This site is intended for UK healthcare professionals only



Keep up to date with the latest news on bladder health, management and treatment.

Have you got a story to share? Email the team today.

AI & bladder cancer diagnosis
24 January 2022
Could artificial intelligence aid bladder cancer diagnosis and outcome prediction?

A recent study reports that AI models have shown great promise in bladder cancer diagnosis and outcome prediction.

A number of studies have used Artificial Intelligence (AI) algorithms for clinical tasks involved in BCa diagnosis and outcome prediction. These include automatic tumour detection, staging, and grading, bladder wall segmentation, as well as prediction of recurrence, response to chemotherapy, and overall survival. There have been promising results reported but AI algorithms have not yet been fully integrated into the clinical workflow.

The study provides an accessible introduction to the fundamental nomenclature and concepts in AI. It also reviews the literature to explore how AI is used for bladder cancer diagnosis and outcome prediction, and it presents a perspective on the obstacles that must be removed before AI algorithms can enter the mainstream of cancer management.

Machine Learning

  • Machine Learning (ML) is a branch of Artificial Intelligence (AI) in which computers, rather than being explicitly programmed with a pre-determined set of rules or instructions, leverage data to ‘learn’ how to perform a given task.

AI for bladder cancer diagnosis

  • Tumour detection – White Light Imaging (WLI) remains the primary mode of cystoscopy in patients suspected of BCa despite relatively high diagnostic error rates. Since AI models are inherently impervious to many sources of human error, their use as a physician assistant tool may prove beneficial to increase diagnostic accuracy during cystoscopy.

  • Bladder segmentation – or the automatic differentiation of bladder from its surrounding anatomical structures in pelvic imaging studies, is a necessary first step towards developing computer-aided diagnosis (CAD) systems for BCa.

  • Tumour staging – Management of patients with BCa relies heavily on the stage of the tumor as determined by the depth of its penetration into various layers of the bladder. In particular, depending on whether the tumor has reached the detrusor muscle or not, patients fall into one of the two categories of muscle-invasive (MIBC) or non-muscle-invasive BCa (NMIBC), with a different set of clinical intervention guidelines recommended for each group.

  • Tumour grading – While choosing the proper BCa treatment strategy is largely driven by the stage of the tumor, its pathological grade also plays an important role in clinical decision-making considering that approximately 30 % of NMIBC cases consist of high grade tumors which can progress to muscle invasion and metastasis during the patient’s lifetime. 

AI for BCa outcome prediction

  • Prediction of survival – The 5-year survival rate of BCa is estimated around 90% for NMIBC patients (Kaufman et al., 2009), but it drops drastically for patients with MIBC as the tumor invades different layers of the bladder. Therefore, development of models that can accurately estimate the risk of disease-specific death for an individual patient would aid oncologists in devising proper treatment strategies.
  • Prediction of recurrence – Accurate and reliable prediction of tumor recurrence is critical in NMIBC and MIBC patients as both groups experience relatively high rates of recurrence. One study built a nonlinear SVM classification model to identify patients at risk of recurrence within two years post-operation.
  • Prediction of response to chemotherapy – it is possible in bladder cancer to develop a reliable predictive model that could identify patients who would only suffer from the adverse side effects of chemotherapeutic drugs without gaining any significant benefits from them. 

The study concluded, “The use of AI technology in BCa diagnostics and prognostics has the potential to improve patients’ quality of life by preventing unnecessary radical cystectomies, and to reduce the financial burden of BCa as the most expensive malignancy to treat over the patients’ lifetime. To realize this potential, further research is required in the form of additional observational studies as well as new multi-phase clinical trials.”

Read the full study here

Read more news from The Bladder Interest Group now.

Latest News

Luja™ shows significant improvement in bladder emptying

Study results show that Coloplast’s new intermittent male catheter with 80+ micro-holes, Luja, achieved complete bladder emptying in one free flow* in 90% of catheterisations. Coloplast has finalised its first pivotal clinical study on Luja, a new intermittent male...

Artificial Intelligence used in bladder cancer treatment assessment

A computerised artificial intelligence (AI)-based decision support system (CDSS-T) enhanced the performance of clinicians when assessing patients’ response to chemotherapy prior to radical cystectomy, according to results from a small multi-institutional observer...

New treatment for urothelial cancer approved by NICE

Over 800 people with locally advanced or metastatic urothelial cancer are set to benefit from a new treatment for urothelial cancer – avelumab – following its recommendation for routine NHS funding in final draft guidance published NICE. The recommendation for this...