Fully automated algorithm for the detection of bone marrow oedema lesions in patients with axial spondyloarthritis – Feasibility study

Rzecki, Krzysztof; Kucybała, Iwona; Gut, Daniel; Jarosz, Aldona; Nabagło, Tomasz; Tabor, Zbisław; Wojciechowski, Wadim

doi:10.1016/j.bbe.2021.05.005

Cited by 16 publications

(21 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Automated models based on deep learning algorithms have recently been proposed as a problem-solving tool for detecting and classifying bone marrow pathology, both in adults and children [ 21 , 22 ]. These studies presupposed a definition of pathological bone marrow signal on MRI, although to date, no such objective definition exists.…”

Section: Introductionmentioning

confidence: 99%

Automated segmentation of magnetic resonance bone marrow signal: a feasibility study

et al. 2022

View full text Add to dashboard Cite

Background Manual assessment of bone marrow signal is time-consuming and requires meticulous standardisation to secure adequate precision of findings. Objective We examined the feasibility of using deep learning for automated segmentation of bone marrow signal in children and adolescents. Materials and methods We selected knee images from 95 whole-body MRI examinations of healthy individuals and of children with chronic non-bacterial osteomyelitis, ages 6–18 years, in a longitudinal prospective multi-centre study cohort. Bone marrow signal on T2-weighted Dixon water-only images was divided into three color-coded intensity-levels: 1 = slightly increased; 2 = mildly increased; 3 = moderately to highly increased, up to fluid-like signal. We trained a convolutional neural network on 85 examinations to perform bone marrow segmentation. Four readers manually segmented a test set of 10 examinations and calculated ground truth using simultaneous truth and performance level estimation (STAPLE). We evaluated model and rater performance through Dice similarity coefficient and in consensus. Results Consensus score of model performance showed acceptable results for all but one examination. Model performance and reader agreement had highest scores for level-1 signal (median Dice 0.68) and lowest scores for level-3 signal (median Dice 0.40), particularly in examinations where this signal was sparse. Conclusion It is feasible to develop a deep-learning-based model for automated segmentation of bone marrow signal in children and adolescents. Our model performed poorest for the highest signal intensity in examinations where this signal was sparse. Further improvement requires training on larger and more balanced datasets and validation against ground truth, which should be established by radiologists from several institutions in consensus.

show abstract

Section: Introductionmentioning

confidence: 99%

Automated segmentation of magnetic resonance bone marrow signal: a feasibility study

et al. 2022

View full text Add to dashboard Cite

show abstract

“…However, these studies relied on manual segmentation to identify an optimal threshold, whereas our data suggest that using manual segmentation as a “gold standard” is problematic and may lead to inconsistent interpretation especially in cases when inflammation is subtle or precise lesion boundary cannot be identified. To highlight this point, a recent study aiming to demonstrate the feasibility of fully-automated segmentation of BME [ 25 ] revised the threshold value developed in earlier work [ 23 ], finding an optimal threshold of 1 compared to 1.5 in the prior study. Clearly, a threshold which depends on reference standard provided by human observers is not desirable.…”

Section: Discussionmentioning

confidence: 99%

Volume of hyperintense inflammation (VHI): A quantitative imaging biomarker of inflammation load in spondyloarthritis, enabled by human-machine cooperation

et al. 2023

View full text Add to dashboard Cite

Qualitative visual assessment of MRI scans is a key mechanism by which inflammation is assessed in clinical practice. For example, in axial spondyloarthritis (axSpA), visual assessment focuses on the identification of regions with increased signal in the bone marrow, known as bone marrow oedema (BMO), on water-sensitive images. The identification of BMO has an important role in the diagnosis, quantification and monitoring of disease in axSpA. However, BMO evaluation depends heavily on the experience and expertise of the image reader, creating substantial imprecision. Deep learning-based segmentation is a natural approach to addressing this imprecision, but purely automated solutions require large training sets that are not currently available, and deep learning solutions with limited data may not be sufficiently trustworthy for use in clinical practice. To address this, we propose a workflow for inflammation segmentation incorporating both deep learning and human input. With this ‘human-machine cooperation’ workflow, a preliminary segmentation is generated automatically by deep learning; a human reader then ‘cleans’ the segmentation by removing extraneous segmented voxels. The final cleaned segmentation defines the volume of hyperintense inflammation (VHI), which is proposed as a quantitative imaging biomarker (QIB) of inflammation load in axSpA. We implemented and evaluated the proposed human-machine workflow in a cohort of 29 patients with axSpA who had undergone prospective MRI scans before and after starting biologic therapy. The performance of the workflow was compared against purely visual assessment in terms of inter-observer/inter-method segmentation overlap, inter-observer agreement and assessment of response to biologic therapy. The human-machine workflow showed superior inter-observer segmentation overlap than purely manual segmentation (Dice score 0.84 versus 0.56). VHI measurements produced by the workflow showed similar or better inter-observer agreement than visual scoring, with similar response assessments. We conclude that the proposed human-machine workflow offers a mechanism to improve the consistency of inflammation assessment, and that VHI could be a valuable QIB of inflammation load in axSpA, as well as offering an exemplar of human-machine cooperation more broadly.

show abstract

“…Automatic segmentation of ilium and sacrum constitutes another innovation in our proposed ML pipeline. Similar to the work of Rzecki et al, 13 we employ a U‐Net segmentation module to delineate the ilium and sacral bone 21 . However, in contrast to the latter study, our method is based on more easily available noisy annotations and subsequently used to narrow down the search area for prediction of inflammatory lesions.…”

Section: Discussionmentioning

confidence: 99%

“…8 Deep neural networks recently showed excellent performance for the classification of radiographic sacroiliitis and detection of incidental sacroiliitis on CT. 9,10 In addition, there is growing interest in automated detection of active sacroiliitis on MRI. [11][12][13] Former studies elaborated on classification of SI joint MRI on a patient level (ie, binary outcome, sacroiliitis yes/no) or image level (ie, signs of sacroiliitis on individual MRI slices) but often required manual segmentation or annotation of the region of interest (ROI). Techniques offering a more detailed (ie, quadrant-level) prediction of inflammatory lesions are scarce.…”

Section: Introductionmentioning

confidence: 99%

Machine Learning Pipeline for Predicting Bone Marrow Edema Along the Sacroiliac Joints on Magnetic Resonance Imaging

Roels

Craemer

Renson

et al. 2023

Arthritis & Rheumatology

View full text Add to dashboard Cite

ObjectiveTo develop and validate a fully automated machine learning (ML) algorithm that predicts bone marrow oedema (BMO) on a quadrant‐level in sacroiliac (SI) joint MRI.MethodsA computer vision workflow automatically locates the SI joints, segments regions of interest (ilium and sacrum), performs objective quadrant extraction and predicts presence of BMO, suggestive of inflammatory lesions, on a quadrant‐level in semi‐coronal slices of T1/T2‐weighted MRI scans. Ground truth was determined by consensus among human readers. The inflammation classifier was trained using a ResNet18 backbone and 5‐fold cross‐validated on scans of spondyloarthritis (SpA) patients (n=279), postpartum (n=71), and healthy subjects (n=114); while independent SpA patient MRIs (n=243) served as test dataset. Patient‐level predictions were derived from aggregating quadrant‐level predictions, i.e. at least one positive quadrant.ResultsThe algorithm automatically detects the SI joints with a precision of 98.4% and segments ilium/sacrum with an intersection‐over‐union of 85.6% and 67.9%, respectively. The inflammation classifier performed well in cross‐validation: area under the curve (AUC) 94.5%, balanced accuracy (B‐ACC) 80.5%, and F1 score 64.1%. In the test dataset, AUC was 88.2%, B‐ACC 72.1%, and F1 score 50.8%. On a patient‐level, the model achieved a B‐ACC of 81.6% and 81.4% in the cross‐validation and test dataset, respectively.ConclusionWe propose a fully automated ML pipeline that enables objective and standardized evaluation of BMO along the SI joints on MRI. This method has the potential to screen large numbers of (suspected) SpA patients and is a step closer towards artificial intelligence assisted diagnosis and follow‐up.

show abstract

Fully automated algorithm for the detection of bone marrow oedema lesions in patients with axial spondyloarthritis – Feasibility study

Cited by 16 publications

References 44 publications

Automated segmentation of magnetic resonance bone marrow signal: a feasibility study

Automated segmentation of magnetic resonance bone marrow signal: a feasibility study

Volume of hyperintense inflammation (VHI): A quantitative imaging biomarker of inflammation load in spondyloarthritis, enabled by human-machine cooperation

Machine Learning Pipeline for Predicting Bone Marrow Edema Along the Sacroiliac Joints on Magnetic Resonance Imaging

Contact Info

Product

Resources

About