PROVIZ - a Machine Learning Software, to Support Targeting of Prostate Biopsies on MR Images in Biopsy-naive Patients

A Proof-of-technology, Pilot, Prospective Clinical Study to Investigate the Feasibility and Performance of PROVIZ-a Radiomics-based Machine Learning Software, to Support Targeting of Prostate Biopsies on MRI Images in Biopsy-naive Patients

To perform a traditional feasibility clinical investigation, as defined in ISO 14155:2020, to investigate preliminary feasibility, safety, and clinical performance information of a near-final design of the investigational software. This will be performed through a prospective clinical study on biopsy naïve men with suspected prostate cancer examined with MRI at St. Olavs Hospital, Trondheim, Norway, in order to adequately plan an appropriate pivotal clinical investigation.

Study Overview

• Status: Completed
• Conditions: Prostate Cancer Diagnosis
• Intervention / Treatment: Device: Automatic image analysis

Detailed Description

In this prospective study, after referral for suspected prostate cancer, the patient is scanned with magnetic resonance imaging (MRI) in accordance with guidelines of the standardized healthcare pathway. For consenting patients, the images are interpreted in two ways: first, in the conventional manner, i.e., manually by a radiologist, to determine whether clinically significant cancer is suspected. If so, the radiologist will delineate the suspicious lesions. In the second approach, the software will perform the same task as the radiologist, but automatically. If either or both interpretations point to significant cancer, the patient will be sent for targeted biopsy sampling. Histopathologic evaluation of the samples will then determine whether significant cancer is present in each of the targeted lesions (delineated by the radiologist, software, or both).

Feasibility is evaluated by measuring the overall failure rate of the software. This is measured by the technical performance log automatically generated by the software, which records all errors and failures during the study. If the record showed an overall failure rate less than 10% across all subjects, the software will be considered feasible.

The safety of the software is evaluated by the records of the serious adverse device effects (SADEs) of the software during the study. The software will be considered safe with no occurrence of death or serious injury.

The results of the histopathological evaluation will be used to evaluate the performance of the investigational software. This allows comparisons to be made between results obtained with the traditional manual approach alone, with the software alone, and with the manual approach assisted by the software. Statistical analysis will be performed to determine if there are significant differences and if the software adds value.

• Study Type: Interventional
• Enrollment (Actual): 82
• Phase: Not Applicable

Contacts and Locations

Study Locations

• Norway
  • Trondheim, Norway
    • St. Olav's University Hospital

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

• Biopsy-naive men undergoing MRI examination for suspected prostate cancer via the Norwegian standardized care pathway.
• Patients who give consent to participate during the enrollment period.

Exclusion Criteria:

• Patients who have undergone a biopsy for prostate cancer in the past 3 years.
• Patients who currently enrolled in an active surveillance program for prostate cancer.
• Patients who have had hip replacements that may affect the quality of the image.
• Patients with claustrophobia.
• Patients who intolerance to glucagon or buscopan.
• Patients who unable to sign the informed consent themselves.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Diagnostic
• Allocation: N/A
• Interventional Model: Single Group Assignment
• Masking: None (Open Label)

Number of Arms

1

Arms and Interventions

Participant Group / Arm

Intervention / Treatment

Experimental: MRI for prostate cancer; Biopsy-naive men undergoing MRI for suspected prostate cancer via the Norwegian standardized care pathway. The MRI images will then be evaluated by the radiologist and a machine-learning based diagnosis and detection system.

Device: Automatic image analysis; After referral for suspected prostate cancer, the patient is scanned with magnetic resonance imaging (MRI) in accordance with guidelines of the standardized healthcare pathway. For consenting patients, the images are interpreted in two ways: first, in the conventional manner, i.e., manually by a radiologist, to determine whether clinically significant cancer is suspected. If so, the radiologist will delineate the suspicious lesions. In the second approach, the software will perform the same task as the radiologist, but automatically. If either or both interpretations point to significant cancer, the patient will be sent for targeted biopsy sampling. Histopathologic evaluation of the samples will then determine whether significant cancer is present in each of the targeted lesions (delineated by the radiologist, software, or both).

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Feasibility of the software in prospective study in a relevant clinical environment.	The percentage of occurrence of a technical problem in the investigational software that hinders its use is less than 10% across all subjects.	Approximately 45 days. Actual time in clinic is approximately 1.5 hours.
Safety of the software in prospective study in a relevant clinical environment.	There are no death or serious harm associated with an adverse device effect (ADE) of the investigational software.	Approximately 45 days. Actual time in clinic is approximately 1.5 hours.

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Performance of the software in prospective study in a relevant clinical environment.	Comparable preliminary detection rate of clinically significant prostate cancer lesions from the software to the radiologist.	Approximately 45 days. Actual time in clinic is approximately 1.5 hours.

Collaborators and Investigators

• Sponsor: Norwegian University of Science and Technology
• Collaborators: St. Olavs Hospital
• Investigators: Principal Investigator: Tone Frost Bathen, Prof., Norwegian University of Science and Technology (NTNU)

Study record dates

Study Major Dates

• Study Start (Actual): December 20, 2023
• Primary Completion (Actual): October 13, 2024
• Study Completion (Actual): October 13, 2024

Study Registration Dates

• First Submitted: August 14, 2023
• First Submitted That Met QC Criteria: August 14, 2023
• First Posted (Actual): August 21, 2023

Study Record Updates

• Last Update Posted (Actual): March 25, 2025
• Last Update Submitted That Met QC Criteria: January 21, 2025
• Last Verified: January 1, 2025

More Information

Terms related to this study

• Keywords: Magnetic Resonance Imaging; Artificial Intelligence; Prostate; Machine Learning; Radiomics; Computer-Aided Diagnosis
• Additional Relevant MeSH Terms: Pathologic Processes; Disease
• Other Study ID Numbers: 479272; CIV-NO-23-04-042915 (Other Identifier: The Norwegian Medicines Agency); REK KULMU A # 479272 (Other Identifier: REK KULMU)

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No