Diagnostic interchangeability of deep convolutional neural networks reconstructed knee MR images: preliminary experience

Abstract

Background: MRI acceleration using deep learning (DL) convolutional neural networks (CNNs) is a novel technique with great promise. Increasing the number of convolutional layers may allow for more accurate image reconstruction. Studies on evaluating the diagnostic interchangeability of DL reconstructed knee magnetic resonance (MR) images are scarce. The purpose of this study was to develop a deep CNN (DCNN) with an optimal number of layers for accelerating knee magnetic resonance imaging (MRI) acquisition by 6-fold and to test the diagnostic interchangeability and image quality of nonaccelerated images versus images reconstructed with a 15-layer DCNN or 3-layer CNN.Methods: For the feasibility portion of this study, 10 patients were randomly selected from the Osteoarthritis Initiative (OAI) cohort. For the interchangeability portion of the study, 40 patients were randomly selected from the OAI cohort. Three readers assessed meniscal and anterior cruciate ligament (ACL) tears and cartilage defects using DCNN, CNN, and nonaccelerated images. Image quality was subjectively graded as nondiagnostic, poor, acceptable, or excellent. Interchangeability was tested by comparing the frequency of agreement when readers used both accelerated and nonaccelerated images to frequency of agreement when readers only used nonaccelerated images. A noninferiority margin of 0.10 was used to ensure type I error ≤5% and power ≥80%. A logistic regression model using generalized estimating equations was used to compare proportions; 95% confidence intervals (CIs) were constructed.Results: DCNN and CNN images were interchangeable with nonaccelerated images for all structures, with excess disagreement values ranging from -2.5% [95% CI: (-6.1, 1.1)] to 3.0% [95% CI: (-0.1, 6.1)]. The quality of DCNN images was graded higher than that of CNN images but less than that of nonaccelerated images [excellent/acceptable quality: DCNN, 95% of cases (114/120); CNN, 60% (72/120); nonaccelerated, 97.5% (117/120)].Conclusions: Six-fold accelerated knee images reconstructed with a DL technique are diagnostically interchangeable with nonaccelerated images and have acceptable image quality when using a 15-layer CNN.

A Torn ACL Mapping in Knee MRI Images Using Deep Convolution Neural Network with Inception-v3

“…DCNN usually performs well with a larger dataset over a smaller one. TL could be useful in the CNN applications where the dataset is not huge [ 13 ]. For applications with comparatively smaller datasets, the concept of TL utilizes the learned model from large datasets such as ImageNet.…”

“…According to [ 13 ], a DCNN model has been developed for comparing the diagnostic interchangeability and image quality of nonaccelerated images to 15-layered DCNNs or 3-layered CNNs images with the optimal number of layers for a sixfold acceleration of the knee MRI dataset, with the optimal number of layers being three.…”

3D MRI of the Knee

“…4. 21,22 In both of these studies, the deep learning algorithms were trained on retrospectively undersampled k-space from 2D sequences using different types of convolutional neural networks. For AI acceleration techniques, although model training is an iterative process that can be time intensive and computationally demanding, once the model is fully trained, image reconstruction is fast and can be performed without the need for special reconstruction hardware.…”

“…evolve, these advanced imaging techniques should become more practical in a clinical workflow. 12,17,[21][22][23][24][25][26][27] In this review, we discuss 3D MRI sequences currently available for the knee, as well as the advantages and limitations of 3D MRI relative to routine 2D knee MRI. We also describe emerging techniques for 3D MRI, including acceleration and automated quantification techniques.…”

Parallel imaging with a combination of sensitivity encoding and generative adversarial networks

“…Ghodrati (17) showed that network training using the perceptual loss function achieved better agreement among radiologist scorings as compared to those networks using L 1 , L 2, or structural similarity (SSIM) objective functions. Subhas (18) demonstrated the feasibility of accelerating knee MRI acquisition 6-fold through the application of a novel CNN architecture with deep layers.…”