Vertebral Detection Research Articles

Automated detection, labelling and radiological grading of clinical spinal MRIs

Spinal magnetic resonance (MR) scans are a vital tool for diagnosing the cause of back pain for many diseases and conditions. However, interpreting clinically useful information from these scans can be challenging, time-consuming and hard to reproduce across different radiologists. In this paper, we alleviate these problems by introducing a multi-stage automated pipeline for analysing spinal MR scans. This pipeline first detects and labels vertebral bodies across several commonly used sequences (e.g. T1w, T2w and STIR) and fields of view (e.g. lumbar, cervical, whole spine). Using these detections it then performs automated diagnosis for several spinal disorders, including intervertebral disc degenerative changes in T1w and T2w lumbar scans, and spinal metastases, cord compression and vertebral fractures. To achieve this, we propose a new method of vertebrae detection and labelling, using vector fields to group together detected vertebral landmarks and a language-modelling inspired beam search to determine the corresponding levels of the detections. We also employ a new transformer-based architecture to perform radiological grading which incorporates context from multiple vertebrae and sequences, as a real radiologist would. The performance of each stage of the pipeline is tested in isolation on several clinical datasets, each consisting of 66 to 421 scans. The outputs are compared to manual annotations of expert radiologists, demonstrating accurate vertebrae detection across a range of scan parameters. Similarly, the model’s grading predictions for various types of disc degeneration and detection of spinal metastases closely match those of an expert radiologist. To aid future research, our code and trained models are made publicly available.

Exploring Neighbor Spatial Relationships for Enhanced Lumbar Vertebrae Detection in X-ray Images

Exploring Neighbor Spatial Relationships for Enhanced Lumbar Vertebrae Detection in X-ray Images

Anatomical Prior and Inter-slice Consistency for Semi-supervised Vertebral Structure Detection in 3D Ultrasound Volume.

Three-dimensional (3D) ultrasound imaging technique has been applied for scoliosis assessment, but the current assessment method only uses coronal projection images and cannot illustrate the 3D deformity and vertebra rotation. The vertebra detection is essential to reveal 3D spine information, but the detection task is challenging due to complex data and limited annotations. We propose VertMatch to detect vertebral structures in 3D ultrasound volume containing a detector and classifier. The detector network finds the potential positions of structures on transverse slice globally, and then the local patches are cropped based on detected positions. The classifier is used to distinguish whether the patches contain real vertebral structures and screen the predicted positions from the detector. VertMatch utilizes unlabeled data in a semi-supervised manner, and we develop two novel techniques for semi-supervised learning: (1) anatomical prior is used to acquire high-quality pseudo labels; (2) inter-slice consistency is used to utilize more unlabeled data by inputting multiple adjacent slices. Experimental results demonstrate that VertMatch can detect vertebra accurately in ultrasound volume and outperforms state-of-the-art methods. Moreover, VertMatch is also validated in automatic spinous process angle measurement on forty subjects with scoliosis, and the results illustrate that it can be a promising approach for the 3D assessment of scoliosis.

Utilizing mobile digital radiography for detection of thoracolumbar vertebrae traits in live donkeys.

It has been well-established that the number of vertebrae is associated with body size and meat productivity. In current study we utilized a digital radiography (DR) technology to detect the number of thoracolumbar vertebrae in live donkeys. For this purpose, we introduced for the first time a groundbreaking device designed by our team for assessing thoracolumbar vertebrae number traits in equids, employing a sample of 1,000 donkeys sourced from five distinct donkey farms. This assessment incorporates a range of crucial body metrics, including body height, length, and various other measurements. Subsequently, our study determined the number of thoracolumbar vertebrae in 112 donkeys, utilizing the DR system. These findings were further validated through post-mortem evaluations conducted by slaughtering the donkeys. Our findings demonstrated a remarkable resemblance between the thoracolumbar vertebrae numbers visualized through the DR system in live donkeys and those obtained via slaughter verification. In conclusion, this research underscores the accuracy and effectiveness of the DR system for the detection of thoracolumbar vertebrae in live donkeys, which might be helpful for assessing the body size and meat productivity. We also recommended the utilization of DR system for counting thoracolumbar vertebrae in other animals in live state and could be a useful addition to livestock business industry for the prediction of body size and meat productivity efficiency.

Auto-CA: Automated Cobb Angle Measurement Based on Vertebrae Detection for Assessment of Spinal Curvature Deformity.

An accurate identification and localization of vertebrae in X-ray images can assist doctors in measuring Cobb angles for treating patients with adolescent idiopathic scoliosis. It is useful for clinical decision support systems for diagnosis, surgery planning, and spinal health analysis. Currently, publicly available annotated datasets on spinal vertebrae are small, making deep-learning-based detection methods that are highly data-dependent less accurate. In this article, we propose an algorithm based on convolutional neural networks that can be trained to detect vertebrae from a small set of images. This method can display critical information on a patient's spine, display vertebrae and their labels on the thoracic and lumbar, calculate the Cobb angle, and evaluate the severity of spinal deformities. The proposed achieved an average accuracy of 0.958 and 0.962 for classifying spinal deformities (i.e., C-shaped, S-shaped type 1, and S-shaped type 2) and severity of Cobb angle (i.e., normal, mild, moderate, and severe), respectively. The Cobb angle measurement had a median difference of less than 5° from the ground-truth with SMAPE of 5.27% and an error on landmark detection of 19.73. In addition, Lenke classification is used to analyze spinal deformities as types A, B, and C, which have an average accuracy of 0.924. Physicians can use the proposed system in clinical practice by providing X-ray images via the user interface.

Deep learning-driven diagnosis of multi-type vertebra diseases based on computed tomography images.

Osteoporotic vertebral compression fractures (OVCFs) are the most common type of fragility fracture. Distinguishing between OVCFs and other types of vertebra diseases, such as old fractures (OFs), Schmorl's node (SN), Kummell's disease (KD), and previous surgery (PS), is critical for subsequent surgery and treatment. Combining with advanced deep learning (DL) technologies, this study plans to develop a DL-driven diagnostic system for diagnosing multi-type vertebra diseases. We established a large-scale dataset based on the computed tomography (CT) images of 1,051 patients with OVCFs from Luhe Hospital and used data of 46 patients from Xuanwu Hospital as alternative hospital validation dataset. Each patient underwent one examination. The dataset contained 11,417 CT slices and 19,718 manually annotated vertebrae with diseases. A two-stage DL-based system was developed to diagnose five vertebra diseases. The proposed system consisted of a vertebra detection module (VDModule) and a vertebra classification module (VCModule). The training and testing dataset for the VDModule consisted of 9,135 and 3,212 vertebrae, respectively. The VDModule using the ResNet18-based Faster region-based convolutional neural network (R-CNN) model achieved an area under the curve (AUC), false-positive (FP) rate, and false-negative (FN) rate of 0.982, 1.52%, and 1.33%, respectively, in the testing dataset. The training dataset for VCModule consisted of 14,584 and 47,604 diseased and normal vertebrae, respectively. The testing dataset consisted of 4,489 and 15,122 diseased and normal vertebrae, respectively. The ResNet50-based VCModule achieved an average sensitivity and specificity of 0.919 and 0.995, respectively, in diagnosing four kinds of vertebra diseases except for SN in the testing dataset. In the alternative hospital validation dataset, the ResNet50-based VCModule achieved an average sensitivity and specificity of 0.891 and 0.989, respectively, in diagnosing four kinds of vertebra diseases except for SN. Our proposed DL system can accurately diagnose four vertebra diseases and has strong potential to facilitate the accurate and rapid diagnosis of vertebral diseases.

VDVM: An automatic vertebrae detection and vertebral segment matching framework for C-arm X-ray image identification.

C-arm fluoroscopy, as an effective diagnosis and treatment method for spine surgery, can help doctors perform surgery procedures more precisely. In clinical surgery, the surgeon often determines the specific surgical location by comparing C-arm X-ray images with digital radiography (DR) images. However, this heavily relies on the doctor's experience. In this study, we design a framework for automatic vertebrae detection as well as vertebral segment matching (VDVM) for the identification of vertebrae in C-arm X-ray images. The proposed VDVM framework is mainly divided into two parts: vertebra detection and vertebra matching. In the first part, a data preprocessing method is used to improve the image quality of C-arm X-ray images and DR images. The YOLOv3 model is then used to detect the vertebrae, and the vertebral regions are extracted based on their position. In the second part, the Mobile-Unet model is first used to segment the vertebrae contour of the C-arm X-ray image and DR image based on vertebral regions respectively. The inclination angle of the contour is then calculated using the minimum bounding rectangle and corrected accordingly. Finally, a multi-vertebra strategy is applied to measure the visual information fidelity for the vertebral region, and the vertebrae are matched based on the measured results. We use 382 C-arm X-ray images and 203 full length X-ray images to train the vertebra detection model, and achieve a mAP of 0.87 in the test dataset of 31 C-arm X-ray images and 0.96 in the test dataset of 31 lumbar DR images. Finally, we achieve a vertebral segment matching accuracy of 0.733 on 31 C-arm X-ray images. A VDVM framework is proposed, which performs well for the detection of vertebrae and achieves good results in vertebral segment matching.

Automatic measurement of the Cobb angle for adolescent idiopathic scoliosis using convolutional neural network

This study proposes a convolutional neural network method for automatic vertebrae detection and Cobb angle (CA) measurement on X-ray images for scoliosis. 1021 full-length X-ray images of the whole spine of patients with adolescent idiopathic scoliosis (AIS) were used for training and segmentation. The proposed AI algorithm's results were compared with those of the manual method by six doctors using the intraclass correlation coefficient (ICC). The ICCs recorded by six doctors and AI were excellent or good, with a value of 0.973 for the major curve in the standing position. The mean error between AI and doctors was not affected by the angle size, with AI tending to measure 1.7°–2.2° smaller than that measured by the doctors. The proposed method showed a high correlation with the doctors’ measurements, regardless of the CA size, doctors’ experience, and patient posture. The proposed method showed excellent reliability, indicating that it is a promising automated method for measuring CA in patients with AIS.

Sequence based local-global information fusion framework for vertebrae detection under pathological and FOV variation challenges.

Automated vertebrae detection (identification and localization) aims to identify vertebrae and locate their centroids in medical images, which is a critical step of spinal computer-aided systems. However, due to unpredictable field-of-view and various pathology cases, the image content is diverse and the vertebral morphology can be abnormal in a variety of ways, which challenges the designed systems. In this paper, we propose an effective sequence-based framework robust to various tough cases for accurate vertebrae identification and localization. Our method consists of three sub-modules: (1) Local Feature Extraction (LFE) module designs a shape-compatible category-balanced sampler to collect patches to train a convolution neural network, which extracts representative local features and generates score maps. (2) Discriminative Sequential Image Description (DSID) module proposes a node screening strategy for reliable vertebral feature sequence construction based on feature maps and score maps. This effectively prevents false positives and false negatives in light-weighted dense prediction schemes and fuses local features into a hierarchical discriminative description of given images. (3) Spinal Pattern Exploitation (SPE) module designs an end-balanced relative position learning scheme to fuse hierarchical local-global information for comprehensively exploiting spinal patterns to overcome the FOV and pathological variation challenges in vertebrae detection. Extensive experiments on a challenging dataset consisting of 450 spinal MRIs show that the identification rate of FSDF reaches 0.974 ±0.025 and the localization error is only 4.742 ±2.928 pixels, which demonstrates the effectiveness of our method with pathological and field-of-view variations and its superiority over other state-of-the-art methods.

BUU-LSPINE: A Thai Open Lumbar Spine Dataset for Spondylolisthesis Detection

(1) Background: Spondylolisthesis, a common disease among older individuals, involves the displacement of vertebrae. The condition may gradually manifest with age, allowing for potential prevention by the research of predictive algorithms. However, one key issue that hinders research in spondylolisthesis prediction algorithms is the need for publicly available spondylolisthesis datasets. (2) Purpose: This paper introduces BUU-LSPINE, a new dataset for the lumbar spine. It includes 3600 patients’ plain film images annotated with vertebral position, spondylolisthesis diagnosis, and lumbosacral transitional vertebrae (LSTV) ground truth. (4) Methods: We established an annotation pipeline to create the BUU-SPINE dataset and evaluated it in three experiments as follows: (1) lumbar vertebrae detection, (2) vertebral corner points extraction, and (3) spondylolisthesis prediction. (5) Results: Lumbar vertebrae detection achieved the highest precision rates of 81.93% on the AP view and 83.45% on the LA view using YOLOv5; vertebral corner point extraction achieved the lowest average error distance of 4.63 mm on the AP view using ResNet152V2 and 4.91 mm on the LA view using DenseNet201. Spondylolisthesis prediction reached the highest accuracy of 95.14% on the AP view and 92.26% on the LA view of a testing set using Support Vector Machine (SVM). (6) Discussions: The results of the three experiments highlight the potential of BUU-LSPINE in developing and evaluating algorithms for lumbar vertebrae detection and spondylolisthesis prediction. These steps are crucial in advancing the creation of a clinical decision support system (CDSS). Additionally, the findings demonstrate the impact of Lumbosacral transitional vertebrae (LSTV) conditions on lumbar detection algorithms.

Integrating biplane information and context for spine landmark detection.

Spine landmark detection is of great significance for spinal morphological parameter assessment and three-dimensional reconstruction of the human spine. This detection task generally involves locating spine landmarks in the anterior-posterior (AP) and lateral (LAT) X-rays of the spine. Recently, the two-stage methods for AP spine landmark detection achieve better performance. However, these methods perform poorly in LAT landmark detection because of poor detection accuracy of LAT vertebra due to occlusion. To solve this problem, this paper proposes a new two-stage spine landmark detection method. In the first stage, this paper propose a biplane vertebra detection network for vertebra detection on AP and X-rays simultaneously. Then an epipolar module and a context enhancement module are proposed to assist LAT vertebra detection by using the biplane information and the context information of the vertebrae respectively. In the second stage, the landmarks can be obtained in the detected vertebrae area. Extensive experiment results conducted on a dataset containing 328 pairs of X-rays demonstrate that our method improves the vertebra and landmark detection accuracy.

3D zero-echo time and 3D T1-weighted gradient-echo MRI sequences as an alternative to CT for the evaluation of the lumbar facet joints and lumbosacral transitional vertebrae.

Computed tomography (CT) is the reference standard for assessment of the bone. Magnetic resonance imaging (MRI) developments enable a CT-like visualization of the osseous structures. To assess the diagnostic performance of 3D zero-echo time (3D-ZTE) and 3D T1-weighted gradient-echo (3D-T1GRE) MRI sequences for the evaluation of lumbar facet joints (LFJs) and the detection of lumbosacral transitional vertebrae (LSTV) using CT as the reference standard. In total, 87 adult patients were included in this prospective study. Evaluation of degenerative changes of the facet joints at the L3/L4, L4/L5, and L5/S1 levels on both sides was performed by two readers using a 4-point Likert scale. LSTV were classified according to Castelvi et al. Image quality was quantitatively measured using the signal-to-noise (SNR) and contrast-to-noise (CNR) ratios. Intra-reader, inter-reader, and inter-modality reliability were calculated using Cohen's kappa statistic. Intra-reader agreement for 3D-ZTE, 3D-T1GRE, and CT was 0.607, 0.751, and 0.856 and inter-reader agreement was 0.535, 0.563, and 0.599, respectively. The inter-modality agreement between 3D-ZTE and CT was 0.631 and between 3D-T1GRE and CT 0.665. A total of LSTV were identified in both MR sequences with overall comparable accuracy compared to CT. Mean SNR for bone, muscle, and fat was highest for 3D-T1GRE and mean CNR was highest for CT. 3D-ZTE and 3D-T1GRE MRI sequences can assess the LFJs and LSTV and may serve as potential alternatives to CT.

Artificial intelligence-aided lytic spinal bone metastasis classification on CT scans.

Spinal bone metastases directly affect quality of life, and patients with lytic-dominant lesions are at high risk for neurological symptoms and fractures. To detect and classify lytic spinal bone metastasis using routine computed tomography (CT) scans, we developed a deep learning (DL)-based computer-aided detection (CAD) system. We retrospectively analyzed 2125 diagnostic and radiotherapeutic CT images of 79 patients. Images annotated as tumor (positive) or not (negative) were randomized into training (1782 images) and test (343 images) datasets. YOLOv5m architecture was used to detect vertebra on whole CT scans. InceptionV3 architecture with the transfer-learning technique was used to classify the presence/absence of lytic lesions on CT images showing the presence of vertebra. The DL models were evaluated via fivefold cross-validation. For vertebra detection, bounding box accuracy was estimated using intersection over union (IoU). We evaluated the area under the curve (AUC) of a receiver operating characteristic curve to classify lesions. Moreover, we determined the accuracy, precision, recall, and F1 score. We used the gradient-weighted class activation mapping (Grad-CAM) technique for visual interpretation. The computation time was 0.44s per image. The average IoU value of the predicted vertebra was 0.923 ± 0.052 (0.684-1.000) for test datasets. In the binary classification task, the accuracy, precision, recall, F1-score, and AUC value for test datasets were 0.872, 0.948, 0.741, 0.832, and 0.941, respectively. Heat maps constructed using the Grad-CAM technique were consistent with the location of lytic lesions. Our artificial intelligence-aided CAD system using two DL models could rapidly identify vertebra bone from whole CT images and detect lytic spinal bone metastasis, although further evaluation of diagnostic accuracy is required with a larger sample size.

Vertebral Center Points Locating and Cobb Angle Measurement Based on Deep Learning

Traditional manual measurement of Cobb angle is a time-consuming process and leads to different results. To address this issue, this paper proposes a deep learning-based method of locating the vertebral center points. The whole X-ray can be input into the network for prediction, without worrying about the detection of cervical vertebrae with similar characteristics to the thoracic and lumbar vertebrae. First, key points predicting and noise points filtering operations are employed to obtain vertebral center points for fitting. Then, the spine curve is fitted, and the slope of the normal line of the spine curve is adjusted according to an empirical formula. Finally, the Cobb angle allowed by the error is calculated. Through the reliability analysis of the traditional manual measurement method and the automatic detection method in this paper, ICC (intraclass correlation coefficient) with the two observers was 0.897 and 0.901, respectively, and MAD (mean absolute deviation) was 3.13° and 3.04° respectively. This indicates that the automatic detecting method by computer has good reliability. Therefore, this method can be used to detect scoliosis quickly and effectively.

Interpretable vertebral fracture quantification via anchor-free landmarks localization.

Vertebral body compression fractures are early signs of osteoporosis. Though these fractures are visible on Computed Tomography (CT) images, they are frequently missed by radiologists in clinical settings. Prior research on automatic methods of vertebral fracture classification proves its reliable quality; however, existing methods provide hard-to-interpret outputs and sometimes fail to process cases with severe abnormalities such as highly pathological vertebrae or scoliosis. We propose a new two-step algorithm to localize the vertebral column in 3D CT images and then detect individual vertebrae and quantify fractures in 2D simultaneously. We train neural networks for both steps using a simple 6-keypoints based annotation scheme, which corresponds precisely to the current clinical recommendation. Our algorithm has no exclusion criteria, processes 3D CT in 2 seconds on a single GPU, and provides an interpretable and verifiable output. The method approaches expert-level performance and demonstrates state-of-the-art results in vertebrae 3D localization (the average error is 1mm), vertebrae 2D detection (precision and recall are 0.99), and fracture identification (ROC AUC at the patient level is up to 0.96). Our anchor-free vertebra detection network shows excellent generalizability on a new domain by achieving ROC AUC 0.95, sensitivity 0.85, specificity 0.9 on a challenging VerSe dataset with many unseen vertebra types.

Unsupervised Domain Adaptation for Vertebrae Detection and Identification in 3D CT Volumes Using a Domain Sanity Loss.

A variety of medical computer vision applications analyze 2D slices of computed tomography (CT) scans, whereas axial slices from the body trunk region are usually identified based on their relative position to the spine. A limitation of such systems is that either the correct slices must be extracted manually or labels of the vertebrae are required for each CT scan to develop an automated extraction system. In this paper, we propose an unsupervised domain adaptation (UDA) approach for vertebrae detection and identification based on a novel Domain Sanity Loss (DSL) function. With UDA the model’s knowledge learned on a publicly available (source) data set can be transferred to the target domain without using target labels, where the target domain is defined by the specific setup (CT modality, study protocols, applied pre- and processing) at the point of use (e.g., a specific clinic with its specific CT study protocols). With our approach, a model is trained on the source and target data set in parallel. The model optimizes a supervised loss for labeled samples from the source domain and the DSL loss function based on domain-specific “sanity checks” for samples from the unlabeled target domain. Without using labels from the target domain, we are able to identify vertebra centroids with an accuracy of %. By adding only ten target labels during training the accuracy increases to %, which is on par with the current state-of-the-art for full supervised learning, while using about 20 times less labels. Thus, our model can be used to extract 2D slices from 3D CT scans on arbitrary data sets fully automatically without requiring an extensive labeling effort, contributing to the clinical adoption of medical imaging by hospitals.

Reliability of coeliac and superior mesenteric artery origin level in lumbosacral transitional vertebrae detection and vertebral numbering

Objective: To evaluate the value of coeliac artery (CA) and superior mesenteric artery (SMA) origin levels as an anatomic marker in the detection of lumbosacral transitional vertebrae (LSTV) and vertebral enumeration. Patients and Methods: Institutional review board approval was obtained. Routine lumbar magnetic resonance (MR) imaging that included sagittal cervicothoracic scout images in 972 cases were evaluated retrospectively. Six segments were created in the vertebral column with 7 lines. CA and SMA ostiums were localized in these segments. Results: Coeliac artery and SMA levels were detected more caudally in lumbarized S1 and more cranially in sacralized L5 cases compared to non-LSTV cases. Conclusion: Coeliac artery and SMA origin levels are not confidential in consequence of wide variability.

MPF-net: An effective framework for automated cobb angle estimation.

In clinical practice, the Cobb angle is the gold standard for idiopathic scoliosis assessment, which can provide an important reference for clinicians to make surgical plan and give medical care to patients. Currently, the Cobb angle is measured manually on both anterior-posterior(AP) view X-rays and lateral(LAT) view X-rays. The clinicians first find four landmarks on each vertebra, and then they extend the line from landmarks and measure the Cobb angle by rules. The whole process is time-consuming and subjective, so that the automated Cobb angle estimation is required for efficient and reliable Cobb angle measurement. The noise in X-rays and the occlusion of vertebras are the main difficulties for automated Cobb angle estimation, and it is challenging to utilize the information between the multi-view X-rays of the same patient. Addressing these problems, in this paper, we propose an effective framework named MPF-net by using deep learning methods for automated Cobb angle estimation. We combine a vertebra detection branch and a landmark prediction branch based on the backbone convolutional neural network, which can provide the bounded area for landmark prediction. Then we propose a proposal correlation module to utilize the information between neighbor vertebras, so that we can find the vertebras hidden by ribcage and arms on LAT X-rays. We also design a feature fusion module to utilize the information in both AP and LAT X-rays for better performance. The experiment results on 2738 pair of X-rays show that our proposed MPF-net achieves precise vertebra detection and landmark prediction performance, and we get impressive 3.52 and 4.05 circular mean absolute errors on AP and LAT X-rays respectively, which is much better than previous methods. Therefore, we can provide clinicians with automated, efficient and reliable Cobb angle measurement.

Spine-transformers: Vertebra labeling and segmentation in arbitrary field-of-view spine CTs via 3D transformers.

In this paper, we address the problem of fully automatic labeling and segmentation of 3D vertebrae in arbitrary Field-Of-View (FOV) CT images. We propose a deep learning-based two-stage solution to tackle these two problems. More specifically, in the first stage, the challenging vertebra labeling problem is solved via a novel transformers-based 3D object detector that views automatic detection of vertebrae in arbitrary FOV CT scans as a one-to-one set prediction problem. The main components of the new method, called Spine-Transformers, are a one-to-one set based global loss that forces unique predictions and a light-weighted 3D transformer architecture equipped with a skip connection and learnable positional embeddings for encoder and decoder, respectively. We additionally propose an inscribed sphere-based object detector to replace the regular box-based object detector for a better handling of volume orientation variations. Our method reasons about the relationships of different levels of vertebrae and the global volume context to directly infer all vertebrae in parallel. In the second stage, the segmentation of the identified vertebrae and the refinement of the detected centers are then done by training one single multi-task encoder-decoder network for all vertebrae as the network does not need to identify which vertebra it is working on. The two tasks share a common encoder path but with different decoder paths. Comprehensive experiments are conducted on two public datasets and one in-house dataset. The experimental results demonstrate the efficacy of the present approach.

3D vertebrae labeling in spine CT: an accurate, memory-efficient (Ortho2D) framework

Purpose. Accurate localization and labeling of vertebrae in computed tomography (CT) is an important step toward more quantitative, automated diagnostic analysis and surgical planning. In this paper, we present a framework (called Ortho2D) for vertebral labeling in CT in a manner that is accurate and memory-efficient. Methods. Ortho2D uses two independent faster R-convolutional neural network networks to detect and classify vertebrae in orthogonal (sagittal and coronal) CT slices. The 2D detections are clustered in 3D to localize vertebrae centroids in the volumetric CT and classify the region (cervical, thoracic, lumbar, or sacral) and vertebral level. A post-process sorting method incorporates the confidence in network output to refine classifications and reduce outliers. Ortho2D was evaluated on a publicly available dataset containing 302 normal and pathological spine CT images with and without surgical instrumentation. Labeling accuracy and memory requirements were assessed in comparison to other recently reported methods. The memory efficiency of Ortho2D permitted extension to high-resolution CT to investigate the potential for further boosts to labeling performance. Results. Ortho2D achieved overall vertebrae detection accuracy of 97.1%, region identification accuracy of 94.3%, and individual vertebral level identification accuracy of 91.0%. The framework achieved 95.8% and 83.6% level identification accuracy in images without and with surgical instrumentation, respectively. Ortho2D met or exceeded the performance of previously reported 2D and 3D labeling methods and reduced memory consumption by a factor of ∼50 (at 1 mm voxel size) compared to a 3D U-Net, allowing extension to higher resolution datasets than normally afforded. The accuracy of level identification increased from 80.1% (for standard/low resolution CT) to 95.1% (for high-resolution CT). Conclusions. The Ortho2D method achieved vertebrae labeling performance that is comparable to other recently reported methods with significant reduction in memory consumption, permitting further performance boosts via application to high-resolution CT.