Explainable artificial intelligence (XAI): closing the gap between image analysis and navigation in complex invasive...

Area of Science:

• Urology
• Medical Technology
• Artificial Intelligence

Background:

• Cystoscopy is the standard for diagnosing bladder cancer (BCa), but its findings can be difficult to interpret, leading to potential misdiagnoses.
• The rates of false negatives and false positives in current cystoscopy practices are not well-defined, posing risks of under- or over-diagnosis.
• These diagnostic inaccuracies can lead to delayed cancer treatment or unnecessary procedures.

Purpose of the Study:

• To explore the potential of explainable artificial intelligence (XAI) robot-assisted cystoscopes to improve bladder cancer diagnosis.
• To establish a framework for semi-autonomous cystoscopy that can be a model for other endoscopic and surgical procedures.
• To address the limitations of conventional cystoscopy by introducing automation and enhancing diagnostic accuracy.

Main Methods:

• Review of current cystoscopy practices and their limitations in diagnosing bladder cancer.
• Conceptualization of XAI robot-assisted cystoscopy systems.
• Discussion of automation levels and the 'human-in-the-loop' approach for safety in semi-autonomous procedures.

Main Results:

• XAI robot-assisted cystoscopy has the potential to mitigate the risks and flaws associated with conventional cystoscopy.
• Semi-autonomous cystoscopy can establish standards for automation in medical procedures.
• A human supervisor remains essential ('human-in-the-loop') for patient safety in robotic cystoscopy.

Conclusions:

• Robot-assisted cystoscopy offers a safer, more accurate method for bladder cancer diagnosis compared to traditional methods.
• The development of standards for semi-autonomous cystoscopy can pave the way for automation in other medical fields.
• Automated diagnostic cystoscopy allows urologists to review findings efficiently, potentially delegating routine procedures to specialized nurses.