ORCHESTRAL FULLY CONVOLUTIONAL NETWORKS FOR SMALL LESION SEGMENTATION IN BRAIN MRI

3-D RoI-Aware U-Net for Accurate and Efficient Colorectal Tumor Segmentation

Segmentation of colorectal cancerous regions from 3-D magnetic resonance (MR) images is a crucial procedure for radiotherapy. Automatic delineation from 3-D whole volumes is in urgent demand yet very challenging. Drawbacks of existing deep-learning-based methods for this task are two-fold: 1) extensive graphics processing unit (GPU) memory footprint of 3-D tensor limits the trainable volume size, shrinks effective receptive field, and therefore, degrades speed and segmentation performance and 2) in-region segmentation methods supported by region-of-interest (RoI) detection are either blind to global contexts, detail richness compromising, or too expensive for 3-D tasks. To tackle these drawbacks, we propose a novel encoder-decoder-based framework for 3-D whole volume segmentation, referred to as 3-D RoI-aware U-Net (3-D RU-Net). 3-D RU-Net fully utilizes the global contexts covering large effective receptive fields. Specifically, the proposed model consists of a global image encoder for global understanding-based RoI localization, and a local region decoder that operates on pyramid-shaped in-region global features, which is GPU memory efficient and thereby enables training and prediction with large 3-D whole volumes. To facilitate the global-to-local learning procedure and enhance contour detail richness, we designed a dice-based multitask hybrid loss function. The efficiency of the proposed framework enables an extensive model ensemble for further performance gain at acceptable extra computational costs. Over a dataset of 64 T2-weighted MR images, the experimental results of four-fold cross-validation show that our method achieved 75.5% dice similarity coefficient (DSC) in 0.61 s per volume on a GPU, which significantly outperforms competing methods in terms of accuracy and efficiency. The code is publicly available.

A deep learning algorithm for white matter hyperintensity lesion detection and segmentation

PURPOSE

White matter hyperintensity (WMHI) lesions on MR images are an important indication of various types of brain diseases that involve inflammation and blood vessel abnormalities. Automated quantification of the WMHI can be valuable for the clinical management of patients, but existing automated software is often developed for a single type of disease and may not be applicable for clinical scans with thick slices and different scanning protocols. The purpose of the study is to develop and validate an algorithm for automatic quantification of white matter hyperintensity suitable for heterogeneous MRI data with different disease types.

METHODS

We developed and evaluated "DeepWML", a deep learning method for fully automated white matter lesion (WML) segmentation of multicentre FLAIR images. We used MRI from 507 patients, including three distinct white matter diseases, obtained in 9 centres, with a wide range of scanners and acquisition protocols. The automated delineation tool was evaluated through quantitative parameters of Dice similarity, sensitivity and precision compared to manual delineation (gold standard).

RESULTS

The overall median Dice similarity coefficient was 0.78 (range 0.64 ~ 0.86) across the three disease types and multiple centres. The median sensitivity and precision were 0.84 (range 0.67 ~ 0.94) and 0.81 (range 0.64 ~ 0.92), respectively. The tool's performance increased with larger lesion volumes.

CONCLUSION

DeepWML was successfully applied to a wide spectrum of MRI data in the three white matter disease types, which has the potential to improve the practical workflow of white matter lesion delineation.

Evaluating deep learning methods in detecting and segmenting different sizes of brain metastases on 3D post-contrast T1-weighted images

Purpose: We investigate the impact of various deep-learning-based methods for detecting and segmenting metastases with different lesion volume sizes on 3D brain MR images. Approach: A 2.5D U-Net and a 3D U-Net were selected. We also evaluated weak learner fusion of the prediction features generated by the 2.5D and the 3D networks. A 3D fully convolutional one-stage (FCOS) detector was selected as a representative of bounding-box regression-based detection methods. A total of 422 3D post-contrast T1-weighted scans from patients with brain metastases were used. Performances were analyzed based on lesion volume, total metastatic volume per patient, and number of lesions per patient. Results: The performance of detection of the 2.5D and 3D U-Net methods had recall of > 0.83 and precision of > 0.44 for lesion volume > 0.3    cm 3 but deteriorated as metastasis size decreased below 0.3    cm 3 to 0.58 to 0.74 in recall and 0.16 to 0.25 in precision. Compared the two U-Nets for detection capability, high precision was achieved by the 2.5D network, but high recall was achieved by the 3D network for all lesion sizes. The weak learner fusion achieved a balanced performance between the 2.5D and 3D U-Nets; particularly, it increased precision to 0.83 for lesion volumes of 0.1 to 0.3    cm 3 but decreased recall to 0.59. The 3D FCOS detector did not outperform the U-Net methods in detecting either the small or large metastases presumably because of the limited data size. Conclusions: Our study provides the performances of four deep learning methods in relationship to lesion size, total metastasis volume, and number of lesions per patient, providing insight into further development of the deep learning networks.

Tract-specific analysis and neurocognitive functioning in sickle cell patients without history of overt stroke

INTRODUCTION

Sickle cell disease (SCD) is a hereditary blood disorder in which the oxygen-carrying hemoglobin molecule in red blood cells is abnormal. SCD patients are at increased risks for strokes and neurocognitive deficit, even though neurovascular screening and treatments have lowered the rate of overt strokes. Tract-specific analysis (TSA) is a statistical method to evaluate microstructural WM damage in neurodegenerative disorders, using diffusion tensor imaging (DTI).

METHODS

We utilized TSA and compared 11 major brain WM tracts between SCD patients with no history of overt stroke, anemic controls, and healthy controls. We additionally examined the relationship between the most commonly used DTI metric of WM tracts and neurocognitive performance in the SCD patients and healthy controls.

RESULTS

Disruption of WM microstructure orientation-dependent metrics for the SCD patients was found in the genu of the corpus callosum (CC), cortico-spinal tract, inferior fronto-occipital fasciculus, right inferior longitudinal fasciculus, superior longitudinal fasciculus, and left uncinate fasciculus. Neurocognitive performance indicated slower processing speed and lower response inhibition skills in SCD patients compared to controls. TSA abnormalities in the CC were significantly associated with measures of processing speed, working memory, and executive functions.

CONCLUSION

Decreased DTI-derived metrics were observed on six tracts in chronically anemic patients, regardless of anemia subtype, while two tracks with decreased measures were unique to SCD patients. Patients with WMHs had more significant FA abnormalities. Decreased FA values in the CC significantly correlated with all nine neurocognitive tests, suggesting a critical importance for CC in core neurocognitive processes.

Cascaded residual U-net for fully automatic segmentation of 3D carotid artery in high-resolution multi-contrast MR images

Accurate and automatic carotid artery segmentation for magnetic resonance (MR) images is eagerly expected, which can greatly assist a comprehensive study of atherosclerosis and accelerate the translation. Although many efforts have been made, identification of the inner lumen and outer wall in diseased vessels is still a challenging task due to complex vascular deformation, blurred wall boundary, and confusing componential expression. In this paper, we introduce a novel fully automatic 3D framework for simultaneously segmenting the carotid artery from high-resolution multi-contrast MR sequences based on deep learning. First, an optimal channel fitting structure is designed for identity mapping, and a novel 3D residual U-net is used as a basic network. Second, high-resolution MR images are trained using both patch-level and global-level strategies, and the two pre-segmentation results are optimized based on structural characteristics. Third, the optimized pre-segmentation results are cascaded with the patch-cropped MR volume data and trained to segment the carotid lumen and wall. Extensive experiments demonstrate the proposed method outperforms the state-of-the-art 3D Unet-based segmentation models.

White matter has impaired resting oxygen delivery in sickle cell patients

Although modern medical management has lowered overt stroke occurrence in patients with sickle cell disease (SCD), progressive white matter (WM) damage remains common. It is known that cerebral blood flow (CBF) increases to compensate for anemia, but sufficiency of cerebral oxygen delivery, especially in the WM, has not been systematically investigated. Cerebral perfusion was measured by arterial spin labeling in 32 SCD patients (age range: 10-42 years old, 14 males, 7 with HbSC, 25 HbSS) and 25 age and race-matched healthy controls (age range: 15-45 years old, 10 males, 12 with HbAS, 13 HbAA); 8/24 SCD patients were receiving regular blood transfusions and 14/24 non-transfused SCD patients were taking hydroxyurea. Imaging data from control subjects were used to calculate maps for CBF and oxygen delivery in SCD patients and their T-score maps. Whole brain CBF was increased in SCD patients with a mean T-score of 0.5 and correlated with lactate dehydrogenase (r2  = 0.58, P < 0.0001). When corrected for oxygen content and arterial saturation, whole brain and gray matter (GM) oxygen delivery were normal in SCD, but WM oxygen delivery was 35% lower than in controls. Age and hematocrit were the strongest predictors for WM CBF and oxygen delivery in patients with SCD. There was spatial co-localization between regions of low oxygen delivery and WM hyperintensities on T2 FLAIR imaging. To conclude, oxygen delivery is preserved in the GM of SCD patients, but is decreased throughout the WM, particularly in areas prone to WM silent strokes.