Implementation of GAN-Based, Synthetic T2-Weighted Fat Saturated Images in the Routine Radiological Workflow Improves Spinal Pathology Detection

Improving machine learning-based bitewing segmentation with synthetic data

OBJECTIVES

Class imbalance in datasets is one of the challenges of machine learning (ML) in medical image analysis. We employed synthetic data to overcome class imbalance when segmenting bitewing radiographs as an exemplary task for using ML.

METHODS

After segmenting bitewings into classes, i.e. dental structures, restorations, and background, the pixel-level representation of implants in the training set (1543 bitewings) and testing set (177 bitewings) was 0.03 % and 0.07 %, respectively. A diffusion model and a generative adversarial network (pix2pix) were used to generate a dataset synthetically enriched in implants. A U-Net segmentation model was trained on (1) the original dataset, (2) the synthetic dataset, (3) on the synthetic dataset and fine-tuned on the original dataset, or (4) on a dataset which was naïvely oversampled with images containing implants.

RESULTS

U-Net trained on the original dataset was unable to segment implants in the testing set. Model performance was significantly improved by naïve over-sampling, achieving the highest precision. The model trained only on synthetic data performed worse than naïve over-sampling in all metrics, but with fine-tuning on original data, it resulted in the highest Dice score, recall, F1 score and ROC AUC, respectively. The performance on other classes than implants was similar for all strategies except training only on synthetic data, which tended to perform worse.

CONCLUSIONS

The use of synthetic data alone may deteriorate the performance of segmentation models. However, fine-tuning on original data could significantly enhance model performance, especially for heavily underrepresented classes.

CLINICAL SIGNIFICANCE

This study explored the use of synthetic data to enhance segmentation of bitewing radiographs, focusing on underrepresented classes like implants. Pre-training on synthetic data followed by fine-tuning on original data yielded the best results, highlighting the potential of synthetic data to advance AI-driven dental imaging and ultimately support clinical decision-making.

Deep learning-based acceleration of muscle water T2 mapping in patients with neuromuscular diseases by more than 50% - translating quantitative MRI from research to clinical routine

BACKGROUND

Quantitative muscle water T2 (T2w) mapping is regarded as a biomarker for disease activity and response to treatment in neuromuscular diseases (NMD). However, the implementation in clinical settings is limited due to long scanning times and low resolution. Using artificial intelligence (AI) to accelerate MR image acquisition offers a possible solution. Combining compressed sensing and parallel imaging with AI-based reconstruction, known as CSAI (SmartSpeed, Philips Healthcare), allows for the generation of high-quality, weighted MR images in a shorter scan time. However, CSAI has not yet been investigated for quantitative MRI. Therefore, in the present work we assessed the performance of CSAI acceleration for T2w mapping compared to standard acceleration with SENSE.

METHODS

T2w mapping of the thigh muscles, based on T2-prepared 3D TSE with SPAIR fat suppression, was performed using standard SENSE (acceleration factor of 2; 04:35 min; SENSE) and CSAI (acceleration factor of 5; 01:57 min; CSAI 5x) in ten patients with facioscapulohumeral muscular dystrophy (FSHD). Subjects were scanned in two consecutive sessions (14 days in between). In each dataset, six regions of interest were placed in three thigh muscles bilaterally. SENSE and CSAI 5x acceleration were compared for i) image quality using apparent signal- and contrast-to-noise ratio (aSNR/aCNR), ii) diagnostic agreement of T2w values, and iii) intra- and inter-session reproducibility.

RESULTS

aSNR and aCNR of SENSE and CSAI 5x scans were not significantly different (p >  0.05). An excellent agreement of SENSE and CSAI 5x T2w values was shown (r =  0.99; ICC =  0.992). T2w mapping with both acceleration methods showed excellent, matching intra-method reproducibility.

CONCLUSION

AI-based acceleration of CS data allows for scan time reduction of more than 50% for T2w mapping in the thigh muscles of NMD patients without compromising quantitative validity.

Generating synthetic high-resolution spinal STIR and T1w images from T2w FSE and low-resolution axial Dixon

OBJECTIVES

To generate sagittal T1-weighted fast spin echo (T1w FSE) and short tau inversion recovery (STIR) images from sagittal T2-weighted (T2w) FSE and axial T1w gradient echo Dixon technique (T1w-Dixon) sequences.

MATERIALS AND METHODS

This retrospective study used three existing datasets: "Study of Health in Pomerania" (SHIP, 3142 subjects, 1.5 Tesla), "German National Cohort" (NAKO, 2000 subjects, 3 Tesla), and an internal dataset (157 patients 1.5/3 Tesla). We generated synthetic sagittal T1w FSE and STIR images from sagittal T2w FSE and low-resolution axial T1w-Dixon sequences based on two successively applied 3D Pix2Pix deep learning models. "Peak signal-to-noise ratio" (PSNR) and "structural similarity index metric" (SSIM) were used to evaluate the generated image quality on an ablations test. A Turing test, where seven radiologists rated 240 images as either natively acquired or generated, was evaluated using misclassification rate and Fleiss kappa interrater agreement.

RESULTS

Including axial T1w-Dixon or T1w FSE images resulted in higher image quality in generated T1w FSE (PSNR = 26.942, SSIM = 0.965) and STIR (PSNR = 28.86, SSIM = 0.948) images compared to using only single T2w images as input (PSNR = 23.076/24.677 SSIM = 0.952/0.928). Radiologists had difficulty identifying generated images (misclassification rate: 0.39 ± 0.09 for T1w FSE, 0.42 ± 0.18 for STIR) and showed low interrater agreement on suspicious images (Fleiss kappa: 0.09 for T1w/STIR).

CONCLUSIONS

Axial T1w-Dixon and sagittal T2w FSE images contain sufficient information to generate sagittal T1w FSE and STIR images.

CLINICAL RELEVANCE STATEMENT

T1w fast spin echo and short tau inversion recovery can be retroactively added to existing datasets, saving MRI time and enabling retrospective analysis, such as evaluating bone marrow pathologies.

KEY POINTS

Sagittal T2-weighted images alone were insufficient for differentiating fat and water and to generate T1-weighted images. Axial T1w Dixon technique, together with a T2-weighted sequence, produced realistic sagittal T1-weighted images. Our approach can be used to retrospectively generate STIR and T1-weighted fast spin echo sequences.

Practical Applications of Artificial Intelligence in Spine Imaging: A Review

Artificial intelligence (AI), a transformative technology with unprecedented potential in medical imaging, can be applied to various spinal pathologies. AI-based approaches may improve imaging efficiency, diagnostic accuracy, and interpretation, which is essential for positive patient outcomes. This review explores AI algorithms, techniques, and applications in spine imaging, highlighting diagnostic impact and challenges with future directions for integrating AI into spine imaging workflow.

Generative adversarial networks for spine imaging: A critical review of current applications

PURPOSE

In recent years, the field of medical imaging has witnessed remarkable advancements, with innovative technologies which revolutionized the visualization and analysis of the human spine. Among the groundbreaking developments in medical imaging, Generative Adversarial Networks (GANs) have emerged as a transformative tool, offering unprecedented possibilities in enhancing spinal imaging techniques and diagnostic outcomes. This review paper aims to provide a comprehensive overview of the use of GANs in spinal imaging, and to emphasize their potential to improve the diagnosis and treatment of spine-related disorders. A specific review focusing on Generative Adversarial Networks (GANs) in the context of medical spine imaging is needed to provide a comprehensive and specialized analysis of the unique challenges, applications, and advancements within this specific domain, which might not be fully addressed in broader reviews covering GANs in general medical imaging. Such a review can offer insights into the tailored solutions and innovations that GANs bring to the field of spinal medical imaging.

METHODS

An extensive literature search from 2017 until July 2023, was conducted using the most important search engines and identified studies that used GANs in spinal imaging.

RESULTS

The implementations include generating fat suppressed T2-weighted (fsT2W) images from T1 and T2-weighted sequences, to reduce scan time. The generated images had a significantly better image quality than true fsT2W images and could improve diagnostic accuracy for certain pathologies. GANs were also utilized in generating virtual thin-slice images of intervertebral spaces, creating digital twins of human vertebrae, and predicting fracture response. Lastly, they could be applied to convert CT to MRI images, with the potential to generate near-MR images from CT without MRI.

CONCLUSIONS

GANs have promising applications in personalized medicine, image augmentation, and improved diagnostic accuracy. However, limitations such as small databases and misalignment in CT-MRI pairs, must be considered.

Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation

BACKGROUND

Automated segmentation of spinal magnetic resonance imaging (MRI) plays a vital role both scientifically and clinically. However, accurately delineating posterior spine structures is challenging.

METHODS

This retrospective study, approved by the ethical committee, involved translating T1-weighted and T2-weighted images into computed tomography (CT) images in a total of 263 pairs of CT/MR series. Landmark-based registration was performed to align image pairs. We compared two-dimensional (2D) paired - Pix2Pix, denoising diffusion implicit models (DDIM) image mode, DDIM noise mode - and unpaired (SynDiff, contrastive unpaired translation) image-to-image translation using "peak signal-to-noise ratio" as quality measure. A publicly available segmentation network segmented the synthesized CT datasets, and Dice similarity coefficients (DSC) were evaluated on in-house test sets and the "MRSpineSeg Challenge" volumes. The 2D findings were extended to three-dimensional (3D) Pix2Pix and DDIM.

RESULTS

2D paired methods and SynDiff exhibited similar translation performance and DCS on paired data. DDIM image mode achieved the highest image quality. SynDiff, Pix2Pix, and DDIM image mode demonstrated similar DSC (0.77). For craniocaudal axis rotations, at least two landmarks per vertebra were required for registration. The 3D translation outperformed the 2D approach, resulting in improved DSC (0.80) and anatomically accurate segmentations with higher spatial resolution than that of the original MRI series.

CONCLUSIONS

Two landmarks per vertebra registration enabled paired image-to-image translation from MRI to CT and outperformed all unpaired approaches. The 3D techniques provided anatomically correct segmentations, avoiding underprediction of small structures like the spinous process.

RELEVANCE STATEMENT

This study addresses the unresolved issue of translating spinal MRI to CT, making CT-based tools usable for MRI data. It generates whole spine segmentation, previously unavailable in MRI, a prerequisite for biomechanical modeling and feature extraction for clinical applications.

KEY POINTS

• Unpaired image translation lacks in converting spine MRI to CT effectively. • Paired translation needs registration with two landmarks per vertebra at least. • Paired image-to-image enables segmentation transfer to other domains. • 3D translation enables super resolution from MRI to CT. • 3D translation prevents underprediction of small structures.