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

Deep learning-based multimodal segmentation of oropharyngeal squamous cell carcinoma on CT and MRI using self-configuring nnU-Net

PURPOSE

To evaluate deep learning-based segmentation models for oropharyngeal squamous cell carcinoma (OPSCC) using CT and MRI with nnU-Net.

METHODS

This single-center retrospective study included 91 patients with OPSCC. The patients were grouped into the development (n = 56), test 1 (n = 13), and test 2 (n = 22) cohorts. In the development cohort, OPSCC was manually segmented on CT, MR, and co-registered CT-MR images, which served as the ground truth. The multimodal and multichannel input images were then trained using a self-configuring nnU-Net. For evaluation metrics, dice similarity coefficient (DSC) and mean Hausdorff distance (HD) were calculated for test cohorts. Pearson's correlation and Bland-Altman analyses were performed between ground truth and prediction volumes. Intraclass correlation coefficients (ICCs) of radiomic features were calculated for reproducibility assessment.

RESULTS

All models achieved robust segmentation performances with DSC of 0.64 ± 0.33 (CT), 0.67 ± 0.27 (MR), and 0.65 ± 0.29 (CT-MR) in test cohort 1 and 0.57 ± 0.31 (CT), 0.77 ± 0.08 (MR), and 0.73 ± 0.18 (CT-MR) in test cohort 2. No significant differences were found in DSC among the models. HD of CT-MR (1.57 ± 1.06 mm) and MR models (1.36 ± 0.61 mm) were significantly lower than that of the CT model (3.48 ± 5.0 mm) (p = 0.037 and p = 0.014, respectively). The correlation coefficients between the ground truth and prediction volumes for CT, MR, and CT-MR models were 0.88, 0.93, and 0.9, respectively. MR models demonstrated excellent mean ICCs of radiomic features (0.91-0.93).

CONCLUSION

The self-configuring nnU-Net demonstrated reliable and accurate segmentation of OPSCC on CT and MRI. The multimodal CT-MR model showed promising results for the simultaneous segmentation on CT and MRI.

CLINICAL RELEVANCE STATEMENT

Deep learning-based automatic detection and segmentation of oropharyngeal squamous cell carcinoma on pre-treatment CT and MRI would facilitate radiologic response assessment and radiotherapy planning.

KEY POINTS

• The nnU-Net framework produced a reliable and accurate segmentation of OPSCC on CT and MRI. • MR and CT-MR models showed higher DSC and lower Hausdorff distance than the CT model. • Correlation coefficients between the ground truth and predicted segmentation volumes were high in all the three models.

Advancing Medical Imaging Research Through Standardization: The Path to Rapid Development, Rigorous Validation, and Robust Reproducibility

ABSTRACT

Artificial intelligence (AI) has made significant advances in radiology. Nonetheless, challenges in AI development, validation, and reproducibility persist, primarily due to the lack of high-quality, large-scale, standardized data across the world. Addressing these challenges requires comprehensive standardization of medical imaging data and seamless integration with structured medical data.Developed by the Observational Health Data Sciences and Informatics community, the OMOP Common Data Model enables large-scale international collaborations with structured medical data. It ensures syntactic and semantic interoperability, while supporting the privacy-protected distribution of research across borders. The recently proposed Medical Imaging Common Data Model is designed to encompass all DICOM-formatted medical imaging data and integrate imaging-derived features with clinical data, ensuring their provenance.The harmonization of medical imaging data and its seamless integration with structured clinical data at a global scale will pave the way for advanced AI research in radiology. This standardization will enable federated learning, ensuring privacy-preserving collaboration across institutions and promoting equitable AI through the inclusion of diverse patient populations. Moreover, it will facilitate the development of foundation models trained on large-scale, multimodal datasets, serving as powerful starting points for specialized AI applications. Objective and transparent algorithm validation on a standardized data infrastructure will enhance reproducibility and interoperability of AI systems, driving innovation and reliability in clinical applications.

Enhancing diagnostic accuracy of multiple myeloma through ML-driven analysis of hematological slides: new dataset and identification model to support hematologists

Multiple Myeloma (MM) is a hematological malignancy characterized by the clonal proliferation of plasma cells within the bone marrow. Diagnosing MM presents considerable challenges, involving the identification of plasma cells in cytology examinations on hematological slides. At present, this is still a time-consuming manual task and has high labor costs. These challenges have adverse implications, which rely heavily on medical professionals' expertise and experience. To tackle these challenges, we present an investigation using Artificial Intelligence, specifically a Machine Learning analysis of hematological slides with a Deep Neural Network (DNN), to support specialists during the process of diagnosing MM. In this sense, the contribution of this study is twofold: in addition to the trained model to diagnose MM, we also make available to the community a fully-curated hematological slide dataset with thousands of images of plasma cells. Taken together, the setup we established here is a framework that researchers and hospitals with limited resources can promptly use. Our contributions provide practical results that have been directly applied in the public health system in Brazil. Given the open-source nature of the project, we anticipate it will be used and extended to diagnose other malignancies.

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.

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.

Development and Evaluation of Machine Learning in Whole-Body Magnetic Resonance Imaging for Detecting Metastases in Patients With Lung or Colon Cancer: A Diagnostic Test Accuracy Study

OBJECTIVES

Whole-body magnetic resonance imaging (WB-MRI) has been demonstrated to be efficient and cost-effective for cancer staging. The study aim was to develop a machine learning (ML) algorithm to improve radiologists' sensitivity and specificity for metastasis detection and reduce reading times.

MATERIALS AND METHODS

A retrospective analysis of 438 prospectively collected WB-MRI scans from multicenter Streamline studies (February 2013-September 2016) was undertaken. Disease sites were manually labeled using Streamline reference standard. Whole-body MRI scans were randomly allocated to training and testing sets. A model for malignant lesion detection was developed based on convolutional neural networks and a 2-stage training strategy. The final algorithm generated lesion probability heat maps. Using a concurrent reader paradigm, 25 radiologists (18 experienced, 7 inexperienced in WB-/MRI) were randomly allocated WB-MRI scans with or without ML support to detect malignant lesions over 2 or 3 reading rounds. Reads were undertaken in the setting of a diagnostic radiology reading room between November 2019 and March 2020. Reading times were recorded by a scribe. Prespecified analysis included sensitivity, specificity, interobserver agreement, and reading time of radiology readers to detect metastases with or without ML support. Reader performance for detection of the primary tumor was also evaluated.

RESULTS

Four hundred thirty-three evaluable WB-MRI scans were allocated to algorithm training (245) or radiology testing (50 patients with metastases, from primary 117 colon [n = 117] or lung [n = 71] cancer). Among a total 562 reads by experienced radiologists over 2 reading rounds, per-patient specificity was 86.2% (ML) and 87.7% (non-ML) (-1.5% difference; 95% confidence interval [CI], -6.4%, 3.5%; P = 0.39). Sensitivity was 66.0% (ML) and 70.0% (non-ML) (-4.0% difference; 95% CI, -13.5%, 5.5%; P = 0.344). Among 161 reads by inexperienced readers, per-patient specificity in both groups was 76.3% (0% difference; 95% CI, -15.0%, 15.0%; P = 0.613), with sensitivity of 73.3% (ML) and 60.0% (non-ML) (13.3% difference; 95% CI, -7.9%, 34.5%; P = 0.313). Per-site specificity was high (>90%) for all metastatic sites and experience levels. There was high sensitivity for the detection of primary tumors (lung cancer detection rate of 98.6% with and without ML [0.0% difference; 95% CI, -2.0%, 2.0%; P = 1.00], colon cancer detection rate of 89.0% with and 90.6% without ML [-1.7% difference; 95% CI, -5.6%, 2.2%; P = 0.65]). When combining all reads from rounds 1 and 2, reading times fell by 6.2% (95% CI, -22.8%, 10.0%) when using ML. Round 2 read-times fell by 32% (95% CI, 20.8%, 42.8%) compared with round 1. Within round 2, there was a significant decrease in read-time when using ML support, estimated as 286 seconds (or 11%) quicker ( P = 0.0281), using regression analysis to account for reader experience, read round, and tumor type. Interobserver variance suggests moderate agreement, Cohen κ = 0.64; 95% CI, 0.47, 0.81 (with ML), and Cohen κ = 0.66; 95% CI, 0.47, 0.81 (without ML).

CONCLUSIONS

There was no evidence of a significant difference in per-patient sensitivity and specificity for detecting metastases or the primary tumor using concurrent ML compared with standard WB-MRI. Radiology read-times with or without ML support fell for round 2 reads compared with round 1, suggesting that readers familiarized themselves with the study reading method. During the second reading round, there was a significant reduction in reading time when using ML support.

Radiomics Analysis for Multiple Myeloma: A Systematic Review with Radiomics Quality Scoring

Multiple myeloma (MM) is one of the most common hematological malignancies affecting the bone marrow. Radiomics analysis has been employed in the literature in an attempt to evaluate the bone marrow of MM patients. This manuscript aimed to systematically review radiomics research on MM while employing a radiomics quality score (RQS) to accurately assess research quality in the field. A systematic search was performed on Web of Science, PubMed, and Scopus. The selected manuscripts were evaluated (data extraction and RQS scoring) by three independent readers (R1, R2, and R3) with experience in radiomics analysis. A total of 23 studies with 2682 patients were included, and the median RQS was 10 for R1 (IQR 5.5-12) and R3 (IQR 8.3-12) and 11 (IQR 7.5-12.5) for R2. RQS was not significantly correlated with any of the assessed bibliometric data (impact factor, quartile, year of publication, and imaging modality) (p > 0.05). Our results demonstrated the low quality of published radiomics research in MM, similarly to other fields of radiomics research, highlighting the need to tighten publication standards.

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.