Prostate Cancer
Conditions
Brief summary
This study is investigating how AI can help doctors outline the prostate on an ultrasound image to make a custom radiation plan during a specialized type of radiation treatment for prostate cancer called brachytherapy.
Detailed description
This is a Phase II prospective study evaluating the standard U-net, a deep learning AI algorithm for auto-contouring of the prostate during HDR prostate brachytherapy with the needles in place by new learners. Contouring will be done on TRUS. The study will be conducted with a randomized design. Each patient will be assigned to a new learner and then randomized to manual versus AI-assisted contouring. The randomization will be stratified by new learner type: resident versus fellow/new attending. The hypothesis is that AI-assisted learner contours will have improved Dice coefficients with respect to clinically approved contours compared with manual learner contours. All brachytherapy contours will undergo review by the treating radiation oncologist who is the experienced clinician for clinical approval prior to patient treatment. The experienced clinician will be blinded to the randomization.
Interventions
OTHERArtificial-intelligence (AI)
Utilizing the addition of artificial intelligence during prostate contouring.
Sponsors
Brigham and Women's Hospital
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
TREATMENT
Masking
SINGLE (Caregiver)
Masking description
The experienced clinician will be blinded to the randomization of the contouring method by the new learner. The prostate contour generated by the new learner will then undergo review by the experienced clinician.
Eligibility
Sex/Gender
MALE
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* 18 years of age and older * Deemed suitable candidates for whole gland HDR prostate brachytherapy under general anesthesia as a monotherapy, boost or salvage treatment.
Exclusion criteria
* Prior permanent seed LDR brachytherapy implant * Prior transurethral resection of the prostate (TURP) * Presence or insertion of a rectal spacer * Focal HDR brachytherapy treatment i.e. not whole prostate
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Clinically approved contour vs. Manual and AI assisted contours | 1 day | Dice coefficient \[0: no match, 1: complete match\] between the final clinically approved brachytherapy prostate contours versus the manual and AI-assisted contours provided by the new learner. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Contouring time | 1 day | Contouring time-to-completion needed to generate and edit manual vs AI-assisted contours for a new learner. |
| Impression of AI or manual contours by new learner | 1 day | Subjective impressions/perception of AI or manual contours by the new learner based on survey responses. |
| Impression of AI or manual contours by experienced clinician | 1 day | Subjective impressions/perception of AI or manual contours by the experienced clinician based on survey responses. |
| Clinician contours vs. Trus images with and without needles | 1 day | Dice coefficients of expert clinician contours between TRUS images with implanted needles and TRUS images after removal of bottom needle rows. |
Countries
United States
Contacts
PRINCIPAL_INVESTIGATORMartin King, MD, PHD
Dana-Farber Cancer Institute / Brigham Women's Hospital
Outcome results
None listed