What benefit can be obtained from magnetic resonance imaging diagnosis with artificial intelligence in prostate cancer compared with clinical assessments?

Table 4 Comparison of AI and clinical assessment methods in the prediction field of PCa

Comparison	AI methods		Clinical assessment methods
Comparison	HC model	DL model	Clinical assessment methods
Overall performance	Relatively high		Relatively poor
SROC-AUC^*	0.86		0.75
Pooled sensitivity^*	0.75		0.68
Pooled specificity^*	0.84		0.79
Qualitative or quantitative	Quantitative		Quantitative/qualitative
Expert dependence	Moderate	Low	High
Consistency	High		Moderate
Manual delineation	Yes	No	No
Features	High-throughput features extracted using specific algorithms (e.g., shape, histogram and textural features)	Automatic extraction of deep and subtle image features using networks with substantial parameters	Clinical characteristics (e.g., PSA, Gleason grade and positive biopsy cores) or features for visual assessments (e.g., location, shape, size, and intensity)

AI artificial intelligence, AP adverse pathology, DL deep learning, HC hand-crafted, PCa prostate cancer, PSA prostate specific antigen, SROC-AUC area under the summary receiver operator characteristic curves
^*Performance indexes pooled across the studies of AP prediction

ISSN: 2054-9369