps from standard cystoscopy videos.
• To enable referenceable bladder mapping without specialized equipment for improved diagnosis and surveillance of bladder carcinoma.
• To assess the algorithm's performance in generating and reevaluating bladder maps in preclinical and clinical settings.

Main Methods:

• A vision-based algorithm was created to stitch 2D image frames from cystoscopy videos based on surface features.
• The algorithm was calibrated using 2D models and a 3D bladder phantom.
• In vivo evaluation was performed in swine, alongside retrospective analysis of human clinical cystoscopy videos.

Main Results:

• The algorithm successfully captured and stitched intravesical images, generating bladder maps from 93% to 99% of frames.
• Accurate frame localization was achieved within 4.5 seconds during reevaluation.
• Generated virtual mucosal surface maps in swine correlated with explant anatomy, and maps were created from archived patient images.

Conclusions:

• The developed algorithm effectively generates referenceable 2D bladder maps from standard cystoscopy videos.
• This tool can aid in recording and referencing pathological findings, biopsy sites, and treatment locations.
• The technology shows potential utility for managing metachronous bladder cancer and in resource-limited settings.