Multiple Sclerosis Lesion Segmentation with Tiramisu and 2.5D Stacked Slices

Assessing robustness and generalization of a deep neural network for brain MS lesion segmentation on real-world data

OBJECTIVES

Evaluate the performance of a deep learning (DL)-based model for multiple sclerosis (MS) lesion segmentation and compare it to other DL and non-DL algorithms.

METHODS

This ambispective, multicenter study assessed the performance of a DL-based model for MS lesion segmentation and compared it to alternative DL- and non-DL-based methods. Models were tested on internal (n = 20) and external (n = 18) datasets from Latin America, and on an external dataset from Europe (n = 49). We also examined robustness by rescanning six patients (n = 6) from our MS clinical cohort. Moreover, we studied inter-human annotator agreement and discussed our findings in light of these results. Performance and robustness were assessed using intraclass correlation coefficient (ICC), Dice coefficient (DC), and coefficient of variation (CV).

RESULTS

Inter-human ICC ranged from 0.89 to 0.95, while spatial agreement among annotators showed a median DC of 0.63. Using expert manual segmentations as ground truth, our DL model achieved a median DC of 0.73 on the internal, 0.66 on the external, and 0.70 on the challenge datasets. The performance of our DL model exceeded that of the alternative algorithms on all datasets. In the robustness experiment, our DL model also achieved higher DC (ranging from 0.82 to 0.90) and lower CV (ranging from 0.7 to 7.9%) when compared to the alternative methods.

CONCLUSION

Our DL-based model outperformed alternative methods for brain MS lesion segmentation. The model also proved to generalize well on unseen data and has a robust performance and low processing times both on real-world and challenge-based data.

CLINICAL RELEVANCE STATEMENT

Our DL-based model demonstrated superior performance in accurately segmenting brain MS lesions compared to alternative methods, indicating its potential for clinical application with improved accuracy, robustness, and efficiency.

KEY POINTS

• Automated lesion load quantification in MS patients is valuable; however, more accurate methods are still necessary. • A novel deep learning model outperformed alternative MS lesion segmentation methods on multisite datasets. • Deep learning models are particularly suitable for MS lesion segmentation in clinical scenarios.

2.5D transfer deep learning model for segmentation of contrast-enhancing lesions on brain magnetic resonance imaging of multiple sclerosis and neuromyelitis optica spectrum disorder

Background

Multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) are the two mimic autoimmune diseases of the central nervous system, which are rare in East Asia. Quantitative detection of contrast-enhancing lesions (CELs) on contrast-enhancing T1-weighted magnetic resonance (MR) images is of great significance for assessing the disease activity of MS and NMOSD. However, it is challenging to develop automatic segmentation algorithms due to the lack of data. In this work, we present an automatic segmentation model of CELs based on Fully Convolutional with Attention DenseNet (FCA-DenseNet) and transfer learning strategy to address the challenge of CEL quantification in small-scale datasets.

Methods

A transfer learning approach was employed in this study, whereby pretraining was conducted using 77 MS subjects from the open access datasets (MICCAI 2016, MICCAI 2017, ISBI 2015) for white matter hyperintensity segmentation, followed by fine-tuning using 24 MS and NMOSD subjects from the local dataset for CEL segmentation. The proposed FCA-DenseNet combined the Fully Convolutional DenseNet and Convolutional Block Attention Module in order to improve the learning capability. A 2.5D data slicing strategy was used to process complex 3D MR images. U-Net, ResUNet, TransUNet, and Attention-UNet are used as comparison models to FCA-DenseNet. Dice similarity coefficient (DSC), positive predictive value (PPV), true positive rate (TPR), and volume difference (VD) are used as evaluation metrics to evaluate the performances of different models.

Results

FCA-DenseNet outperforms all other models in terms of all evaluation metrics, with a DSC of 0.661±0.187, PPV of 0.719±0.201, TPR of 0.680±0.254, and VD of 0.388±0.334. Transfer learning strategy has achieved success in building segmentation models on a small-scale local dataset where traditional deep learning approaches fail to train effectively.

Conclusions

The improved FCA-DenseNet, combined with transfer learning strategy and 2.5D data slicing strategy, has successfully addressed the challenges in constructing deep learning models on small-scale datasets, making it conducive to clinical quantification of brain CELs and diagnosis of MS and NMOSD.

How far MS lesion detection and segmentation are integrated into the clinical workflow? A systematic review

INTRODUCTION

Over the past few years, the deep learning community has developed and validated a plethora of tools for lesion detection and segmentation in Multiple Sclerosis (MS). However, there is an important gap between validating models technically and clinically. To this end, a six-step framework necessary for the development, validation, and integration of quantitative tools in the clinic was recently proposed under the name of the Quantitative Neuroradiology Initiative (QNI).

AIMS

Investigate to what extent automatic tools in MS fulfill the QNI framework necessary to integrate automated detection and segmentation into the clinical neuroradiology workflow.

METHODS

Adopting the systematic Cochrane literature review methodology, we screened and summarised published scientific articles that perform automatic MS lesions detection and segmentation. We categorised the retrieved studies based on their degree of fulfillment of QNI's six-steps, which include a tool's technical assessment, clinical validation, and integration.

RESULTS

We found 156 studies; 146/156 (94%) fullfilled the first QNI step, 155/156 (99%) the second, 8/156 (5%) the third, 3/156 (2%) the fourth, 5/156 (3%) the fifth and only one the sixth.

CONCLUSIONS

To date, little has been done to evaluate the clinical performance and the integration in the clinical workflow of available methods for MS lesion detection/segmentation. In addition, the socio-economic effects and the impact on patients' management of such tools remain almost unexplored.

Multi-arm U-Net with dense input and skip connectivity for T2 lesion segmentation in clinical trials of multiple sclerosis

T2 lesion quantification plays a crucial role in monitoring disease progression and evaluating treatment response in multiple sclerosis (MS). We developed a 3D, multi-arm U-Net for T2 lesion segmentation, which was trained on a large, multicenter clinical trial dataset of relapsing MS. We investigated its generalization to other relapsing and primary progressive MS clinical trial datasets, and to an external dataset from the MICCAI 2016 MS lesion segmentation challenge. Additionally, we assessed the model's ability to reproduce the separation of T2 lesion volumes between treatment and control arms; and the association of baseline T2 lesion volumes with clinical disability scores compared with manual lesion annotations. The trained model achieved a mean dice coefficient of ≥ 0.66 and a lesion detection sensitivity of ≥ 0.72 across the internal test datasets. On the external test dataset, the model achieved a mean dice coefficient of 0.62, which is comparable to 0.59 from the best model in the challenge, and a lesion detection sensitivity of 0.68. Lesion detection performance was reduced for smaller lesions (≤ 30 μL, 3-10 voxels). The model successfully maintained the separation of the longitudinal changes in T2 lesion volumes between the treatment and control arms. Such tools could facilitate semi-automated MS lesion quantification; and reduce rater burden in clinical trials.

HC-Net: A hybrid convolutional network for non-human primate brain extraction

Brain extraction (skull stripping) is an essential step in the magnetic resonance imaging (MRI) analysis of brain sciences. However, most of the current brain extraction methods that achieve satisfactory results for human brains are often challenged by non-human primate brains. Due to the small sample characteristics and the nature of thick-slice scanning of macaque MRI data, traditional deep convolutional neural networks (DCNNs) are unable to obtain excellent results. To overcome this challenge, this study proposed a symmetrical end-to-end trainable hybrid convolutional neural network (HC-Net). It makes full use of the spatial information between adjacent slices of the MRI image sequence and combines three consecutive slices from three axes for 3D convolutions, which reduces the calculation consumption and promotes accuracy. The HC-Net consists of encoding and decoding structures of 3D convolutions and 2D convolutions in series. The effective use of 2D convolutions and 3D convolutions relieves the underfitting of 2D convolutions to spatial features and the overfitting of 3D convolutions to small samples. After evaluating macaque brain data from different sites, the results showed that HC-Net performed better in inference time (approximately 13 s per volume) and accuracy (mean Dice coefficient reached 95.46%). The HC-Net model also had good generalization ability and stability in different modes of brain extraction tasks.

CAT-Net: A Cross-Slice Attention Transformer Model for Prostate Zonal Segmentation in MRI

Prostate cancer is the second leading cause of cancer death among men in the United States. The diagnosis of prostate MRI often relies on accurate prostate zonal segmentation. However, state-of-the-art automatic segmentation methods often fail to produce well-contained volumetric segmentation of the prostate zones since certain slices of prostate MRI, such as base and apex slices, are harder to segment than other slices. This difficulty can be overcome by leveraging important multi-scale image-based information from adjacent slices, but current methods do not fully learn and exploit such cross-slice information. In this paper, we propose a novel cross-slice attention mechanism, which we use in a Transformer module to systematically learn cross-slice information at multiple scales. The module can be utilized in any existing deep-learning-based segmentation framework with skip connections. Experiments show that our cross-slice attention is able to capture cross-slice information significant for prostate zonal segmentation in order to improve the performance of current state-of-the-art methods. Cross-slice attention improves segmentation accuracy in the peripheral zones, such that segmentation results are consistent across all the prostate slices (apex, mid-gland, and base). The code for the proposed model is available at https://bit.ly/CAT-Net.

Improving the detection of new lesions in multiple sclerosis with a cascaded 3D fully convolutional neural network approach

Longitudinal magnetic resonance imaging (MRI) has an important role in multiple sclerosis (MS) diagnosis and follow-up. Specifically, the presence of new lesions on brain MRI scans is considered a robust predictive biomarker for the disease progression. New lesions are a high-impact prognostic factor to predict evolution to MS or risk of disability accumulation over time. However, the detection of this disease activity is performed visually by comparing the follow-up and baseline scans. Due to the presence of small lesions, misregistration, and high inter-/intra-observer variability, this detection of new lesions is prone to errors. In this direction, one of the last Medical Image Computing and Computer Assisted Intervention (MICCAI) challenges was dealing with this automatic new lesion quantification. The MSSEG-2: MS new lesions segmentation challenge offers an evaluation framework for this new lesion segmentation task with a large database (100 patients, each with two-time points) compiled from the OFSEP (Observatoire français de la sclérose en plaques) cohort, the French MS registry, including 3D T2-w fluid-attenuated inversion recovery (T2-FLAIR) images from different centers and scanners. Apart from a change in centers, MRI scanners, and acquisition protocols, there are more challenges that hinder the automated detection process of new lesions such as the need for large annotated datasets, which may be not easily available, or the fact that new lesions are small areas producing a class imbalance problem that could bias trained models toward the non-lesion class. In this article, we present a novel automated method for new lesion detection of MS patient images. Our approach is based on a cascade of two 3D patch-wise fully convolutional neural networks (FCNNs). The first FCNN is trained to be more sensitive revealing possible candidate new lesion voxels, while the second FCNN is trained to reduce the number of misclassified voxels coming from the first network. 3D T2-FLAIR images from the two-time points were pre-processed and linearly co-registered. Afterward, a fully CNN, where its inputs were only the baseline and follow-up images, was trained to detect new MS lesions. Our approach obtained a mean segmentation dice similarity coefficient of 0.42 with a detection F1-score of 0.5. Compared to the challenge participants, we obtained one of the highest precision scores (PPVL = 0.52), the best PPVL rate (0.53), and a lesion detection sensitivity (SensL of 0.53).

New lesion segmentation for multiple sclerosis brain images with imaging and lesion-aware augmentation

Multiple sclerosis (MS) is an inflammatory and demyelinating neurological disease of the central nervous system. Image-based biomarkers, such as lesions defined on magnetic resonance imaging (MRI), play an important role in MS diagnosis and patient monitoring. The detection of newly formed lesions provides crucial information for assessing disease progression and treatment outcome. Here, we propose a deep learning-based pipeline for new MS lesion detection and segmentation, which is built upon the nnU-Net framework. In addition to conventional data augmentation, we employ imaging and lesion-aware data augmentation methods, axial subsampling and CarveMix, to generate diverse samples and improve segmentation performance. The proposed pipeline is evaluated on the MICCAI 2021 MS new lesion segmentation challenge (MSSEG-2) dataset. It achieves an average Dice score of 0.510 and F 1 score of 0.552 on cases with new lesions, and an average false positive lesion number n FP of 0.036 and false positive lesion volume V FP of 0.192 mm 3 on cases with no new lesions. Our method outperforms other participating methods in the challenge and several state-of-the-art network architectures.

Recent advances in the longitudinal segmentation of multiple sclerosis lesions on magnetic resonance imaging: a review

Multiple sclerosis (MS) is a chronic autoimmune disease characterized by demyelinating lesions that are often visible on magnetic resonance imaging (MRI). Segmentation of these lesions can provide imaging biomarkers of disease burden that can help monitor disease progression and the imaging response to treatment. Manual delineation of MRI lesions is tedious and prone to subjective bias, while automated lesion segmentation methods offer objectivity and speed, the latter being particularly important when analysing large datasets. Lesion segmentation can be broadly categorised into two groups: cross-sectional methods, which use imaging data acquired at a single time-point to characterise MRI lesions; and longitudinal methods, which use imaging data from the same subject acquired at two or more different time-points to characterise lesions over time. The main objective of longitudinal segmentation approaches is to more accurately detect the presence of new MS lesions and the growth or remission of existing lesions, which may be effective biomarkers of disease progression and treatment response. This paper reviews articles on longitudinal MS lesion segmentation methods published over the past 10 years. These are divided into traditional machine learning methods and deep learning techniques. PubMed articles using longitudinal information and comparing fully automatic two time point segmentations in any step of the process were selected. Nineteen articles were reviewed. There is an increasing number of deep learning techniques for longitudinal MS lesion segmentation that are promising to help better understand disease progression.

Robust Multiple Sclerosis Lesion Inpainting with Edge Prior

Inpainting lesions is an important preprocessing task for algorithms analyzing brain MRIs of multiple sclerosis (MS) patients, such as tissue segmentation and cortical surface reconstruction. We propose a new deep learning approach for this task. Unlike existing inpainting approaches which ignore the lesion areas of the input image, we leverage the edge information around the lesions as a prior to help the inpainting process. Thus, the input of this network includes the T1-w image, lesion mask and the edge map computed from the T1-w image, and the output is the lesion-free image. The introduction of the edge prior is based on our observation that the edge detection results of the MRI scans will usually contain the contour of white matter (WM) and grey matter (GM), even though some undesired edges appear near the lesions. Instead of losing all the information around the neighborhood of lesions, our approach preserves the local tissue shape (brain/WM/GM) with the guidance of the input edges. The qualitative results show that our pipeline inpaints the lesion areas in a realistic and shape-consistent way. Our quantitative evaluation shows that our approach outperforms the existing state-of-the-art inpainting methods in both image-based metrics and in FreeSurfer segmentation accuracy. Furthermore, our approach demonstrates robustness to inaccurate lesion mask inputs. This is important for practical usability, because it allows for a generous over-segmentation of lesions instead of requiring precise boundaries, while still yielding accurate results.

Evaluating White Matter Lesion Segmentations with Refined Sørensen-Dice Analysis

The Sørensen-Dice index (SDI) is a widely used measure for evaluating medical image segmentation algorithms. It offers a standardized measure of segmentation accuracy which has proven useful. However, it offers diminishing insight when the number of objects is unknown, such as in white matter lesion segmentation of multiple sclerosis (MS) patients. We present a refinement for finer grained parsing of SDI results in situations where the number of objects is unknown. We explore these ideas with two case studies showing what can be learned from our two presented studies. Our first study explores an inter-rater comparison, showing that smaller lesions cannot be reliably identified. In our second case study, we demonstrate fusing multiple MS lesion segmentation algorithms based on the insights into the algorithms provided by our analysis to generate a segmentation that exhibits improved performance. This work demonstrates the wealth of information that can be learned from refined analysis of medical image segmentations.