DicomAnnotator: a Configurable Open-Source Software Program for Efficient DICOM Image Annotation

Using an Ensemble of Segmentation Methods to Detect Vertebral Bodies on Radiographs

BACKGROUND AND PURPOSE

Vertebral compression fractures may indicate osteoporosis but are underdiagnosed and underreported by radiologists. We have developed an ensemble of vertebral body (VB) segmentation models for lateral radiographs as a critical component of an automated, opportunistic screening tool. Our goal is to detect the approximate location of thoracic and lumbar VBs, including fractured vertebra, on lateral radiographs.

MATERIALS AND METHODS

The Osteoporotic Fractures in Men Study (MrOS) data set includes spine radiographs of 5994 men aged ≥65 years from 6 clinical centers. Two segmentation models, U-Net and Mask-RCNN (Region-based Convolutional Neural Network), were independently trained on the MrOS data set retrospectively, and an ensemble was created by combining them. Primary performance metrics for VB detection success included precision, recall, and F1 score for object detection on a held-out test set. Intersection over union (IoU) and Dice coefficient were also calculated as secondary metrics of performance for the test set. A separate external data set from a quaternary health care enterprise was acquired to test generalizability, comprising diagnostic clinical radiographs from men and women aged ≥65 years.

RESULTS

The trained models achieved F1 score of U-Net = 83.42%, Mask-RCNN = 86.30%, and ensemble = 88.34% in detecting all VBs, and F1 score of U-Net = 87.88%, Mask-RCNN = 92.31%, and ensemble = 97.14% in detecting severely fractured vertebrae. The trained models achieved an average IoU per VB of 0.759 for U-Net and 0.709 for Mask-RCNN. The trained models achieved F1 score of U-Net = 81.11%, Mask-RCNN = 79.24%, and ensemble = 87.72% in detecting all VBs in the external data set.

CONCLUSIONS

An ensemble model combining predictions from U-Net and Mask-RCNN resulted in the best performance in detecting VBs on lateral radiographs and generalized well to an external data set. This model could be a key component of a pipeline to detect fractures on all vertebrae in a radiograph in an automated, opportunistic screening tool under development.

SCO-spondin knockout mice exhibit small brain ventricles and mild spine deformation

Reissner's fiber (RF) is an extracellular polymer comprising the large monomeric protein SCO-spondin (SSPO) secreted by the subcommissural organ (SCO) that extends through cerebrospinal fluid (CSF)-filled ventricles into the central canal of the spinal cord. In zebrafish, RF and CSF-contacting neurons (CSF-cNs) form an axial sensory system that detects spinal curvature, instructs morphogenesis of the body axis, and enables proper alignment of the spine. In mammalian models, RF has been implicated in CSF circulation. However, challenges in manipulating Sspo, an exceptionally large gene of 15,719 nucleotides, with traditional approaches has limited progress. Here, we generated a Sspo knockout mouse model using CRISPR/Cas9-mediated genome-editing. Sspo knockout mice lacked RF-positive material in the SCO and fibrillar condensates in the brain ventricles. Remarkably, Sspo knockout brain ventricle sizes were reduced compared to littermate controls. Minor defects in thoracic spine curvature were detected in Sspo knockouts, which did not alter basic motor behaviors tested. Altogether, our work in mouse demonstrates that SSPO and RF regulate ventricle size during development but only moderately impact spine geometry.

Generalizability of Deep Learning Classification of Spinal Osteoporotic Compression Fractures on Radiographs Using an Adaptation of the Modified-2 Algorithm-Based Qualitative Criteria

RATIONALE AND OBJECTIVES

Spinal osteoporotic compression fractures (OCFs) can be an early biomarker for osteoporosis but are often subtle, incidental, and underreported. To ensure early diagnosis and treatment of osteoporosis, we aimed to build a deep learning vertebral body classifier for OCFs as a critical component of our future automated opportunistic screening tool.

MATERIALS AND METHODS

We retrospectively assembled a local dataset, including 1790 subjects and 15,050 vertebral bodies (thoracic and lumbar). Each vertebral body was annotated using an adaption of the modified-2 algorithm-based qualitative criteria. The Osteoporotic Fractures in Men (MrOS) Study dataset provided thoracic and lumbar spine radiographs of 5994 men from six clinical centers. Using both datasets, five deep learning algorithms were trained to classify each individual vertebral body of the spine radiographs. Classification performance was compared for these models using multiple metrics, including the area under the receiver operating characteristic curve (AUC-ROC), sensitivity, specificity, and positive predictive value (PPV).

RESULTS

Our best model, built with ensemble averaging, achieved an AUC-ROC of 0.948 and 0.936 on the local dataset's test set and the MrOS dataset's test set, respectively. After setting the cutoff threshold to prioritize PPV, this model achieved a sensitivity of 54.5% and 47.8%, a specificity of 99.7% and 99.6%, and a PPV of 89.8% and 94.8%.

CONCLUSION

Our model achieved an AUC-ROC>0.90 on both datasets. This testing shows some generalizability to real-world clinical datasets and a suitable performance for a future opportunistic osteoporosis screening tool.

SCO-spondin knockout mice exhibit small brain ventricles and mild spine deformation

Reissner's fiber (RF) is an extracellular polymer comprising the large monomeric protein SCO-spondin (SSPO) secreted by the subcommissural organ (SCO) that extends through cerebrospinal fluid (CSF)-filled ventricles into the central canal of the spinal cord. In zebrafish, RF and CSF-contacting neurons (CSF-cNs) form an axial sensory system that detects spinal curvature, instructs morphogenesis of the body axis, and enables proper alignment of the spine. In mammalian models, RF has been implicated in CSF circulation. However, challenges in manipulating Sspo , an exceptionally large gene of 15,719 nucleotides, with traditional approaches has limited progress. Here, we generated a Sspo knockout mouse model using CRISPR/Cas9-mediated genome-editing. Sspo knockout mice lacked RF-positive material in the SCO and fibrillar condensates in the brain ventricles. Remarkably, Sspo knockout brain ventricle sizes were reduced compared to littermate controls. Minor defects in thoracic spine curvature were detected in Sspo knockouts, which did not alter basic motor behaviors tested. Altogether, our work in mouse demonstrates that SSPO and RF regulate ventricle size during development but only moderately impact spine geometry.

An Orchestration Platform that Puts Radiologists in the Driver's Seat of AI Innovation: a Methodological Approach

Current AI-driven research in radiology requires resources and expertise that are often inaccessible to small and resource-limited labs. The clinicians who are able to participate in AI research are frequently well-funded, well-staffed, and either have significant experience with AI and computing, or have access to colleagues or facilities that do. Current imaging data is clinician-oriented and is not easily amenable to machine learning initiatives, resulting in inefficient, time consuming, and costly efforts that rely upon a crew of data engineers and machine learning scientists, and all too often preclude radiologists from driving AI research and innovation. We present the system and methodology we have developed to address infrastructure and platform needs, while reducing the staffing and resource barriers to entry. We emphasize a data-first and modular approach that streamlines the AI development and deployment process while providing efficient and familiar interfaces for radiologists, such that they can be the drivers of new AI innovations.

m2ABQ-a proposed refinement of the modified algorithm-based qualitative classification of osteoporotic vertebral fractures

Currently, there is no reproducible, widely accepted gold standard to classify osteoporotic vertebral body fractures (OVFs). The purpose of this study is to refine a method with clear rules to classify OVFs for machine learning purposes. The method was found to have moderate interobserver agreement that improved with training.

INTRODUCTION

The current methods to classify osteoporotic vertebral body fractures are considered ambiguous; there is no reproducible, accepted gold standard. The purpose of this study is to refine classification methodology by introducing clear, unambiguous rules and a refined flowchart to allow consistent classification of osteoporotic vertebral body fractures.

METHODS

We developed a set of rules and refinements that we called m2ABQ to classify vertebrae into five categories. A fracture-enriched database of thoracic and lumbar spine radiographs of patients 65 years of age and older was retrospectively obtained from clinical institutional radiology records using natural language processing. Five raters independently classified each vertebral body using the m2ABQ system. After each annotation round, consensus sessions that included all raters were held to discuss and finalize a consensus annotation for each vertebral body where individual raters' evaluations differed. This process led to further refinement and development of the rules.

RESULTS

Each annotation round showed increase in Fleiss kappa both for presence vs absence of fracture 0.62 (0.56-0.68) to 0.70 (0.65-0.75), as well as for the whole m2ABQ scale 0.29 (0.25-0.33) to 0.54 (0.51-0.58).

CONCLUSION

The m2ABQ system demonstrates moderate interobserver agreement and practical feasibility for classifying osteoporotic vertebral body fractures. Future studies to compare the method to existing studies are warranted, as well as further development of its use in machine learning purposes.

AnatomySketch: An Extensible Open-Source Software Platform for Medical Image Analysis Algorithm Development

The development of medical image analysis algorithm is a complex process including the multiple sub-steps of model training, data visualization, human-computer interaction and graphical user interface (GUI) construction. To accelerate the development process, algorithm developers need a software tool to assist with all the sub-steps so that they can focus on the core function implementation. Especially, for the development of deep learning (DL) algorithms, a software tool supporting training data annotation and GUI construction is highly desired. In this work, we constructed AnatomySketch, an extensible open-source software platform with a friendly GUI and a flexible plugin interface for integrating user-developed algorithm modules. Through the plugin interface, algorithm developers can quickly create a GUI-based software prototype for clinical validation. AnatomySketch supports image annotation using the stylus and multi-touch screen. It also provides efficient tools to facilitate the collaboration between human experts and artificial intelligent (AI) algorithms. We demonstrate four exemplar applications including customized MRI image diagnosis, interactive lung lobe segmentation, human-AI collaborated spine disc segmentation and Annotation-by-iterative-Deep-Learning (AID) for DL model training. Using AnatomySketch, the gap between laboratory prototyping and clinical testing is bridged and the development of MIA algorithms is accelerated. The software is opened at https://github.com/DlutMedimgGroup/AnatomySketch-Software .