Cancer of the Prostate, Neoplasms, Prostate, Prostate Cancer, Prostatic Neoplasm
Conditions
Keywords
Artificial Intelligence, Magnetic Resonance Imaging, Biopsy
Brief summary
This study will enroll participants who are undergoing an MRI before a prostate biopsy due to suspected prostate cancer. The purpose of this study is to see if the use of Artificial Intelligence (AI) helps detect lesions on an MRI better than a radiologist not using AI. The AI Rad Companion (AIRC) Prostate MRI application is a software that uses measurements of the prostate and will be utilized in this study to help detect potential cancerous lesions. The AI software will assign the lesions a PI-RADS score, which is a way to measure the chance of the lesion being cancer. There are two parts to this study. The first part involves comparing the interpretation of prostate MRI images by a radiologist alone, a radiologist aided by AI, and AI alone. A systematic biopsy will be completed per standard of care. The radiologist may opt to include up to 2 additional AI-identified targets to biopsy in addition to those biopsied for standard of care. The second part of the study involves utilizing the MRI images from the first part of the study in addition to retrospective prostate MRI images. These de-identified images, along with Prostate Image Quality (PI-QUAL) scores, clinical data, and biopsy results will be sent to Siemens in order to aid in the development of methods to identify good or bad image quality in prostate MRI images.
Detailed description
Prostate cancer is the most diagnosed cancer among men in the United States and the second most prevalent cancer in men worldwide (1,2). MRI can better triage patients to undergo biopsy, while providing information about tumors and treatment progress (4). However, there are challenges with MRI image quality, the interpretation of the images, and the potential for false positives and negatives. With the annual number of prostate exams rapidly increasing (5), there is a need for accurate and reliable prostate MRI interpretation. One proposed approach is to use an AI-based lesion detection software in addition to radiologists' interpretations. The Siemens Healthineers ("Siemens") business lines Digital & Automation ("D&A") and Magnetic Resonance ("MR") have developed the AI Rad Companion Prostate MRI (AIRC), which aids in the detection of potential prostate cancer. The goal of this study is to investigate the effects of the use of AIRC in improving the standard of care in prostate cancer detection and treatment. Participation in the study will be about an hour to review the consent form. All other study procedures (MRI, biopsy) will be standard of care.
Interventions
DEVICEAI
Following the completion of a pre-biopsy prostate MRI, the radiologist will interpret the MRI. Once interpreted by the radiologist alone, the radiologist will interpret the scan while aided by AI. A systematic biopsy in conjunction with radiologist-identified targets will be completed per standard of care, with the optional inclusion of up to 2 AI-detected targets. When completing a biopsy per SOC, the prostate is divided into quadrants. In addition to noted targets, samples are taken systematically from each quadrant. If targets are detected by AI that were not identified by the physician when reviewing the MRI, these targets will be sampled. Sampling of these targets will not be in addition to the systematic sampling in each quadrant, but in place of up to two of the samples biopsied systematically
Sponsors
Case Comprehensive Cancer Center
Siemens Medical Solutions
Study design
Allocation
NA
Intervention model
SINGLE_GROUP
Primary purpose
SCREENING
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
55 Years to 80 Years
Healthy volunteers
No
Inclusion criteria
* Plan of care is to undergo a biopsy of the prostate after a pre-biopsy MRI * Age 55-80 * Prostate-specific antigen (PSA) between 3-10 ng/mL * No prior diagnosis or treatment of prostate cancer
Exclusion criteria
* Pre-biopsy MRI is of low quality * PI-QUAL score of 1 using PI-QUAL version 2
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Readers' (radiologists') mean quadrant-level area under the receiver operating characteristic curve (AUC) in predicting the presence or absence of clinically significant prostate cancer (csPCa) | One-time MRI, up to 30 days post-enrollment in study. | csPCa is defined as Gleason grade group ≥ 2. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Inter-reader agreement on the presence/absence of csPCa at the participant-level | One-time MRI, up to 30 days post-enrollment in study. | Agreement is defined as agreement between unaided vs aided reads. If the highest lesion-level PI-RADS assigned to a patient is greater than or equal to 3, the participant is considered positive. |
| Inter-reader agreement on the presence/absence of csPCa at the quadrant-level | One-time MRI, up to 30 days post-enrollment in study. | Agreement is defined as agreement between unaided vs aided reads. If the highest lesion-level PI-RADS assigned to a quadrant is greater than or equal to 3, the quadrant is considered positive. |
| Participant-level sensitivity for detection of clinically significant prostate cancer | One-time MRI, up to 30 days post-enrollment in study. | If the highest lesion-level PI-RADS assigned to a patient is greater than or equal to 3, the patient is considered positive. |
| Participant-level specificity for detection of clinically significant prostate cancer | One-time MRI, up to 30 days post-enrollment in study. | If the highest lesion-level PI-RADS assigned to a patient is greater than or equal to 3, the patient is considered positive. |
| Quadrant-level sensitivity for detection of clinically significant prostate cancer | One-time MRI, up to 30 days post-enrollment in study. | If the highest lesion-level PI-RADS assigned to a quadrant is greater than or equal to 3, the quadrant is considered positive. |
| Quadrant-level specificity for detection of clinically significant prostate cancer | One-time MRI, up to 30 days post-enrollment in study. | If the highest lesion-level PI-RADS assigned to a quadrant is greater than or equal to 3, the quadrant is considered positive. |
| Free-response ROC curve for the aided and unaided reads | One-time MRI, up to 30 days post-enrollment in study. | The average number of false positives per patient will be on the x-axis, while lesion-level sensitivity will be on the y-axis. |
| Comparison of the quadrant-level AUC of the unaided vs AI read | One-time MRI, up to 30 days post-enrollment in study. | — |
Countries
United States
Contacts
CONTACTAndrei Purysko, MD
PRINCIPAL_INVESTIGATORAndrei Purysko, MD
Case Comprehensive Cancer Center, Cleveland Clinic
Outcome results
None listed