- 50 views
- 56 downloads
Automatic Migration Percentage Measurements on Ultrasound Images
-
- Author / Creator
- Yousefvand, Reza
-
Hip displacement is a prevalent disorder in children with cerebral palsy, defined as the lateral displacement of the femur head from under the acetabulum, and leads to severe pain and difficulties in daily activities. As a result, hip surveillance programs have been developed to monitor and diagnose hip displacement in children with cerebral palsy, and ensure that appropriate interventions are made at the right time. These programs involve regular assessments of hip displacement. Migration Percentage (MP), defined as the ratio of the distance between the lateral borders of the femur head and the acetabulum (A) to the total width of the femur head (B), is the gold standard parameter of hip displacement measurement and is measured using anteroposterior X-ray imaging (MP=A/B). However, the frequent X-ray imaging in hip surveillance programs exposes children to ionizing radiation, increasing the risk of cancer development later in life. Recently, ultrasound (US) has been proposed as a non-invasive and widely accessible alternative for hip assessments. Yet, the inherently fuzzy and noisy nature of US images makes accurately identifying edges and landmarks challenging, leading to time-consuming and user-dependent manual measurements. To tackle these issues, this study aimed to develop a fast, reliable, and fully automatic algorithm to measure the MP from US images.
In the developed methodology, for each of the “A” and “B” measurements, UNets were trained to segment the hip features on the coronal hip scans. A convolution neural network was trained to score and select frames for measurement. More UNets were trained to identify the measurement features on the selected frames, and statistical analysis was applied to aggregate and finalize the measurements.
To verify the developed method, a total of 38 children with an average age of 9 ± 3.4 years old were recruited, and 62 hips were scanned in total. From the 62 scanned hips, 36 were utilized for training, 8 for validation, and 18 for testing. An experienced rater provided the X-ray measurements for all scanned hips as the ground truth. The mean absolute difference (MAD) and intra-class correlation coefficient (ICC(2,1)) of the test measurements were 6.5% ± 5.5% and 0.86. The clinical acceptance rate was 72%, and the sensitivity and specificity of classification of displaced hips (MP>30) were found to be 100% and 93%. The measurement time for each hip was 105.6 seconds on average, which was 3 times faster than manual measurements. Hence, the developed method demonstrated good reliability, accuracy, and speed in MP measurements, marking a significant step towards replacing X-ray with US in hip assessments. -
- Subjects / Keywords
-
- Graduation date
- Fall 2024
-
- Type of Item
- Thesis
-
- Degree
- Master of Science
-
- License
- This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.