# NIH grant: Advancing Clinically Translatable Deep Learning for Prostate Cancer Detection on MRI (1R01CA307281-01A1) — retained record

> Retained us-gov-api record from us-gov-api:nih-reporter.

Canonical: https://highsignal.app/data/records/69cbfe8dbcdf8b0a

[← sources](https://highsignal.app/data)


source: [us-gov-api](https://highsignal.app/data/us-gov-api) (us-gov-api:nih-reporter)  published 2026-07-16 00:00 ingested 2026-07-20 08:52

[original source ↗](https://reporter.nih.gov/project-details/1R01CA307281-01A1)

## retained text



```
NIH grant 1R01CA307281-01A1: Advancing Clinically Translatable Deep Learning for Prostate Cancer Detection on MRI
Institute: {'code': 'CA', 'abbreviation': 'NCI', 'name': 'National Cancer Institute', 'admin_org_id': '12500008', 'admin_funding_url': ''}. PI: Patricia  Johnson. Award: $700171.
Start: 2026-07-16T00:00:00.
SUMMARY
Prostate cancer (PCa) is one of the most common malignancies and a leading cause of mortality among men in the United States. Magnetic Resonance Imaging (MRI) has transformed PCa detection by improving the identification of clinically significant prostate cancer (csPCa), reducing unnecessary biopsies, and improving PCa management and surveillance. However, interpretation of prostate MRI remains challenging, with false positives leading to unnecessary interventions, false negatives delaying critical diagnoses, and significant variability among radiologists. These issues hinder the full potential of MRI in clinical practice, emphasizing the urgent need for tools that improve consistency, reliability, and efficiency in MRI interpretation. Deep learning (DL) models hold immense promise for addressing these challenges by enhancing diagnostic accuracy. However, existing models have not seen widespread clinical use due to concerns about their reliability, generalizability, and inability to provide actionable confidence measures for predictions. Furthermore, the lack of mechanisms to adapt to evolving clinical environments and demographics limits the long-term utility of current DL solutions. This project seeks to overcome these barriers by advancing DL-based csPCa detection on MRI through three specific aims. First, we will train a novel and robust DL model on diverse imaging datasets, ensuring consistent performance across a wide range of scanners, protocols, and patient populations (Aim 1). Next, we will incorporate uncertainty quantification and out-of-distribution detection to enhance the reliability of model predictions and provide actionable insights for clinical decision-making (Aim 2). Finally, we will evaluate the clinical utility of AI assistance and confidence visualization in a multi-reader study (Aim 3). By addressing these critical barriers, this work aims to establish DL as a reliable tool for improving prostate MRI interpretation, reducing inter-reader variability, and increasing diagnostic precision. The resulting framework will not only enhance the safety and consistency of csPCa detection but also provide a scalable foundation for integrating AI into routine clinical workflows. Beyond prostate cancer, these methods can be adapted to other imaging-based disease detection tasks, advancing the broader application of artificial intelligence in healthcare.
```



document fetched 2026-07-26 08:23