Repeatability and Reproducibility of ADC Measurements and MRI Signal Intensity Measurements of Bone Marrow in Monoclonal Plasma Cell Disorders: A Prospective Bi-institutional Multiscanner, Multiprotocol Study

Deep learning-based automated high-accuracy location and identification of fresh vertebral compression fractures from spinal radiographs: a multicenter cohort study

Background

Digital radiography (DR) is a common and widely available examination. However, spinal DR cannot detect bone marrow edema, therefore, determining vertebral compression fractures (VCFs), especially fresh VCFs, remains challenging for clinicians.

Methods

We trained, validated, and externally tested the deep residual network (DRN) model that automated the detection and identification of fresh VCFs from spinal DR images. A total of 1,747 participants from five institutions were enrolled in this study and divided into the training cohort, validation cohort and external test cohorts (YHDH and BMUH cohorts). We evaluated the performance of DRN model based on the area under the receiver operating characteristic curve (AUC), feature attention maps, sensitivity, specificity, and accuracy. We compared it with five other deep learning models and validated and tested the model internally and externally and explored whether it remains highly accurate for an external test cohort. In addition, the influence of old VCFs on the performance of the DRN model was assessed.

Results

The AUC was 0.99, 0.89, and 0.88 in the validation, YHDH, and BMUH cohorts, respectively, for the DRN model for detecting and discriminating fresh VCFs. The accuracies were 81.45% and 72.90%, sensitivities were 84.75% and 91.43%, and specificities were 80.25% and 63.89% in the YHDH and BMUH cohorts, respectively. The DRN model generated correct activation on the fresh VCFs and accurate peak responses on the area of the target vertebral body parts and demonstrated better feature representation learning and classification performance. The AUC was 0.90 (95% confidence interval [CI] 0.84-0.95) and 0.84 (95% CI 0.72-0.93) in the non-old VCFs and old VCFs groups, respectively, in the YHDH cohort (p = 0.067). The AUC was 0.89 (95% CI 0.84-0.94) and 0.85 (95% CI 0.72-0.95) in the non-old VCFs and old VCFs groups, respectively, in the BMUH cohort (p = 0.051).

Conclusion

In present study, we developed the DRN model for automated diagnosis and identification of fresh VCFs from spinal DR images. The DRN model can provide interpretable attention maps to support the excellent prediction results, which is the key that most clinicians care about when using the model to assist decision-making.

Reproducible Radiomics Features from Multi-MRI-Scanner Test-Retest-Study: Influence on Performance and Generalizability of Models

BACKGROUND

Radiomics models trained on data from one center typically show a decline of performance when applied to data from external centers, hindering their introduction into large-scale clinical practice. Current expert recommendations suggest to use only reproducible radiomics features isolated by multiscanner test-retest experiments, which might help to overcome the problem of limited generalizability to external data.

PURPOSE

To evaluate the influence of using only a subset of robust radiomics features, defined in a prior in vivo multi-MRI-scanner test-retest-study, on the performance and generalizability of radiomics models.

STUDY TYPE

Retrospective.

POPULATION

Patients with monoclonal plasma cell disorders. Training set (117 MRIs from center 1); internal test set (42 MRIs from center 1); external test set (143 MRIs from center 2-8).

FIELD STRENGTH/SEQUENCE

1.5T and 3.0T; T1-weighted turbo spin echo.

ASSESSMENT

The task for the radiomics models was to predict plasma cell infiltration, determined by bone marrow biopsy, noninvasively from MRI. Radiomics machine learning models, including linear regressor, support vector regressor (SVR), and random forest regressor (RFR), were trained on data from center 1, using either all radiomics features, or using only reproducible radiomics features. Models were tested on an internal (center 1) and a multicentric external data set (center 2-8).

STATISTICAL TESTS

Pearson correlation coefficient r and mean absolute error (MAE) between predicted and actual plasma cell infiltration. Fisher's z-transformation, Wilcoxon signed-rank test, Wilcoxon rank-sum test; significance level P < 0.05.

RESULTS

When using only reproducible features compared with all features, the performance of the SVR on the external test set significantly improved (r = 0.43 vs. r = 0.18 and MAE = 22.6 vs. MAE = 28.2). For the RFR, the performance on the external test set deteriorated when using only reproducible instead of all radiomics features (r = 0.33 vs. r = 0.44, P = 0.29 and MAE = 21.9 vs. MAE = 20.5, P = 0.10).

CONCLUSION

Using only reproducible radiomics features improves the external performance of some, but not all machine learning models, and did not automatically lead to an improvement of the external performance of the overall best radiomics model.

LEVEL OF EVIDENCE: 3

TECHNICAL EFFICACY

Stage 2.

Physics-Informed Discretization for Reproducible and Robust Radiomic Feature Extraction Using Quantitative MRI

OBJECTIVE

Given the limited repeatability and reproducibility of radiomic features derived from weighted magnetic resonance imaging (MRI), there may be significant advantages to using radiomics in conjunction with quantitative MRI. This study introduces a novel physics-informed discretization (PID) method for reproducible radiomic feature extraction and evaluates its performance using quantitative MRI sequences including magnetic resonance fingerprinting (MRF) and apparent diffusion coefficient (ADC) mapping.

MATERIALS AND METHODS

A multiscanner, scan-rescan dataset comprising whole-brain 3D quantitative (MRF T1, MRF T2, and ADC) and weighted MRI (T1w MPRAGE, T2w SPACE, and T2w FLAIR) from 5 healthy subjects was prospectively acquired. Subjects underwent 2 repeated acquisitions on 3 distinct 3 T scanners each, for a total of 6 scans per subject (30 total scans). First-order statistical (n = 23) and second-order texture (n = 74) radiomic features were extracted from 56 brain tissue regions of interest using the proposed PID method (for quantitative MRI) and conventional fixed bin number (FBN) discretization (for quantitative MRI and weighted MRI). Interscanner radiomic feature reproducibility was measured using the intraclass correlation coefficient (ICC), and the effect of image sequence (eg, MRF T1 vs T1w MPRAGE), as well as image discretization method (ie, PID vs FBN), on radiomic feature reproducibility was assessed using repeated measures analysis of variance. The robustness of PID and FBN discretization to segmentation error was evaluated by simulating segmentation differences in brainstem regions of interest. Radiomic features with ICCs greater than 0.75 following simulated segmentation were determined to be robust to segmentation.

RESULTS

First-order features demonstrated higher reproducibility in quantitative MRI than weighted MRI sequences, with 30% (n = 7/23) features being more reproducible in MRF T1 and MRF T2 than weighted MRI. Gray level co-occurrence matrix (GLCM) texture features extracted from MRF T1 and MRF T2 were significantly more reproducible using PID compared with FBN discretization; for all quantitative MRI sequences, PID yielded the highest number of texture features with excellent reproducibility (ICC > 0.9). Comparing texture reproducibility of quantitative and weighted MRI, a greater proportion of MRF T1 (n = 225/370, 61%) and MRF T2 (n = 150/370, 41%) texture features had excellent reproducibility (ICC > 0.9) compared with T1w MPRAGE (n = 148/370, 40%), ADC (n = 115/370, 32%), T2w SPACE (n = 98/370, 27%), and FLAIR (n = 102/370, 28%). Physics-informed discretization was also more robust than FBN discretization to segmentation error, as 46% (n = 103/222, 46%) of texture features extracted from quantitative MRI using PID were robust to simulated 6 mm segmentation shift compared with 19% (n = 42/222, 19%) of weighted MRI texture features extracted using FBN discretization.

CONCLUSIONS

The proposed PID method yields radiomic features extracted from quantitative MRI sequences that are more reproducible and robust than radiomic features extracted from weighted MRI using conventional (FBN) discretization approaches. Quantitative MRI sequences also demonstrated greater scan-rescan robustness and first-order feature reproducibility than weighted MRI.

Normal Brain and Brain Tumor ADC: Changes Resulting From Variation of Diffusion Time and/or Echo Time in Pulsed-Gradient Spin Echo Diffusion Imaging

OBJECTIVES

Increasing gradient performance on modern magnetic resonance imaging scanners has profoundly reduced the attainable diffusion and echo times for clinically available pulsed-gradient spin echo (PGSE) sequences. This study investigated how this may impact the measured apparent diffusion coefficient (ADC), which is considered an important diagnostic marker for differentiation between normal and abnormal brain tissue and for therapeutic follow-up.

MATERIALS AND METHODS

Diffusion time and echo time dependence of the ADC were evaluated on a high-performance 3 T magnetic resonance imaging scanner. Diffusion PGSE brain scans were performed in 10 healthy volunteers and in 10 brain tumor patients using diffusion times of 16, 40, and 70 ms, echo times of 60, 75, and 104 ms at 3 b-values (0, 100, and 1000 s/mm 2 ), and a maximum gradient amplitude of 68 mT/m. A low gradient performance system was also emulated by reducing the diffusion encoding gradient amplitude to 19 mT/m. In healthy subjects, the ADC was measured in 6 deep gray matter regions and in 6 white matter regions. In patients, the ADC was measured in the solid part of the tumor.

RESULTS

With increasing diffusion time, a small but significant ADC increase of up to 2.5% was observed for 6 aggregate deep gray matter structures. With increasing echo time or reduced gradient performance, a small but significant ADC decrease of up to 2.6% was observed for 6 aggregate white matter structures. In tumors, diffusion time-related ADC changes were inconsistent without clear trend. For tumors with diffusivity above 1.0 μm 2 /ms, with prolonged echo time, there was a pronounced ADC increase of up to 12%. Meanwhile, for tumors with diffusivity at or below 1.0 μm 2 /ms, no change or a reduction was observed. Similar results were observed for gradient performance reduction, with an increase of up to 21%. The coefficient of variation determined in repeat experiments was 2.4%.

CONCLUSIONS

For PGSE and the explored parameter range, normal tissue ADC changes seem negligible. Meanwhile, observed tumor ADC changes can be relevant if ADC is used as a quantitative biomarker and not merely assessed by visual inspection. This highlights the importance of reporting all pertinent timing parameters in ADC studies and of considering these effects when building scan protocols for use in multicenter investigations.

3D CNN-based Deep Learning Model-based Explanatory Prognostication in Patients with Multiple Myeloma using Whole-body MRI

Although magnetic resonance imaging (MRI) data of patients with multiple myeloma (MM) are used to predict prognosis, few reports have applied artificial intelligence (AI) techniques for this purpose. We aimed to analyze whole-body diffusion-weighted MRI data using three-dimensional (3D) convolutional neural networks (CNNs) and Gradient-weighted Class Activation Mapping (Grad-CAM), an explainable AI, to predict prognosis and explore the factors involved in prediction. We retrospectively analyzed the MRI data of a total of 142 patients with MM obtained from two medical centers. We defined the occurrence of progressive disease after MRI evaluation within 12 months as a poor prognosis and constructed a 3D CNN-based deep learning model to predict prognosis. Images from 111 cases were used as the training and internal validation data; images from 31 cases were used as the external validation data. Internal validation of the AI model with stratified 5-fold cross-validation resulted in a significant difference in progression-free survival (PFS) between good and poor prognostic cases (2-year PFS, 91.2% versus [vs.] 61.1%, P = 0.0002). The AI model clearly stratified good and poor prognostic cases in the external validation cohort (2-year PFS, 92.9% vs. 55.6%, P = 0.004), with an area under the receiver operating characteristic curve of 0.804. According to Grad-CAM, the MRI signals of the spleen and bones of the vertebrae and pelvis contributed to prognosis prediction. This study is the first to show that image analysis of whole-body MRI using a 3D CNN without any other clinical data is effective in predicting the prognosis of patients with MM.

Deep Learning for Delineation of the Spinal Canal in Whole-Body Diffusion-Weighted Imaging: Normalising Inter- and Intra-Patient Intensity Signal in Multi-Centre Datasets

BACKGROUND

Whole-Body Diffusion-Weighted Imaging (WBDWI) is an established technique for staging and evaluating treatment response in patients with multiple myeloma (MM) and advanced prostate cancer (APC). However, WBDWI scans show inter- and intra-patient intensity signal variability. This variability poses challenges in accurately quantifying bone disease, tracking changes over follow-up scans, and developing automated tools for bone lesion delineation. Here, we propose a novel automated pipeline for inter-station, inter-scan image signal standardisation on WBDWI that utilizes robust segmentation of the spinal canal through deep learning.

METHODS

We trained and validated a supervised 2D U-Net model to automatically delineate the spinal canal (both the spinal cord and surrounding cerebrospinal fluid, CSF) in an initial cohort of 40 patients who underwent WBDWI for treatment response evaluation (80 scans in total). Expert-validated contours were used as the target standard. The algorithm was further semi-quantitatively validated on four additional datasets (three internal, one external, 207 scans total) by comparing the distributions of average apparent diffusion coefficient (ADC) and volume of the spinal cord derived from a two-component Gaussian mixture model of segmented regions. Our pipeline subsequently standardises WBDWI signal intensity through two stages: (i) normalisation of signal between imaging stations within each patient through histogram equalisation of slices acquired on either side of the station gap, and (ii) inter-scan normalisation through histogram equalisation of the signal derived within segmented spinal canal regions. This approach was semi-quantitatively validated in all scans available to the study (N = 287).

RESULTS

The test dice score, precision, and recall of the spinal canal segmentation model were all above 0.87 when compared to manual delineation. The average ADC for the spinal cord (1.7 × 10-3 mm2/s) showed no significant difference from the manual contours. Furthermore, no significant differences were found between the average ADC values of the spinal cord across the additional four datasets. The signal-normalised, high-b-value images were visualised using a fixed contrast window level and demonstrated qualitatively better signal homogeneity across scans than scans that were not signal-normalised.

CONCLUSION

Our proposed intensity signal WBDWI normalisation pipeline successfully harmonises intensity values across multi-centre cohorts. The computational time required is less than 10 s, preserving contrast-to-noise and signal-to-noise ratios in axial diffusion-weighted images. Importantly, no changes to the clinical MRI protocol are expected, and there is no need for additional reference MRI data or follow-up scans.

Radiomics signature based on robust features derived from diffusion data for differentiation between benign and malignant solitary pulmonary lesions

BACKGROUND

Classifying and characterizing pulmonary lesions are critical for clinical decision-making process to identify optimal therapeutic strategies. The purpose of this study was to develop and validate a radiomics nomogram for distinguishing between benign and malignant pulmonary lesions based on robust features derived from diffusion images.

MATERIAL AND METHODS

The study was conducted in two phases. In the first phase, we prospectively collected 30 patients with pulmonary nodule/mass who underwent twice EPI-DWI scans. The robustness of features between the two scans was evaluated using the concordance correlation coefficient (CCC) and dynamic range (DR). In the second phase, 139 patients who underwent pulmonary DWI were randomly divided into training and test sets in a 7:3 ratio. Maximum relevance minimum redundancy, least absolute shrinkage and selection operator, and logistic regression were used for feature selection and construction of radiomics signatures. Nomograms were established incorporating clinical features, radiomics signatures, and ADC(0, 800). The diagnostic efficiency of different models was evaluated using the area under the curve (AUC) and decision curve analysis.

RESULTS

Among the features extracted from DWI and ADC images, 42.7% and 37.4% were stable (both CCC and DR ≥ 0.85). The AUCs for distinguishing pulmonary lesions in the test set for clinical model, ADC, ADC radiomics signatures, and DWI radiomics signatures were 0.694, 0.802, 0.885, and 0.767, respectively. The nomogram exhibited the best differentiation performance (AUC = 0.923). The decision curve showed that the nomogram consistently outperformed ADC value and clinical model in lesion differentiation.

CONCLUSION

Our study demonstrates the robustness of radiomics features derived from lung DWI. The ADC radiomics nomogram shows superior clinical net benefits compared to conventional clinical models or ADC values alone in distinguishing solitary pulmonary lesions, offering a promising tool for noninvasive, precision diagnosis in lung cancer.

Repeatability of quantitative individual lesion and total disease multiparametric whole-body MRI measurements in prostate cancer bone metastases

OBJECTIVES

To assess the repeatability of quantitative multiparametric whole-body MRI (mpWB-MRI) parameters in advanced prostate cancer (APC) bone metastases.

METHODS

1.5T MRI was performed twice on the same day in 10 APC patients. MpWB-MRI-included diffusion weighted imaging (DWI) and T1-weighted gradient-echo 2-point Dixon sequences. ADC and relative fat-fraction percentage (rFF%) maps were calculated, respectively. A radiologist delineated up to 10 target bone metastases per study. Means of ADC, b900 signal intensity(SI), normalised b900 SI, rFF% and maximum diameter (MD) for each target lesion and overall parameter averages across all targets per patient were recorded. The total disease volume (tDV in ml) was manually delineated on b900 images and mean global (g)ADC was derived. Bland-Altman analyses were performed with calculation of 95% repeatability coefficients (RC).

RESULTS

Seventy-three individual targets (median MD 26 mm) were included. Lesion mean ADC RC was 12.5%, mean b900 SI RC 137%, normalised mean b900 SI RC 110%, rFF% RC 3.2 and target MD RC 5.5 mm (16.3%). Patient target lesion average mean ADC RC was 6.4%, b900 SI RC 104% and normalised mean b900 SI RC 39.6%. Target average rFF% RC was 1.8, average MD RC 1.3 mm (4.8%). tDV segmentation RC was 6.4% and mean gADC RC 5.3%.

CONCLUSIONS

APC bone metastases' ADC, rFF% and maximum diameter, tDV and gADC show good repeatability.

ADVANCES IN KNOWLEDGE

APC bone metastases' mean ADC and rFF% measurements of single lesions and global disease volumes are repeatable, supporting their potential role as quantitative biomarkers in metastatic bone disease.

Image quality in whole-body MRI using the MY-RADS protocol in a prospective multi-centre multiple myeloma study

BACKGROUND

The Myeloma Response Assessment and Diagnosis System (MY-RADS) guidelines establish a standardised acquisition and analysis pipeline for whole-body MRI (WB-MRI) in patients with myeloma. This is the first study to assess image quality in a multi-centre prospective trial using MY-RADS.

METHODS

The cohort consisted of 121 examinations acquired across ten sites with a range of prior WB-MRI experience, three scanner manufacturers and two field strengths. Image quality was evaluated qualitatively by a radiologist and quantitatively using a semi-automated pipeline to quantify common artefacts and image quality issues. The intra- and inter-rater repeatability of qualitative and quantitative scoring was also assessed.

RESULTS

Qualitative radiological scoring found that the image quality was generally good, with 94% of examinations rated as good or excellent and only one examination rated as non-diagnostic. There was a significant correlation between radiological and quantitative scoring for most measures, and intra- and inter-rater repeatability were generally good. When the quality of an overall examination was low, this was often due to low quality diffusion-weighted imaging (DWI), where signal to noise ratio (SNR), anterior thoracic signal loss and brain geometric distortion were found as significant predictors of examination quality.

CONCLUSIONS

It is possible to successfully deliver a multi-centre WB-MRI study using the MY-RADS protocol involving scanners with a range of manufacturers, models and field strengths. Quantitative measures of image quality were developed and shown to be significantly correlated with radiological assessment. The SNR of DW images was identified as a significant factor affecting overall examination quality.

TRIAL REGISTRATION

ClinicalTrials.gov, NCT03188172 , Registered on 15 June 2017.

CRITICAL RELEVANCE STATEMENT

Good overall image quality, assessed both qualitatively and quantitatively, can be achieved in a multi-centre whole-body MRI study using the MY-RADS guidelines.

KEY POINTS

• A prospective multi-centre WB-MRI study using MY-RADS can be successfully delivered. • Quantitative image quality metrics were developed and correlated with radiological assessment. • SNR in DWI was identified as a significant predictor of quality, allowing for rapid quality adjustment.

Identification of focal lesion characteristics in MRI which indicate presence of corresponding osteolytic lesion in CT in patients with multiple myeloma

PURPOSE

The presence of bone marrow focal lesions and osteolytic lesions in patients with multiple myeloma (MM) is of high prognostic significance for their individual outcome. It is not known yet why some focal lesions seen in MRI, reflecting localized bone marrow infiltration of myeloma cells, remain non-lytic, whereas others are associated with destruction of mineralized bone. In this study, we analyzed MRI characteristics of manually segmented focal lesions in MM patients to identify possible features that might discriminate lytic and non-lytic lesions.

METHOD

The initial cohort included a total of 140 patients with different stages of MM who had undergone both whole-body MRI and whole-body low-dose CT within 30 days, and of which 29 satisfied the inclusion criteria for this study. Focal lesions in MRI and corresponding osteolytic areas in CT were segmented manually. Analysis of the lesions included volume, location and first order texture features analysis.

RESULTS

There were significantly more lytic lesions in the axial skeleton than in the appendicular skeleton (p = 0.037). Out of 926 focal lesions in the axial skeleton seen on MRI, 544 (59.3 %) were osteolytic. Analysis of volume and first order texture features showed differences in texture and volume between focal lesions in MRI with and without local bone destruction in CT, but these findings were not statistically significant.

CONCLUSIONS

Neither morphological imaging characteristics like size and location nor first order texture features could predict whether focal lesions seen in MRI would exhibit corresponding bone destruction in CT. Studies performing biopsies of such lesions are ongoing.

Same-day repeatability and Between-Sequence reproducibility of Mean ADC in PI-RADS lesions

PURPOSE

This study aimed to assess repeatability after repositioning (inter-scan), intra-rater, inter-rater and inter-sequence variability of mean apparent diffusion coefficient (ADC) measurements in MRI-detected prostate lesions.

METHOD

Forty-three patients with suspicion for prostate cancer were included and received a clinical prostate bi-/multiparametric MRI examination with repeat scans of the T2-weighted and two DWI-weighted sequences (ssEPI and rsEPI). Two raters (R1 and R2) performed single-slice, 2D regions of interest (2D-ROIs) and 3D-segmentation-ROIs (3D-ROIs). Mean bias, corresponding limits of agreement (LoA), mean absolute difference, within-subject coefficient of variation (CoV) and repeatability/reproducibility coefficient (RC/RDC) were calculated. Bradley & Blackwood test was used for variance comparison. Linear mixed models (LMM) were used to account for multiple lesions per patient.

RESULTS

Inter-scan repeatability, intra-rater and inter-sequence reproducibility analysis of ADC showed no significant bias. 3D-ROIs demonstrated significantly less variability than 2D-ROIs (p < 0.01). Inter-rater comparison demonstrated small significant systematic bias of 57 × 10-6 mm2/s for 3D-ROIs (p < 0.001). Intra-rater RC, with the lowest variation, was 145 and 189 × 10-6 mm2/s for 3D- and 2D-ROIs, respectively. For 3D-ROIs of ssEPI, RCs and RDCs were 190-198 × 10-6 mm2/s for inter-scan, inter-rater and inter-sequence variation. No significant differences were found for inter-scan, inter-rater and inter-sequence variability.

CONCLUSIONS

In a single-scanner setting, single-slice ADC measurements showed considerable variation, which may be lowered using 3D-ROIs. For 3D-ROIs, we propose a cut-off of ∼ 200 × 10-6 mm2/s for differences introduced by repositioning, rater or sequence effects. The results suggest that follow-up measurements should be possible by different raters or sequences.

An MRI-based radiomics nomogram for differentiating spinal metastases from multiple myeloma

BACKGROUND

Spinal metastasis and multiple myeloma share many overlapping conventional radiographic imaging characteristics, thus, their differentiation may be challenging. The purpose of this study was to develop and validate an MRI-based radiomics nomogram for the differentiation of spinal metastasis and multiple myeloma.

MATERIALS AND METHODS

A total of 312 patients (training set: n = 146, validation set: n = 65, our center; external test set: n = 101, two other centers) with spinal metastasis (n = 196) and multiple myeloma (n = 116) were retrospectively enrolled. Demographics and MRI findings were assessed to build a clinical factor model. Radiomics features were extracted from MRI images. A radiomics model was constructed by the least absolute shrinkage and selection operator method. A radiomics nomogram combining the radiomics signature and independent clinical factors was constructed. And, one experienced radiologist reviewed the MRI images for all case. The diagnostic performance of the different models was evaluated by receiver operating characteristic curves.

RESULTS

A clinical factors model was built based on heterogeneous appearance and shape. Twenty-one features were used to build the radiomics signature. The area under the curve (AUC) values of the radiomics nomogram (0.853 and 0.762, respectively) were significantly higher than that of the clinical factor model (0.692 and 0.540, respectively) in both validation (p = 0.048) and external test (p < 0.001) sets. The AUC values of the radiomics nomogram model were higher than that of radiologist in training, validation and external test sets (all p < 0.05). Moreover, no significant difference in AUC values of radiomics nomogram model was found between the validation set and external test set (p = 0.212).

CONCLUSION

The radiomics nomogram can differentiate spinal metastasis and multiple myeloma with a moderate to good performance, and may be as a valuable method to assist in the clinical diagnosis and preoperative decision-making.

Differentiation of benign versus malignant indistinguishable vertebral compression fractures by different machine learning with MRI-based radiomic features

OBJECTIVES

To explore an optimal machine learning (ML) model trained on MRI-based radiomic features to differentiate benign from malignant indistinguishable vertebral compression fractures (VCFs).

METHODS

This retrospective study included patients within 6 weeks of back pain (non-traumatic) who underwent MRI and were diagnosed with benign and malignant indistinguishable VCFs. The two cohorts were retrospectively recruited from the Affiliated Hospital of Qingdao University (QUH) and Qinghai Red Cross Hospital (QRCH). Three hundred seventy-six participants from QUH were divided into the training (n = 263) and validation (n = 113) cohort based on the date of MRI examination. One hundred three participants from QRCH were used to evaluate the external generalizability of our prediction models. A total of 1045 radiomic features were extracted from each region of interest (ROI) and used to establish the models. The prediction models were established based on 7 different classifiers.

RESULTS

These models showed favorable efficacy in differentiating benign from malignant indistinguishable VCFs. However, our Gaussian naïve Bayes (GNB) model attained higher AUC and accuracy (0.86, 87.61%) than the other classifiers in validation cohort. It also remains the high accuracy and sensitivity for the external test cohort.

CONCLUSIONS

Our GNB model performed better than the other models in the present study, suggesting that it may be more useful for differentiating indistinguishable benign form malignant VCFs.

KEY POINTS

• The differential diagnosis of benign and malignant indistinguishable VCFs based on MRI is rather difficult for spine surgeons or radiologists. • Our ML models facilitate the differential diagnosis of benign and malignant indistinguishable VCFs with improved diagnostic efficacy. • Our GNB model had the high accuracy and sensitivity for clinical application.

Texture Analysis for the Bone Age Assessment from MRI Images of Adolescent Wrists in Boys

Currently, bone age is assessed by X-rays. It enables the evaluation of the child's development and is an important diagnostic factor. However, it is not sufficient to diagnose a specific disease because the diagnoses and prognoses may arise depending on how much the given case differs from the norms of bone age.

BACKGROUND

The use of magnetic resonance images (MRI) to assess the age of the patient would extend diagnostic possibilities. The bone age test could then become a routine screening test. Changing the method of determining the bone age would also prevent the patient from taking a dose of ionizing radiation, making the test less invasive.

METHODS

The regions of interest containing the wrist area and the epiphyses of the radius are marked on the magnetic resonance imaging of the non-dominant hand of boys aged 9 to 17 years. Textural features are computed for these regions, as it is assumed that the texture of the wrist image contains information about bone age.

RESULTS

The regression analysis revealed that there is a high correlation between the bone age of a patient and the MRI-derived textural features derived from MRI. For DICOM T1-weighted data, the best scores reached 0.94 R2, 0.46 RMSE, 0.21 MSE, and 0.33 MAE.

CONCLUSIONS

The experiments performed have shown that using the MRI images gives reliable results in the assessment of bone age while not exposing the patient to ionizing radiation.

In Vivo Repeatability and Multiscanner Reproducibility of MRI Radiomics Features in Patients With Monoclonal Plasma Cell Disorders: A Prospective Bi-institutional Study

OBJECTIVES

Despite the extensive number of publications in the field of radiomics, radiomics algorithms barely enter large-scale clinical application. Supposedly, the low external generalizability of radiomics models is one of the main reasons, which hinders the translation from research to clinical application. The objectives of this study were to investigate reproducibility of radiomics features (RFs) in vivo under variation of patient positioning, magnetic resonance imaging (MRI) sequence, and MRI scanners, and to identify a subgroup of RFs that shows acceptable reproducibility across all different acquisition scenarios.

MATERIALS AND METHODS

Between November 30, 2020 and February 16, 2021, 55 patients with monoclonal plasma cell disorders were included in this prospective, bi-institutional, single-vendor study. Participants underwent one reference scan at a 1.5 T MRI scanner and several retest scans: once after simple repositioning, once with a second MRI protocol, once at another 1.5 T scanner, and once at a 3 T scanner. Radiomics feature from the bone marrow of the left hip bone were extracted, both from original scans and after different image normalizations. Intraclass correlation coefficient (ICC) was used to assess RF repeatability and reproducibility.

RESULTS

Fifty-five participants (mean age, 59 ± 7 years; 36 men) were enrolled. For T1-weighted images after muscle normalization, in the simple test-retest experiment, 110 (37%) of 295 RFs showed an ICC ≥0.8: 54 (61%) of 89 first-order features (FOFs), 35 (95%) of 37 volume and shape features, and 21 (12%) of 169 texture features (TFs). When the retest was performed with different technical settings, even after muscle normalization, the number of FOF/TF with an ICC ≥0.8 declined to 58/13 for the second protocol, 29/7 for the second 1.5 T scanner, and 49/7 for the 3 T scanner, respectively. Twenty-five (28%) of the 89 FOFs and 6 (4%) of the 169 TFs from muscle-normalized T1-weighted images showed an ICC ≥0.8 throughout all repeatability and reproducibility experiments.

CONCLUSIONS

In vivo, only few RFs are reproducible with different MRI sequences or different MRI scanners, even after application of a simple image normalization. Radiomics features selected by a repeatability experiment only are not necessarily suited to build radiomics models for multicenter clinical application. This study isolated a subset of RFs, which are robust to variations in MRI acquisition observed in scanners from 1 vendor, and therefore are candidates to build reproducible radiomics models for monoclonal plasma cell disorders for multicentric applications, at least when centers are equipped with scanners from this vendor.

Deep Learning for Automatic Bone Marrow Apparent Diffusion Coefficient Measurements From Whole-Body Magnetic Resonance Imaging in Patients With Multiple Myeloma: A Retrospective Multicenter Study

OBJECTIVES

Diffusion-weighted magnetic resonance imaging (MRI) is increasingly important in patients with multiple myeloma (MM). The objective of this study was to train and test an algorithm for automatic pelvic bone marrow analysis from whole-body apparent diffusion coefficient (ADC) maps in patients with MM, which automatically segments pelvic bones and subsequently extracts objective, representative ADC measurements from each bone.

MATERIALS AND METHODS

In this retrospective multicentric study, 180 MRIs from 54 patients were annotated (semi)manually and used to train an nnU-Net for automatic, individual segmentation of the right hip bone, the left hip bone, and the sacral bone. The quality of the automatic segmentation was evaluated on 15 manually segmented whole-body MRIs from 3 centers using the dice score. In 3 independent test sets from 3 centers, which comprised a total of 312 whole-body MRIs, agreement between automatically extracted mean ADC values from the nnU-Net segmentation and manual ADC measurements from 2 independent radiologists was evaluated. Bland-Altman plots were constructed, and absolute bias, relative bias to mean, limits of agreement, and coefficients of variation were calculated. In 56 patients with newly diagnosed MM who had undergone bone marrow biopsy, ADC measurements were correlated with biopsy results using Spearman correlation.

RESULTS

The ADC-nnU-Net achieved automatic segmentations with mean dice scores of 0.92, 0.93, and 0.85 for the right pelvis, the left pelvis, and the sacral bone, whereas the interrater experiment gave mean dice scores of 0.86, 0.86, and 0.77, respectively. The agreement between radiologists' manual ADC measurements and automatic ADC measurements was as follows: the bias between the first reader and the automatic approach was 49 × 10 -6 mm 2 /s, 7 × 10 -6 mm 2 /s, and -58 × 10 -6 mm 2 /s, and the bias between the second reader and the automatic approach was 12 × 10 -6 mm 2 /s, 2 × 10 -6 mm 2 /s, and -66 × 10 -6 mm 2 /s for the right pelvis, the left pelvis, and the sacral bone, respectively. The bias between reader 1 and reader 2 was 40 × 10 -6 mm 2 /s, 8 × 10 -6 mm 2 /s, and 7 × 10 -6 mm 2 /s, and the mean absolute difference between manual readers was 84 × 10 -6 mm 2 /s, 65 × 10 -6 mm 2 /s, and 75 × 10 -6 mm 2 /s. Automatically extracted ADC values significantly correlated with bone marrow plasma cell infiltration ( R = 0.36, P = 0.007).

CONCLUSIONS

In this study, a nnU-Net was trained that can automatically segment pelvic bone marrow from whole-body ADC maps in multicentric data sets with a quality comparable to manual segmentations. This approach allows automatic, objective bone marrow ADC measurements, which agree well with manual ADC measurements and can help to overcome interrater variability or nonrepresentative measurements. Automatically extracted ADC values significantly correlate with bone marrow plasma cell infiltration and might be of value for automatic staging, risk stratification, or therapy response assessment.

Influence of Image Processing on Radiomic Features From Magnetic Resonance Imaging

OBJECTIVE

Before implementing radiomics in routine clinical practice, comprehensive knowledge about the repeatability and reproducibility of radiomic features is required. The aim of this study was to systematically investigate the influence of image processing parameters on radiomic features from magnetic resonance imaging (MRI) in terms of feature values as well as test-retest repeatability.

MATERIALS AND METHODS

Utilizing a phantom consisting of 4 onions, 4 limes, 4 kiwifruits, and 4 apples, we acquired a test-retest dataset featuring 3 of the most commonly used MRI sequences on a 3 T scanner, namely, a T1-weighted, a T2-weighted, and a fluid-attenuated inversion recovery sequence, each at high and low resolution. After semiautomatic image segmentation, image processing with systematic variation of image processing parameters was performed, including spatial resampling, intensity discretization, and intensity rescaling. For each respective image processing setting, a total of 45 radiomic features were extracted, corresponding to the following 7 matrices/feature classes: conventional indices, histogram matrix, shape matrix, gray-level zone length matrix, gray-level run length matrix, neighboring gray-level dependence matrix, and gray-level cooccurrence matrix. Systematic differences of individual features between different resampling steps were assessed using 1-way analysis of variance with Tukey-type post hoc comparisons to adjust for multiple testing. Test-retest repeatability of radiomic features was measured using the concordance correlation coefficient, dynamic range, and intraclass correlation coefficient.

RESULTS

Image processing influenced radiological feature values. Regardless of the acquired sequence and feature class, significant differences ( P < 0.05) in feature values were found when the size of the resampled voxels was too large, that is, bigger than 3 mm. Almost all higher-order features depended strongly on intensity discretization. The effects of intensity rescaling were negligible except for some features derived from T1-weighted sequences. For all sequences, the percentage of repeatable features (concordance correlation coefficient and dynamic range ≥ 0.9) varied considerably depending on the image processing settings. The optimal image processing setting to achieve the highest percentage of stable features varied per sequence. Irrespective of image processing, the fluid-attenuated inversion recovery sequence in high-resolution overall yielded the highest number of stable features in comparison with the other sequences (89% vs 64%-78% for the respective optimal image processing settings). Across all sequences, the most repeatable features were generally obtained for a spatial resampling close to the originally acquired voxel size and an intensity discretization to at least 32 bins.

CONCLUSION

Variation of image processing parameters has a significant impact on the values of radiomic features as well as their repeatability. Furthermore, the optimal image processing parameters differ for each MRI sequence. Therefore, it is recommended that these processing parameters be determined in corresponding test-retest scans before clinical application. Extensive repeatability, reproducibility, and validation studies as well as standardization are required before quantitative image analysis and radiomics can be reliably translated into routine clinical care.

Whole Body MRI with DWI in People with NF1 and Schwannomatosis: Are Qualitative and Quantitative Imaging Features of Peripheral Lesions Comparable to Localized MRI?

PURPOSE

To compare the qualitative and quantitative features of peripheral lesions on localized (L) and whole-body (WB) magnetic resonance imaging (MRI) in people with neurofibromatosis type 1 (NF1) and schwannomatosis.

MATERIALS AND METHODS

This is a retrospective, HIPAA compliant study with twenty-seven patients (14 women, 13 men; mean age (years): 38 (3-67)) who underwent both L-MRI and WB-MRI without interval treatment. WB-MRI and L-MRI were comprised of T1-weighted, fat suppressed (FS) T2-weighted or short tau inversion recovery (STIR), diffusion-weighted imaging (DWI) using b-values of 50, 400, and 800 s/mm2, apparent diffusion coefficient (ADC) mapping and pre- and post-contrast FST1 sequences. Two readers recorded qualitative (T1 and T2/STIR signal intensity and heterogeneity, contrast enhancement and heterogeneity, perilesional enhancement, presence of a target sign and perilesional edema) and quantitative (size, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), ADC) features of peripheral lesions on L-MRI and WB-MRI.Descriptive statistics, Wilcoxon signed-rank test and McNemar's test were used.

RESULTS

There were 31 peripheral lesions identified in 27 subjects, (mean size: 3.1 cm (range: 1-8.1 cm) on both L-MRI and WB-MRI).There were no differences in T1 signal and heterogeneity and T2/STIR signal and heterogeneity between WB-MRI and L-MRI ((p = 0.180, 0.083, 0.317 and 0.157 respectively). There were also no differences in contrast enhancement, heterogeneity and perilesional enhancement between WB-MRI and L-MRI (p = 1.000, 0.380 and 1.000 respectively). Presence of a target sign and perilesional edema did not differ between WB-MRI and L-MRI (p = 1.000 and 0.500 respectively). Craniocaudal (CC), mediolateral (ML) and anteroposterior (AP) size measurements on WB-MRI did not differ from CC, ML and AP size measurements on L-MRI (p = 0.597, 0.128 and 0.783 respectively). SNR on WB-DWI did not differ from SNR on L-DWI for b50, b400 and b800 images (p = 0.285, 0.166, and 0.974 respectively), and CNR on WB-DWI did not differ from CNR on L-DWI for b50, b400 and b800 images (p = 0.600, 0.124, and 0.787 respectively). There was no significant difference in minimum, mean and maximum ADC values between WB-DWI and L-DWI (p = 0.234, 0.481, and 0.441 respectively). Median minimum, mean and maximum ADC (×10(-3)mm(2)/s) differences between WB-DWI and L-DWI were 0.0 (range -1 to 0.7), 0.0 (range -0.5 to 0.6), and 0.1 (range -1.2 to 0.8) respectively. Relative ADC difference averages were 29.1% for minimum values, 10.1% for mean values, and 14.8% for maximum values.

CONCLUSION

WB-MRI yields qualitative and quantitative features for peripheral lesions, including DWI and ADC measurements, that are comparable to L-MRI scans. WB-DWI can be reliably used for the assessment of peripheral nerve sheath tumors, obviating the need for a repeat follow-up L-DWI acquisition.

Quantification and reduction of cross-vendor variation in multicenter DWI MR imaging: results of the Cancer Core Europe imaging task force

OBJECTIVES

In the Cancer Core Europe Consortium (CCE), standardized biomarkers are required for therapy monitoring oncologic multicenter clinical trials. Multiparametric functional MRI and particularly diffusion-weighted MRI offer evident advantages for noninvasive characterization of tumor viability compared to CT and RECIST. A quantification of the inter- and intraindividual variation occurring in this setting using different hardware is missing. In this study, the MRI protocol including DWI was standardized and the residual variability of measurement parameters quantified.

METHODS

Phantom and volunteer measurements (single-shot T2w and DW-EPI) were performed at the seven CCE sites using the MR hardware produced by three different vendors. Repeated measurements were performed at the sites and across the sites including a traveling volunteer, comparing qualitative and quantitative ROI-based results including an explorative radiomics analysis.

RESULTS

For DWI/ADC phantom measurements using a central post-processing algorithm, the maximum deviation could be decreased to 2%. However, there is no significant difference compared to a decentralized ADC value calculation at the respective MRI devices. In volunteers, the measurement variation in 2 repeated scans did not exceed 11% for ADC and is below 20% for single-shot T2w in systematic liver ROIs. The measurement variation between sites amounted to 20% for ADC and < 25% for single-shot T2w. Explorative radiomics classification experiments yield better results for ADC than for single-shot T2w.

CONCLUSION

Harmonization of MR acquisition and post-processing parameters results in acceptable standard deviations for MR/DW imaging. MRI could be the tool in oncologic multicenter trials to overcome the limitations of RECIST-based response evaluation.

KEY POINTS

• Harmonizing acquisition parameters and post-processing homogenization, standardized protocols result in acceptable standard deviations for multicenter MR-DWI studies. • Total measurement variation does not to exceed 11% for ADC in repeated measurements in repeated MR acquisitions, and below 20% for an identical volunteer travelling between sites. • Radiomic classification experiments were able to identify stable features allowing for reliable discrimination of different physiological tissue samples, even when using heterogeneous imaging data.

Repeatability and test-retest reproducibility of mean apparent diffusion coefficient measurements of focal and diffuse disease in relapsed multiple myeloma at 3T whole body diffusion-weighted MRI (WB-DW-MRI)

OBJECTIVE

To assess the test-retest reproducibility and intra/interobserver agreement of apparent diffusion coefficient (ADC) measurements of myeloma lesions using whole body diffusion-weighted MRI (WB-DW-MRI) at 3T MRI.

METHODS

Following ethical approval, 11 consenting patients with relapsed multiple myeloma were prospectively recruited and underwent baseline WB-DW-MRI. For a single bed position, axial DWI was repeated after a short interval to permit test-retest measurements.Mean ADC measurement was performed by two experienced observers. Intra- and interobserver agreement and test-retest reproducibility were assessed, using coefficient of variation (CV) and interclass correlation coefficient (ICC) measures, for diffuse and focal lesions (small ≤10 mm and large >10 mm).

RESULTS

47 sites of disease were outlined (23 focal, 24 diffuse) in different bed positions (pelvis = 22, thorax = 20, head and neck = 5). For all lesions, there was excellent intraobserver agreement with ICC of 0.99 (0.98-0.99) and COV of 5%. For interobserver agreement, ICC was 0.89 (0.8-0.934) and COV was 17%. There was poor interobserver agreement for diffuse disease (ICC = 0.46) and small lesions (ICC = 0.54).For test-retest reproducibility, excellent ICC (0.916) and COV (14.5%) values for mean ADC measurements were observed. ICCs of test-retest were similar between focal lesions (0.83) and diffuse infiltration (0.80), while ICCs were higher in pelvic (0.95) compared to thoracic (0.81) region and in small (0.96) compared to large (0.8) lesions.

CONCLUSION

ADC measurements of focal lesions in multiple myeloma are repeatable and reproducible, while there is more variation in ADC measurements of diffuse disease in patients with multiple myeloma.

ADVANCES IN KNOWLEDGE

Mean ADC measurements are repeatable and reproducible in focal lesions in multiple myeloma, while the ADC measurements of diffuse disease in multiple myeloma are more subject to variation. The evidence supports the future potential role of ADC measurements as predictive quantitative biomarker in multiple myeloma.

Whole-body magnetic resonance imaging plus serological follow-up for early identification of progression in smouldering myeloma patients to prevent development of end-organ damage

The definition of multiple myeloma (MM) was updated in 2014, with the intent to enable earlier treatment and thereby avoid appearance of end-organ damage at progression from smouldering multiple myeloma (SMM) to MM. The purpose of this study was to investigate to which extent the development of end-organ damage at progression to MM was reduced under the updated guidelines. In this prospective observational cohort study (ClinicalTrials.gov Identifier: NCT01374412), between 2014 and 2020, 96 SMM patients prospectively underwent whole-body magnetic resonance imaging (wb-MRI) and serological follow-up at baseline and every 6 months thereafter. A total of 22 patients progressed into MM during follow-up, of which seven (32%) showed SLiM-criteria only but no end-organ damage. Four (57%) of the seven patients who progressed by SLiM-criteria only progressed with >1 focal lesion (FL) or a growing FL, and three (43%) due to serum free light-chain-ratio ≥100. Fifteen (68%) out of 22 patients who progressed still suffered from end-organ damage at progression. The updated disease definition reduced the proportion of SMM patients suffering from end-organ damage at progression to MM by one third. wb-MRI is an important tool for detection of SMM patients who progress to MM without end-organ damage.

[Potential of radiomics and artificial intelligence in myeloma imaging : Development of automatic, comprehensive, objective skeletal analyses from whole-body imaging data]

CLINICAL/METHODICAL ISSUE

Multiple myeloma can affect the complete skeleton, which makes whole-body imaging necessary. With the current assessment of these complex datasets by radiologists, only a small part of the accessible information is assessed and reported.

STANDARD RADIOLOGICAL METHODS

Depending on the question and availability, computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET) is performed and the results are then visually examined by radiologists.

METHODOLOGICAL INNOVATIONS

A combination of automatic skeletal segmentation using artificial intelligence and subsequent radiomics analysis of each individual bone have the potential to provide automatic, comprehensive, and objective skeletal analyses.

PERFORMANCE

A few automatic skeletal segmentation algorithms for CT already show promising results. In addition, first studies indicate correlations between radiomics features of bone and bone marrow with established disease markers and therapy response.

ACHIEVEMENTS

Artificial intelligence (AI) and radiomics algorithms for automatic skeletal analysis from whole-body imaging are currently in an early phase of development.