-
Unsupervised Domain Adaptation for Automated Knee Osteoarthritis Phenotype Classification
Authors:
Junru Zhong,
Yongcheng Yao,
Donal G. Cahill,
Fan Xiao,
Siyue Li,
Jack Lee,
Kevin Ki-Wai Ho,
Michael Tim-Yun Ong,
James F. Griffith,
Weitian Chen
Abstract:
Purpose: The aim of this study was to demonstrate the utility of unsupervised domain adaptation (UDA) in automated knee osteoarthritis (OA) phenotype classification using a small dataset (n=50). Materials and Methods: For this retrospective study, we collected 3,166 three-dimensional (3D) double-echo steady-state magnetic resonance (MR) images from the Osteoarthritis Initiative dataset and 50 3D t…
▽ More
Purpose: The aim of this study was to demonstrate the utility of unsupervised domain adaptation (UDA) in automated knee osteoarthritis (OA) phenotype classification using a small dataset (n=50). Materials and Methods: For this retrospective study, we collected 3,166 three-dimensional (3D) double-echo steady-state magnetic resonance (MR) images from the Osteoarthritis Initiative dataset and 50 3D turbo/fast spin-echo MR images from our institute (in 2020 and 2021) as the source and target datasets, respectively. For each patient, the degree of knee OA was initially graded according to the MRI Osteoarthritis Knee Score (MOAKS) before being converted to binary OA phenotype labels. The proposed UDA pipeline included (a) pre-processing, which involved automatic segmentation and region-of-interest crop**; (b) source classifier training, which involved pre-training phenotype classifiers on the source dataset; (c) target encoder adaptation, which involved unsupervised adaption of the source encoder to the target encoder and (d) target classifier validation, which involved statistical analysis of the target classification performance evaluated by the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity and accuracy. Additionally, a classifier was trained without UDA for comparison. Results: The target classifier trained with UDA achieved improved AUROC, sensitivity, specificity and accuracy for both knee OA phenotypes compared with the classifier trained without UDA. Conclusion: The proposed UDA approach improves the performance of automated knee OA phenotype classification for small target datasets by utilising a large, high-quality source dataset for training. The results successfully demonstrated the advantages of the UDA approach in classification on small datasets.
△ Less
Submitted 13 December, 2022;
originally announced December 2022.
-
Denoising of Three-Dimensional Fast Spin Echo Magnetic Resonance Images of Knee Joints using Spatial-Variant Noise-Relevant Residual Learning of Convolution Neural Network
Authors:
Shutian Zhao,
Donal G. Cahill,
Siyue Li,
Fan Xiao,
Thierry Blu,
James F Griffith,
Weitian Chen
Abstract:
Two-dimensional (2D) fast spin echo (FSE) techniques play a central role in the clinical magnetic resonance imaging (MRI) of knee joints. Moreover, three-dimensional (3D) FSE provides high-isotropic-resolution magnetic resonance (MR) images of knee joints, but it has a reduced signal-to-noise ratio compared to 2D FSE. Deep-learning denoising methods are a promising approach for denoising MR images…
▽ More
Two-dimensional (2D) fast spin echo (FSE) techniques play a central role in the clinical magnetic resonance imaging (MRI) of knee joints. Moreover, three-dimensional (3D) FSE provides high-isotropic-resolution magnetic resonance (MR) images of knee joints, but it has a reduced signal-to-noise ratio compared to 2D FSE. Deep-learning denoising methods are a promising approach for denoising MR images, but they are often trained using synthetic noise due to challenges in obtaining true noise distributions for MR images. In this study, inherent true noise information from 2-NEX acquisition was used to develop a deep-learning model based on residual learning of convolutional neural network (CNN), and this model was used to suppress the noise in 3D FSE MR images of knee joints. The proposed CNN used two-step residual learning over parallel transporting and residual blocks and was designed to comprehensively learn real noise features from 2-NEX training data. The results of an ablation study validated the network design. The new method achieved improved denoising performance of 3D FSE knee MR images compared with current state-of-the-art methods, based on the peak signal-to-noise ratio and structural similarity index measure. The improved image quality after denoising using the new method was verified by radiological evaluation. A deep CNN using the inherent spatial-varying noise information in 2-NEX acquisitions was developed. This method showed promise for clinical MRI assessments of the knee, and has potential applications for the assessment of other anatomical structures.
△ Less
Submitted 20 April, 2022;
originally announced April 2022.
-
Prevalence of algorithm-based qualitative (ABQ) method osteoporotic vertebral fracture in elderly Chinese men and women with reference to semi-quantitative (SQ) method: Mr. Os and Ms Os. (Hong Kong) studies
Authors:
Xian Jun Zeng,
Min Deng,
Yi Xiang Wang,
James F. Griffith,
Lai Chang He,
Anthony W. L. Kwok,
Jason C. S. Leung,
Timothy Kwok,
** Chung Leung
Abstract:
Introduction: This study evaluated algorithm-based qualitative (ABQ) method for vertebral fracture (VF) evaluation with reference to semi-quantitative (SQ) method and bone mineral density (BMD) measurement. Methods: Mr. OS (Hong Kong) and Ms. OS (Hong Kong) represent the first large-scale cohort studies on bone health in elderly Chinese men and women. The current study compared Genant's SQ method…
▽ More
Introduction: This study evaluated algorithm-based qualitative (ABQ) method for vertebral fracture (VF) evaluation with reference to semi-quantitative (SQ) method and bone mineral density (BMD) measurement. Methods: Mr. OS (Hong Kong) and Ms. OS (Hong Kong) represent the first large-scale cohort studies on bone health in elderly Chinese men and women. The current study compared Genant's SQ method and ABQ method in these two cohorts. Based on quantitative measurement, the severity of ABQ method detected fractures was additionally classified into grade-1, grad-2, and grade-3 according to SQ's deformity criteria. The radiographs of 1,954 elderly Chinese men (mean: 72.3 years) and 1,953 elderly Chinese women (mean: 72.5 years) were evaluated. Results: according to ABQ, grade-1,-2,-3 VFs accounted for 1.89%, 1.74%, 2.25% in men, and 3.33%, 3.07%, and 5.53% in women. In men and women, 15.7% (35/223) and 34.5% (48/139) of vertebrae with SQ grade-1 deformity were ABQ(+, with fracture) respectively. In men and women, 89.7% (35/39) and 66.7% (48/72) of vertebrae with ABQ grade-1 fracture had SQ grade-1 deformity. For grade-1 change, SQ (-, negative without fracture) & ABQ (+, positive with vertebral cortex line fracture) subjects tend to have a lower BMD than the SQ(+)& ABQ(-) subjects. In subjects with SQ grade-2 deformity, those were also ABQ(+) tended to have a lower BMD than those were ABQ(-). In all grades, SQ(-)&ABQ(-) subjects tended to have highest BMD, while SQ(+)&ABQ(+)subjects tended to have lowest BMD. Conclusion: ABQ method may be more sensitive to VF associated mild lower BMD than SQ method.
△ Less
Submitted 10 November, 2016;
originally announced November 2016.
