Automated thalamic nuclei segmentation using multi-planar cascaded convolutional neural networks

Mutually communicated model based on multi-parametric MRI for automated segmentation and classification of prostate cancer

BACKGROUND

Multiparametric magnetic resonance imaging (mp-MRI) is introduced and established as a noninvasive alternative for prostate cancer (PCa) detection and characterization.

PURPOSE

To develop and evaluate a mutually communicated deep learning segmentation and classification network (MC-DSCN) based on mp-MRI for prostate segmentation and PCa diagnosis.

METHODS

The proposed MC-DSCN can transfer mutual information between segmentation and classification components and facilitate each other in a bootstrapping way. For classification task, the MC-DSCN can transfer the masks produced by the coarse segmentation component to the classification component to exclude irrelevant regions and facilitate classification. For segmentation task, this model can transfer the high-quality localization information learned by the classification component to the fine segmentation component to mitigate the impact of inaccurate localization on segmentation results. Consecutive MRI exams of patients were retrospectively collected from two medical centers (referred to as center A and B). Two experienced radiologists segmented the prostate regions, and the ground truth of the classification refers to the prostate biopsy results. MC-DSCN was designed, trained, and validated using different combinations of distinct MRI sequences as input (e.g., T2-weighted and apparent diffusion coefficient) and the effect of different architectures on the network's performance was tested and discussed. Data from center A were used for training, validation, and internal testing, while another center's data were used for external testing. The statistical analysis is performed to evaluate the performance of the MC-DSCN. The DeLong test and paired t-test were used to assess the performance of classification and segmentation, respectively.

RESULTS

In total, 134 patients were included. The proposed MC-DSCN outperforms the networks that were designed solely for segmentation or classification. Regarding the segmentation task, the classification localization information helped to improve the IOU in center A: from 84.5% to 87.8% (p < 0.01) and in center B: from 83.8% to 87.1% (p < 0.01), while the area under curve (AUC) of PCa classification was improved in center A: from 0.946 to 0.991 (p < 0.02) and in center B: from 0.926 to 0.955 (p < 0.01) as a result of the additional information provided by the prostate segmentation.

CONCLUSION

The proposed architecture could effectively transfer mutual information between segmentation and classification components and facilitate each other in a bootstrapping way, thus outperforming the networks designed to perform only one task.

Thalamic nuclei atrophy at high and heterogenous rates during cognitively unimpaired human aging

The thalamus is a central integration structure in the brain, receiving and distributing information among the cerebral cortex, subcortical structures, and the peripheral nervous system. Prior studies clearly show that the thalamus atrophies in cognitively unimpaired aging. However, the thalamus is comprised of multiple nuclei involved in a wide range of functions, and the age-related atrophy of individual thalamic nuclei remains unknown. Using a recently developed automated method of identifying thalamic nuclei (3T or 7T MRI with white-matter-nulled MPRAGE contrast and THOMAS segmentation) and a cross-sectional design, we evaluated the age-related atrophy rate for 10 thalamic nuclei (AV, CM, VA, VLA, VLP, VPL, pulvinar, LGN, MGN, MD) and an epithalamic nucleus (habenula). We also used T1-weighted images with the FreeSurfer SAMSEG segmentation method to identify and measure age-related atrophy for 11 extra-thalamic structures (cerebral cortex, cerebral white matter, cerebellar cortex, cerebellar white matter, amygdala, hippocampus, caudate, putamen, nucleus accumbens, pallidum, and lateral ventricle). In 198 cognitively unimpaired participants with ages spanning 20-88 years, we found that the whole thalamus atrophied at a rate of 0.45% per year, and that thalamic nuclei had widely varying age-related atrophy rates, ranging from 0.06% to 1.18% per year. A functional grouping analysis revealed that the thalamic nuclei involved in cognitive (AV, MD; 0.53% atrophy per year), visual (LGN, pulvinar; 0.62% atrophy per year), and auditory/vestibular (MGN; 0.64% atrophy per year) functions atrophied at significantly higher rates than those involved in motor (VA, VLA, VLP, and CM; 0.37% atrophy per year) and somatosensory (VPL; 0.32% atrophy per year) functions. A proximity-to-CSF analysis showed that the group of thalamic nuclei situated immediately adjacent to CSF atrophied at a significantly greater atrophy rate (0.59% atrophy per year) than that of the group of nuclei located farther from CSF (0.36% atrophy per year), supporting a growing hypothesis that CSF-mediated factors contribute to neurodegeneration. We did not find any significant hemispheric differences in these rates of change for thalamic nuclei. Only the CM thalamic nucleus showed a sex-specific difference in atrophy rates, atrophying at a greater rate in male versus female participants. Roughly half of the thalamic nuclei showed greater atrophy than all extra-thalamic structures examined (0% to 0.54% per year). These results show the value of white-matter-nulled MPRAGE imaging and THOMAS segmentation for measuring distinct thalamic nuclei and for characterizing the high and heterogeneous atrophy rates of the thalamus and its nuclei across the adult lifespan. Collectively, these methods and results advance our understanding of the role of thalamic substructures in neurocognitive and disease-related changes that occur with aging.

Convolutional Neural Network Based Frameworks for Fast Automatic Segmentation of Thalamic Nuclei from Native and Synthesized Contrast Structural MRI

Thalamic nuclei have been implicated in several neurological diseases. Thalamic nuclei parcellation from structural MRI is challenging due to poor intra-thalamic nuclear contrast while methods based on diffusion and functional MRI are affected by limited spatial resolution and image distortion. Existing multi-atlas based techniques are often computationally intensive and time-consuming. In this work, we propose a 3D convolutional neural network (CNN) based framework for thalamic nuclei parcellation using T1-weighted Magnetization Prepared Rapid Gradient Echo (MPRAGE) images. Transformation of images to an efficient representation has been proposed to improve the performance of subsequent classification tasks especially when working with limited labeled data. We investigate this by transforming the MPRAGE images to White-Matter-nulled MPRAGE (WMn-MPRAGE) contrast, previously shown to exhibit good intra-thalamic nuclear contrast, prior to the segmentation step. We trained two 3D segmentation frameworks using MPRAGE images (n = 35 subjects): (a) a native contrast segmentation (NCS) on MPRAGE images and (b) a synthesized contrast segmentation (SCS) where synthesized WMn-MPRAGE representation generated by a contrast synthesis CNN were used. Thalamic nuclei labels were generated using THOMAS, a multi-atlas segmentation technique proposed for WMn-MPRAGE images. The segmentation accuracy and clinical utility were evaluated on a healthy cohort (n = 12) and a cohort (n = 45) comprising of healthy subjects and patients with alcohol use disorder (AUD), respectively. Both the segmentation CNNs yielded comparable performances on most thalamic nuclei with Dice scores greater than 0.84 for larger nuclei and at least 0.7 for smaller nuclei. However, for some nuclei, the SCS CNN yielded significant improvements in Dice scores (medial geniculate nucleus, P = 0.003, centromedian nucleus, P = 0.01) and percent volume difference (ventral anterior, P = 0.001, ventral posterior lateral, P = 0.01) over NCS. In the AUD cohort, the SCS CNN demonstrated a significant atrophy in ventral lateral posterior nucleus in AUD patients compared to healthy age-matched controls (P = 0.01), agreeing with previous studies on thalamic atrophy in alcoholism, whereas the NCS CNN showed spurious atrophy of the ventral posterior lateral nucleus. CNN-based segmentation of thalamic nuclei provides a fast and automated technique for thalamic nuclei prediction in MPRAGE images. The transformation of images to an efficient representation, such as WMn-MPRAGE, can provide further improvements in segmentation performance.

Structural assessment of thalamus morphology in brain disorders: A review and recommendation of thalamic nucleus segmentation and shape analysis

The thalamus is a central brain structure crucially involved in cognitive, emotional, sensory, and motor functions and is often reported to be involved in the pathophysiology of neurological and psychiatric disorders. The functional subdivision of the thalamus warrants morphological investigation on the level of individual subnuclei. In addition to volumetric measures, the investigation of other morphological features may give additional insights into thalamic morphology. For instance, shape features offer a higher spatial resolution by revealing small, regional differences that are left undetected in volumetric analyses. In this review, we discuss the benefits and limitations of recent advances in neuroimaging techniques to investigate thalamic morphology in vivo, leading to our proposed methodology. This methodology consists of available pipelines for volume and shape analysis, focussing on the morphological features of volume, thickness, and surface area. We demonstrate this combined approach in a Parkinson's disease cohort to illustrate their complementarity. Considering our findings, we recommend a combined methodology as it allows for more sensitive investigation of thalamic morphology in clinical populations.

The specificity of thalamic alterations in Korsakoff's syndrome: Implications for the study of amnesia

The pathophysiological mechanisms behind amnesia are still unknown. Recent literature, through the study of patients with Alcohol Use Disorder with and without Korsakoff's syndrome, increasingly shows that physiological alterations to the thalamus have an important role in the development of amnesia. This review gives an overview of neuropsychological, neuropathological and neuroimaging contributions to the understanding of Korsakoff's syndrome, highlighting the central role of the thalamus in this amnesia. The thalamus being a multi-nucleus structure, the limitations regarding the loci, nature and alterations to specific nuclei are discussed, along with potential solutions. Finally, future directions for clinical research are laid out to unravel the intricacies inherent to amnesia. They consider the need to evaluate the physiological role of the thalamus, not only as an entity but also as part of a brain circuit through a more integrative approach.