Association between artificial intelligence–derived tumor volume and oncologic outcomes in localized prostate cancer.
Bibliographic record
Abstract
280 Background: Characteristics of the intraprostatic tumor (e.g., PI-RADS scores) from multi-parametric magnetic resonance imaging (mpMRI) are prognostic but exhibit significant inter-observer variability. An artificial intelligence (AI)-based method for measuring intraprostatic tumor volume from mpMRI may provide prognostic information in a systematic manner. We sought to evaluate whether intraprostatic tumor volume (VAI) from AI-segmented lesions provides prognostic information independent of clinical risk groups for contemporary patients with localized prostate cancer (PCa) treated with radiation therapy (RT) or radical prostatectomy (RP). Methods: We retrospectively identified 732 patients with cT1-3N0M0 PCa who underwent high B-value, 3-Tesla mpMRI and were treated with RT (n=438) in 2010-2017 or RP (n=294) in 2015-2017 at a single academic institution. The RT cohort was randomly divided into cross-validation RT (CVRT, n=288) and TestRT (n=150) cohorts. A deep learning model (nnU-Net) was trained to delineate PI-RADS 3-5 lesions from the CVRT cohort using 5-fold cross validation before providing delineations for TestRT and RP cohorts. The F1 score (geometric mean of precision and recall) for identifying patients with PI-RADS 3-5 lesions was calculated for each cohort. Multivariable Cox regression models were used to evaluate the association between VAI and time to biochemical failure and metastasis, while adjusting for 4-tiered National Cancer Center Network risk groups. The RT (CVRT and TestRT) and RP cohorts were analyzed separately. Results: F1-scores for the CVRT, TestRT, and RP datasets were 87.4%, 83.8%, and 85.3%, respectively. With a median follow-up of 6.9 years for the RT cohort, VAI was significantly associated with biochemical failure (adjusted hazard ratio [AHR] 1.58, 95% confidence interval [CI] 1.29-1.94, p<0.001) and metastasis (AHR 1.70, 95% CI 1.24-2.34, p=0.001). With a median follow-up of 5.5 years for the RP cohort, VAI was similarly significantly associated with biochemical failure (AHR 1.33, 95% CI 1.12-1.59, p=0.001) and metastasis (AHR 1.93, 95% CI 1.19-3.13, p=0.008). The areas under the receiver operating characteristic curves for 7-year metastasis for the RT cohort were 0.867 for VAI versus 0.821 for NCCN (p=0.03). Corresponding values at 5 years for the RP cohort were 0.888 vs 0.789 (p=0.25). Conclusions: An AI model can delineate the intraprostatic tumor with excellent performance. The intraprostatic tumor volume of AI-segmented lesions is an independent, highly prognostic factor that can be derived from standard mpMRI, even prior to biopsy. If further validated, this noninvasive biomarker may enable further treatment personalization for localized PCa.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".