Automatic Labeling of Cortical Sulci Using Spherical Convolutional Neural Networks in a Developmental Cohort

Hao, Lingyan; Bao, Shunxing; Tang, Yucheng; Gao, Riqiang; Parvathaneni, Prasanna; Miller, Jacob A.; Voorhies, Willa I.; Yao, Jewelia K.; Bunge, Silvia A.; Weiner, Kevin S.; Landman, Bennett A.; Lyu, Ilwoo

doi:10.1109/isbi45749.2020.9098414

Cited by 8 publications

(9 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moving forward, we hope to leverage the manual labeling performed here to develop better automated algorithms for sulcal labeling within individuals. Future work using deep learning algorithms data may help to identify tertiary structures in novel brains without manual labeling or intervention (Borne et al, 2020;Hao et al, 2020). Such automated tools have translational applications as tertiary sulci are largely hominoidspecific structures (Amiez et al, 2019) located in association cortices associated with pathology in many neurological disorders.…”

Section: Discussionmentioning

confidence: 99%

“…Indeed, 60-70% of the cortex is buried in sulci and some sulci serve as landmarks that identify different cortical areas, especially in primary sensory cortices (Van Essen and Dierker, 2007;Zilles et al, 2013). In these cases, merely identifying a sulcus provides functional insight (Hinds et al, 2008). Despite this widely replicated relationship between sulcal morphology and functional representations in primary sensory cortices, much less is known regarding the predictability between shallow, tertiary sulci and functional representations in association cortex, especially LPFC.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Overlooked Tertiary Sulci Serve as a Meso-Scale Link between Microstructural and Functional Properties of Human Lateral Prefrontal Cortex

et al. 2021

Self Cite

View full text Add to dashboard Cite

Understanding the relationship between neuroanatomy and function in portions of human cortex that are expanded compared to other mammals such as lateral prefrontal cortex (LPFC) is of major interest in systems and cognitive neuroscience. When considering neuroanatomicalfunctional relationships in LPFC, shallow indentations in cortex known as tertiary sulci have been largely ignored. Here, by implementing a multi-modal approach and manually defining 936 neuroanatomical structures in 72 hemispheres (males and females), we show that a subset of these overlooked tertiary sulci serve as a meso-scale link between microstructural (myelin content) and functional (network connectivity) properties of human LPFC in individuals. For example, the posterior middle frontal sulcus (pmfs) is a tertiary sulcus with three components that differ in their myelin content, resting state connectivity profiles, and engagement across metaanalyses of 83 cognitive tasks. Further, generating microstructural profiles of myelin content across cortical depths for each pmfs component and the surrounding middle frontal gyrus (MFG) shows that both gyral and sulcal components of the MFG have greater myelin content in deeper compared to superficial layers and that the myelin content in superficial layers of the gyral components is greater than sulcal components. These findings support a classic, yet largely unconsidered theory that tertiary sulci may serve as landmarks in association cortices, as well as a modern cognitive neuroscience theory proposing a functional hierarchy in LPFC. As there is a growing need for computational tools that automatically define tertiary sulci throughout cortex, we share pmfs probabilistic sulcal maps with the field..

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Overlooked Tertiary Sulci Serve as a Meso-Scale Link between Microstructural and Functional Properties of Human Lateral Prefrontal Cortex

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“… Simple graph convolution network [ 39 , 40 ]. Graph-based segmentation (e.g., 3D Unet-graph [ 24 , 41 ], Spherical Unet [ 19 , 22 ]). Attention mechanisms for feature representation [ 42 ].…”

Section: Graph Neural Network Backgroundmentioning

confidence: 99%

“…GCNs, however, can be applied to graphs with a varying number of nodes and connectivity [ 20 ]. Spherical CNN architectures can render valid parametrizations in a spherical space without introducing spatial distortions on the sphere (spherical mapping) [ 21 ], and geometric features can be augmented by utilizing surface registration methods [ 22 ]. GCNs can also offer more flexibility to parcellate the cerebral cortex (surface segmentation) by providing better generalization on target-domain datasets where surface data are aligned differently, without the need for manual annotations or explicit alignment of these surfaces [ 23 ].…”

Section: Introductionmentioning

confidence: 99%

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Ahmedt-Aristizabal

Armin

Denman

et al. 2021

Sensors

114

View full text Add to dashboard Cite

With the advances of data-driven machine learning research, a wide variety of prediction problems have been tackled. It has become critical to explore how machine learning and specifically deep learning methods can be exploited to analyse healthcare data. A major limitation of existing methods has been the focus on grid-like data; however, the structure of physiological recordings are often irregular and unordered, which makes it difficult to conceptualise them as a matrix. As such, graph neural networks have attracted significant attention by exploiting implicit information that resides in a biological system, with interacting nodes connected by edges whose weights can be determined by either temporal associations or anatomical junctions. In this survey, we thoroughly review the different types of graph architectures and their applications in healthcare. We provide an overview of these methods in a systematic manner, organized by their domain of application including functional connectivity, anatomical structure, and electrical-based analysis. We also outline the limitations of existing techniques and discuss potential directions for future research.

show abstract

“…Ongoing work is already underway to develop deep learning algorithms to accurately define tertiary sulci automatically in individual subjects, and initial results are promising. 70,71 In summary, using a data-driven, model-based approach, we provide cognitive insights from evolutionarily new brain structures in human LPFC for the first time. After manually defining 1,320 LPFC sulci, our approach revealed that the depths of tertiary sulci reliably predicted reasoning skills above and beyond age.…”

Section: An Immediate Question Generated From Our Findings Is: What Umentioning

confidence: 99%

Cognitive insights from evolutionarily new brain structures in prefrontal cortex

Voorhies

Miller

Yao

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

While the disproportionate expansion of lateral prefrontal cortex (LPFC) throughout evolution is commonly accepted, the relationship between evolutionarily new LPFC brain structures and uniquely human cognitive skills is largely unknown. Here, we tested the relationship between variability in evolutionarily new LPFC tertiary sulci and reasoning skills in a pediatric cohort. A novel data-driven approach in independent discovery and replication samples revealed that the depth of specific LPFC tertiary sulci predicts individual differences in reasoning skills beyond age. These findings support a classic, yet untested, theory linking the protracted development of tertiary sulci to late-developing cognitive processes. We conclude by proposing a mechanistic hypothesis relating the depth of LPFC tertiary sulci to anatomical connections. We suggest that deeper LPFC tertiary sulci reflect reduced short-range connections in white matter, which in turn, improve the efficiency of local neural signals underlying cognitive skills such as reasoning that are central to human cognitive development.

show abstract

Automatic Labeling of Cortical Sulci Using Spherical Convolutional Neural Networks in a Developmental Cohort

Cited by 8 publications

References 14 publications

Overlooked Tertiary Sulci Serve as a Meso-Scale Link between Microstructural and Functional Properties of Human Lateral Prefrontal Cortex

Overlooked Tertiary Sulci Serve as a Meso-Scale Link between Microstructural and Functional Properties of Human Lateral Prefrontal Cortex

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Cognitive insights from evolutionarily new brain structures in prefrontal cortex

Contact Info

Product

Resources

About