Automatic semantic segmentation of the lumbar spine: Clinical applicability in a multi-parametric and multi-center study on magnetic resonance images

SPINEPS-automatic whole spine segmentation of T2-weighted MR images using a two-phase approach to multi-class semantic and instance segmentation

OBJECTIVES

Introducing SPINEPS, a deep learning method for semantic and instance segmentation of 14 spinal structures (ten vertebra substructures, intervertebral discs, spinal cord, spinal canal, and sacrum) in whole-body sagittal T2-weighted turbo spin echo images.

MATERIAL AND METHODS

This local ethics committee-approved study utilized a public dataset (train/test 179/39 subjects, 137 female), a German National Cohort (NAKO) subset (train/test 1412/65 subjects, mean age 53, 694 female), and an in-house dataset (test 10 subjects, mean age 70, 5 female). SPINEPS is a semantic segmentation model, followed by a sliding window approach utilizing a second model to create instance masks from the semantic ones. Segmentation evaluation metrics included the Dice score and average symmetrical surface distance (ASSD). Statistical significance was assessed using the Wilcoxon signed-rank test.

RESULTS

On the public dataset, SPINEPS outperformed a nnUNet baseline on every structure and metric (e.g., an average over vertebra instances: dice 0.933 vs 0.911, p < 0.001, ASSD 0.21 vs 0.435, p < 0.001). SPINEPS trained on automated annotations of the NAKO achieves an average global Dice score of 0.918 on the combined NAKO and in-house test split. Adding the training data from the public dataset outperforms this (average instance-wise Dice score over the vertebra substructures 0.803 vs 0.778, average global Dice score 0.931 vs 0.918).

CONCLUSION

SPINEPS offers segmentation of 14 spinal structures in T2w sagittal images. It provides a semantic mask and an instance mask separating the vertebrae and intervertebral discs. This is the first publicly available algorithm to enable this segmentation.

KEY POINTS

Question No publicly available automatic approach can yield semantic and instance segmentation masks for the whole spine (including posterior elements) in T2-weighted sagittal TSE images. Findings Segmenting semantically first and then instance-wise outperforms a baseline trained directly on instance segmentation. The developed model produces high-resolution MRI segmentations for the whole spine. Clinical relevance This study introduces an automatic approach to whole spine segmentation, including posterior elements, in arbitrary fields of view T2w sagittal MR images, enabling easy biomarker extraction, automatic localization of pathologies and degenerative diseases, and quantifying analyses as downstream research.

Neural network segmentation of disc volume from magnetic resonance images and the effect of degeneration and spinal level

Background

Magnetic resonance imaging (MRI) noninvasively quantifies disc structure but requires segmentation that is both time intensive and susceptible to human error. Recent advances in neural networks can improve on manual segmentation. The aim of this study was to establish a method for automatic slice-wise segmentation of 3D disc volumes from subjects with a wide range of age and degrees of disc degeneration. A U-Net convolutional neural network was trained to segment 3D T1-weighted spine MRI.

Methods

Lumbar spine MRIs were acquired from 43 subjects (23-83 years old) and manually segmented. A U-Net architecture was trained using the TensorFlow framework. Two rounds of model tuning were performed. The performance of the model was measured using a validation set that did not cross over from the training set. The model version with the best Dice similarity coefficient (DSC) was selected in each tuning round. After model development was complete and a final U-Net model was selected, performance of this model was compared between disc levels and degeneration grades.

Results

Performance of the final model was equivalent to manual segmentation, with a mean DSC = 0.935 ± 0.014 for degeneration grades I-IV. Neither the manual segmentation nor the U-Net model performed as well for grade V disc segmentation. Compared with the baseline model at the beginning of round 1, the best model had fewer filters/parameters (75%), was trained using only slices with at least one disc-labeled pixel, applied contrast stretching to its input images, and used a greater dropout rate.

Conclusion

This study successfully trained a U-Net model for automatic slice-wise segmentation of 3D disc volumes from populations with a wide range of ages and disc degeneration. The final trained model is available to support scientific use.

Deep learning assisted segmentation of the lumbar intervertebral disc: a systematic review and meta-analysis

BACKGROUND

In recent years, deep learning (DL) technology has been increasingly used for the diagnosis and treatment of lumbar intervertebral disc (IVD) degeneration. This study aims to evaluate the performance of DL technology for IVD segmentation in magnetic resonance (MR) images and explore improvement strategies.

METHODS

We developed a PRISMA systematic review protocol and systematically reviewed studies that used DL algorithm frameworks to perform IVD segmentation based on MR images published up to April 10, 2024. The Quality Assessment of Diagnostic Accuracy Studies-2 tool was used to assess methodological quality, and the pooled dice similarity coefficient (DSC) score and Intersection over Union (IoU) were calculated to evaluate segmentation performance.

RESULTS

45 studies were included in this systematic review, of which 16 provided complete segmentation performance data and were included in the quantitative meta-analysis. The results indicated that DL models showed satisfactory IVD segmentation performance, with a pooled DSC of 0.900 (95% confidence interval [CI]: 0.887-0.914) and IoU of 0.863 (95% CI: 0.730-0.995). However, the subgroup analysis did not show significant effects of factors on IVD segmentation performance, including network dimensionality, algorithm type, publication year, number of patients, scanning direction, data augmentation, and cross-validation.

CONCLUSIONS

This study highlights the potential of DL technology in IVD segmentation and its further applications. However, due to the heterogeneity in algorithm frameworks and result reporting of the included studies, the conclusions should be interpreted with caution. Future research should focus on training generalized models on large-scale datasets to enhance their clinical application.

Post-secondary classroom teaching quality evaluation using small object detection model

The classroom video has a complex background and dense targets. This study utilizes small object detection technology to analyze and evaluate students' behavior in the classroom, aiming to objectively and accurately assess classroom quality. Firstly, noise is removed from the images using a median filter, and the contrast of the images is enhanced through histogram equalization. Label smoothing is applied to reduce the model's sensitivity to labels. Then, features are extracted from the preprocessed images, and multi-scale feature fusion is employed to enhance semantic expression across multiple scales. Finally, a combination loss function is utilized to improve the accuracy of multi-object recognition tasks. Real-time detection of students' behaviors in the classroom is performed based on the small object detection model. The average head-up rate in the classroom is calculated, and the quality of teaching is evaluated and analyzed. This study explores the methods and applications of small object detection technology based on actual teaching cases and analyzes and evaluates its effectiveness in evaluating the quality of higher education classroom teaching. The research findings demonstrate the significant importance of small object detection technology in effectively evaluating students' learning conditions in higher education classrooms, leading to improved teaching quality and personalized education.

Data pyramid structure for optimizing EUS-based GISTs diagnosis in multi-center analysis with missing label

This study introduces the Data Pyramid Structure (DPS) to address data sparsity and missing labels in medical image analysis. The DPS optimizes multi-task learning and enables sustainable expansion of multi-center data analysis. Specifically, It facilitates attribute prediction and malignant tumor diagnosis tasks by implementing a segmentation and aggregation strategy on data with absent attribute labels. To leverage multi-center data, we propose the Unified Ensemble Learning Framework (UELF) and the Unified Federated Learning Framework (UFLF), which incorporate strategies for data transfer and incremental learning in scenarios with missing labels. The proposed method was evaluated on a challenging EUS patient dataset from five centers, achieving promising diagnostic performance. The average accuracy was 0.984 with an AUC of 0.927 for multi-center analysis, surpassing state-of-the-art approaches. The interpretability of the predictions further highlights the potential clinical relevance of our method.

Facial wrinkle segmentation using weighted deep supervision and semi-automatic labeling

Facial wrinkles are important indicators of human aging. Recently, a method using deep learning and a semi-automatic labeling was proposed to segment facial wrinkles, which showed much better performance than conventional image-processing-based methods. However, the difficulty of wrinkle segmentation remains challenging due to the thinness of wrinkles and their small proportion in the entire image. Therefore, performance improvement in wrinkle segmentation is still necessary. To address this issue, we propose a novel loss function that takes into account the thickness of wrinkles based on the semi-automatic labeling approach. First, considering the different spatial dimensions of the decoder in the U-Net architecture, we generated weighted wrinkle maps from ground truth. These weighted wrinkle maps were used to calculate the training losses more accurately than the existing deep supervision approach. This new loss computation approach is defined as weighted deep supervision in our study. The proposed method was evaluated using an image dataset obtained from a professional skin analysis device and labeled using semi-automatic labeling. In our experiment, the proposed weighted deep supervision showed higher Jaccard Similarity Index (JSI) performance for wrinkle segmentation compared to conventional deep supervision and traditional image processing methods. Additionally, we conducted experiments on the labeling using a semi-automatic labeling approach, which had not been explored in previous research, and compared it with human labeling. The semi-automatic labeling technology showed more consistent wrinkle labels than human-made labels. Furthermore, to assess the scalability of the proposed method to other domains, we applied it to retinal vessel segmentation. The results demonstrated superior performance of the proposed method compared to existing retinal vessel segmentation approaches. In conclusion, the proposed method offers high performance and can be easily applied to various biomedical domains and U-Net-based architectures. Therefore, the proposed approach will be beneficial for various biomedical imaging approaches. To facilitate this, we have made the source code of the proposed method publicly available at: https://github.com/resemin/WeightedDeepSupervision.