Open-sourced semi-automatic program for ultrasound assessments of femoral trochlea cartilage health.
McKenzie S WhiteRiann M Palmieri-SmithLindsey K LepleyPublished in: Computer methods in biomechanics and biomedical engineering (2023)
We developed an open-sourced program that accounts for the current limitations of B-mode ultrasound and extends the clinical utility of ultrasound for assessing femoral trochlear cartilage thickness. Images were collected on 18 patients with a history of knee surgery. By failing to account for ultrasound acoustics and beam refraction, cartilage thickness was underestimated by 26% and overestimated by 0-4%, respectively ( p < 0.001). Comprehensive thickness measurements achieved by measuring the Euclidean distance between every point were significantly different than traditional single-location measurements or by using the area/length ( p = 0.004-0.006). Sub-regions were significantly different than all whole regions of interest ( p = 0.001-0.012).
Keyphrases
- magnetic resonance imaging
- optical coherence tomography
- minimally invasive
- contrast enhanced ultrasound
- healthcare
- quality improvement
- deep learning
- extracellular matrix
- public health
- mental health
- machine learning
- risk assessment
- convolutional neural network
- coronary artery disease
- social media
- acute coronary syndrome
- coronary artery bypass
- atrial fibrillation
- human health