Gynaecologic Organ Segmentation and Motion Tracking Using Ultrasound
Trial Snapshot
- Phase
- Not Applicable
- Enrollment
- 25
- Locations
- 1
- Primary Endpoint
- Percentage of ultrasound scans with dice similarity coefficient (DSC) of >0.90 between algorithm-derived contour and ground-truth contour
Study Overview
Brief Summary
This study will establish the feasibility and accuracy of localising and outlining the uterus on ultrasound images using both healthy volunteer and cervical patient cohorts. This will enable us to determine whether or not ultrasound will be a good option to ensure the correct patient position prior to radiotherapy for cervical cancer patients.
Detailed Description
Radiotherapy is an established and effective method for treating cancer, but it can have negative side effects if healthy tissues are also exposed to radiation. In the case of cervical cancer, the bladder, small bowel, and bone marrow are often irradiated, which can result in both short-term and long-term side effects. There are sophisticated radiation delivery techniques (known as IMRT and VMAT) which are capable of administering the radiation dose in a precise and controlled manner so that the intended target is irradiated and the healthy tissue is spared. However, the only way for these sophisticated radiation delivery methods to work in cervical cancer is to localise the uterus prior to radiotherapy treatment so that the lineup between the radiation beam and the target is ensured. The uterus is a relatively mobile organ, and can have very different positions within the body depending on the patient position, bladder volume, tumour site, etc. Currently, we are unable to take advantage of the sophisticated radiation delivery techniques as we have no reliable way of seeing where the uterus is. Ultrasound is a promising way to overcome this problem.
Ultrasound is non-ionising, has good soft-tissue contrast, and can easily be incorporated into the radiation treatment room. We wish to establish the feasibility and accuracy of localising and outlining the uterus on ultrasound images. This will enable us to determine whether or not ultrasound will be a good option to ensure the correct patient position prior to radiotherapy for cervical cancer patients.
Study Design
- Study Type
- Observational
- Observational Model
- Cohort
- Time Perspective
- Prospective
Eligibility Criteria
- Ages
- 18 Years to — (Adult, Older Adult)
- Sex
- Female
- Accepts Healthy Volunteers
- Yes
Inclusion Criteria
- •healthy volunteer cohort: Healthy female adult volunteers
- •patient volunteer cohort: Patients receiving radiotherapy for cervical cancer who will receive a planning CT scan and daily CBCT scanning as part of their treatment.
Exclusion Criteria
- •both cohorts: Women who have received hysterectomies and women who are pregnant at the time of the study
Outcomes
Primary Outcomes
Percentage of ultrasound scans with dice similarity coefficient (DSC) of >0.90 between algorithm-derived contour and ground-truth contour
Time Frame: DSC calculated at 24 time points for each healthy volunteer (3 days)
The DSC between the ground truth contour and the algorithm derived contour will indicate the accuracy of the segmentation algorithm.
Coefficient of variation between of expert-drawn contours on ultrasound and expert-drawn contours on CBCT
Time Frame: COV calculated at a minimum of 3 time points during 6 weeks that the patient is receiving radiation therapy
A small COV will indicate greater clinician/expert confidence in segmenting the uterus
Secondary Outcomes
- Bland-Altman plot (healthy volunteer cohort)(1 day)
- Uterine center of mass displacement on ultrasound (healthy volunteer cohort)(3 days)
- Expert confidence scores for each imaging modality (ultrasound and MR) (healthy volunteer cohort)(from data collected over 3 days)
- Bland-Altman plot (patient volunteer cohort)(6 weeks)
- Expert confidence scores for each imaging modality (ultrasound, CT, and CBCT) (patient volunteer cohort)(from data collected over 6 weeks)