A new study suggests that artificial intelligence (AI) can significantly enhance the use of apparent diffusion coefficient (ADC) mapping in detecting prostate cancer (pCa) with magnetic resonance imaging (MRI).
For the study, recently published in Academic Radiology, researchers assessed the impact of AI-generated ADC mapping from MRI to conventional ADC mapping in a cohort of 178 patients. In addition to T2-weighted 3T MRI, study participants had high b-value diffusion-weighted biparametric MRI (bpMRI) prior to biopsy or radical prostatectomy, according to the study.
The researchers found that AI-generated ADC mapping offered comparable sensitivity to conventional ADC mapping for (93 percent vs. 94 percent) for PCa along with enhanced accuracy (78 percent vs. 70 percent), positive predictive value (PPV) (78 percent vs. 71 percent) and negative predictive value (NPV) (78 percent vs. 65 percent).
AI-generated ADC mapping also offered more than double the specificity of conventional ADC mapping (47 percent vs. 22 percent), according to the study authors.
“ … Conventional ADC maps are often degraded by noise and artifacts, especially at high b-values, which can obscure the true diffusion characteristics of malignant lesions and limit diagnostic accuracy. AI-ADC maps potentially overcome these limitations by learning to emphasize biologically meaningful contrast patterns while suppressing irrelevant noise or benign confounders. This enhancement may contribute to improved tumor delineation, particularly in the peripheral zone where inflammation or benign prostatic hyperplasia can mimic cancer on standard imaging,” wrote lead study author Kutsev B. Ozyaruk, Ph.D., who is affiliated with the Molecular Imaging Branch of the National Cancer Institute in Bethesda, Md., and colleagues.
(Editor’s note: For additional content on prostate cancer imaging, click here.)
Three Key Takeaways
1. Improved diagnostic accuracy and specificity. AI-generated ADC maps provided significantly improved overall accuracy (78 percent vs. 70 percent) and more than double the specificity (47 percent vs. 22 percent) compared to conventional ADC maps, reducing false positives, particularly in benign conditions like prostatitis or hyperplasia.
2. Noise reduction and biological relevance. AI mapping improved lesion visualization by suppressing noise and benign confounders, addressing the limitations of standard high b-value DWI that often degrade image quality, especially in the prostate's peripheral zone.
3. Enhanced performance in early-stage disease. The AI-ADC approach doubled the sensitivity in detecting Gleason Grade Group 1 cancers (80 percent vs. 40 percent), suggesting better detection of early, potentially curable prostate cancers.
Addressing the increased specificity with the generative AI model, the researchers noted the use of spatial and channel attention mechanisms facilitated a 13 percent increase with NPV and a 25 percent increase in identifying negative cases. The study authors also noted a doubling of sensitivity for patients with a Gleason grade group 1 presentation.
“Specifically, we examined the sensitivity for Gleason Grade Group 1, where conventional ADC maps demonstrated a sensitivity of 40%, compared to an 80% sensitivity achieved using AI-ADC maps,” added Ozyaruk and colleagues.
(Editor’s note: For related content, see “Study: MRI-Based AI Enhances Detection of Seminal Vesicle Invasion in Prostate Cancer,” “New bpMRI Study Suggests AI Offers Comparable Results to Radiologists for PCa Detection” and “AI and bpMRI for csPCa Detection: What a New Meta-Analysis Reveals.”)
Beyond the inherent limitations of a single-center study, the authors acknowledged that MRI scans with endorectal coils weren’t included in the training dataset but were included in the evaluation of the AI ADC model. They also conceded the possibility of undetected PCa in some control patients despite negative results with 12-core systematic biopsies.
