Clinical applications of artificial intelligence in urologic oncology

Assessment of Prostate and Bladder Cancer Genomic Biomarkers Using Artificial Intelligence: a Systematic Review

PURPOSE OF REVIEW

The aim of the systematic review is to assess AI's capabilities in the genetics of prostate cancer (PCa) and bladder cancer (BCa) to evaluate target groups for such analysis as well as to assess its prospects in daily practice.

RECENT FINDINGS

In total, our analysis included 27 articles: 10 articles have reported on PCa and 17 on BCa, respectively. The AI algorithms added clinical value and demonstrated promising results in several fields, including cancer detection, assessment of cancer development risk, risk stratification in terms of survival and relapse, and prediction of response to a specific therapy. Besides clinical applications, genetic analysis aided by the AI shed light on the basic urologic cancer biology. We believe, our results of the AI application to the analysis of PCa, BCa data sets will help to identify new targets for urological cancer therapy. The integration of AI in genomic research for screening and clinical applications will evolve with time to help personalizing chemotherapy, prediction of survival and relapse, aid treatment strategies such as reducing frequency of diagnostic cystoscopies, and clinical decision support, e.g., by predicting immunotherapy response. These factors will ultimately lead to personalized and precision medicine thereby improving patient outcomes.

Artificial intelligence to improve cytology performance in urothelial carcinoma diagnosis: results from validation phase of the French, multicenter, prospective VISIOCYT1 trial

PURPOSE

Cytology and cystoscopy, the current gold standard for diagnosing urothelial carcinomas, have limits: cytology has high interobserver variability with moderate or not optimal sensitivity (particularly for low-grade tumors); while cystoscopy is expensive, invasive, and operator dependent. The VISIOCYT1 study assessed the benefit of VisioCyt® for diagnosing urothelial carcinoma.

METHODS

VISIOCYT1 was a French prospective clinical trial conducted in 14 centers. The trial enrolled adults undergoing endoscopy for suspected bladder cancer or to explore the lower urinary tract. Participants were allocated either Group 1: with bladder cancer, i.e., with positive cystoscopy or with negative cystoscopy but positive cytology, or Group 2: without bladder cancer. Before cystoscopy and histopathology, slides were prepared for cytology and the VisioCyt® test from urine samples. The diagnostic performance of VisioCyt® was assessed using sensitivity (primary objective, 70% lower-bound threshold) and specificity (75% lower-bound threshold). Sensitivity was also assessed by tumor grade and T-staging. VisioCyt® and cytology performance were evaluated relative to the histopathological assessments.

RESULTS

Between October 2017 and December 2019, 391 participants (170 in Group 1 and 149 in Group 2) were enrolled. VisioCyt®'s sensitivity was 80.9% (95% CI 73.9-86.4%) and specificity was 61.8% (95% CI 53.4-69.5%). In high-grade tumors, the sensitivity was 93.7% (95% CI 86.0-97.3%) and in low-grade tumors 66.7% (95% CI 55.2-76.5%). Sensitivity by T-staging, compared to the overall sensitivity, was higher in high-grade tumors and lower in low-grade tumors.

CONCLUSION

VisioCyt® is a promising diagnostic tool for urothelial cancers with improved sensitivities for high-grade tumors and notably for low-grade tumors.

Beyond high hopes: A scoping review of the 2019-2021 scientific discourse on machine learning in medical imaging

Machine learning has become a key driver of the digital health revolution. That comes with a fair share of high hopes and hype. We conducted a scoping review on machine learning in medical imaging, providing a comprehensive outlook of the field's potential, limitations, and future directions. Most reported strengths and promises included: improved (a) analytic power, (b) efficiency (c) decision making, and (d) equity. Most reported challenges included: (a) structural barriers and imaging heterogeneity, (b) scarcity of well-annotated, representative and interconnected imaging datasets (c) validity and performance limitations, including bias and equity issues, and (d) the still missing clinical integration. The boundaries between strengths and challenges, with cross-cutting ethical and regulatory implications, remain blurred. The literature emphasizes explainability and trustworthiness, with a largely missing discussion about the specific technical and regulatory challenges surrounding these concepts. Future trends are expected to shift towards multi-source models, combining imaging with an array of other data, in a more open access, and explainable manner.