Publications

Detailed Information

Hyperbolic embedding of functional brain network and its applications : ๋‡Œ ๊ธฐ๋Šฅ์  ๋„คํŠธ์›Œํฌ์˜ ์Œ๊ณก๊ธฐํ•˜ ๊ณต๊ฐ„์—์˜ ์ž„๋ฒ ๋”ฉ ๋ฐ ๊ทธ ์ ์šฉ: ํœด์‹์ƒํƒœ ๋‡Œ fMRI์— ๊ทผ๊ฑฐํ•˜์—ฌ
based on resting state fMRI

DC Field Value Language
dc.contributor.advisor์ด๋™์ˆ˜-
dc.contributor.author์œ„์›์„-
dc.date.accessioned2022-04-05T06:44:30Z-
dc.date.available2022-04-05T06:44:30Z-
dc.date.issued2021-
dc.identifier.other000000166411-
dc.identifier.urihttps://hdl.handle.net/10371/177803-
dc.identifier.urihttps://dcollection.snu.ac.kr/common/orgView/000000166411ko_KR
dc.descriptionํ•™์œ„๋…ผ๋ฌธ(๋ฐ•์‚ฌ) -- ์„œ์šธ๋Œ€ํ•™๊ต๋Œ€ํ•™์› : ์œตํ•ฉ๊ณผํ•™๊ธฐ์ˆ ๋Œ€ํ•™์› ๋ถ„์ž์˜ํ•™ ๋ฐ ๋ฐ”์ด์˜ค์ œ์•ฝํ•™๊ณผ, 2021.8. ์œ„์›์„.-
dc.description.abstract๋Œ€๋ถ€๋ถ„์˜ ์‹ค์„ธ๊ณ„ ๋„คํŠธ์›Œํฌ์—์„œ ๋„คํŠธ์›Œํฌ์˜ ๊ตฌ์„ฑ์— ์žˆ์–ด์„œ ๊ธฐํ•˜ํ•™์ด ์ค‘์š”ํ•œ ์—ญํ• ์„ ํ•˜๋ฉฐ, ์ตœ๊ทผ ์—ฐ๊ตฌ์—์„œ ๊ตฌ์กฐ์  ๋‡Œ ๋„คํŠธ์›Œํฌ๋Š” ์Œ๊ณก๊ธฐํ•˜์  ํŠน์„ฑ์„ ๊ฐ€์ง€๊ณ  ์žˆ์Œ์ด ๋ฐํ˜€์กŒ๋‹ค. ๋‡Œ์˜ ๊ตฌ์กฐ์™€ ๊ธฐ๋Šฅ์€ ๋ฐ€์ ‘ํ•œ ์—ฐ๊ด€์„ ์ง€๋‹ˆ๊ณ  ์žˆ์œผ๋ฏ€๋กœ, ๊ธฐ๋Šฅ์  ๋‡Œ ๋„คํŠธ์›Œํฌ ์—ญ์‹œ ์Œ๊ณก๊ธฐํ•˜์  ํŠน์„ฑ์„ ์ง€๋‹ˆ๊ณ  ์žˆ์Œ์„ ์ถ”์ •ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋ฒˆ ์—ฐ๊ตฌ์—์„œ, ์šฐ๋ฆฌ๋Š” ํœด์‹๊ธฐ ๋‡Œ ์ž๊ธฐ๊ณต๋ช…์˜์ƒ(rs-fMRI)์„ ํ†ตํ•ด ์ถ”์ถœํ•œ ๊ธฐ๋Šฅ์  ๋‡Œ ์ปค๋„ฅํ†ฐ(connectome)์„ ๋ถ„์„ํ•˜์—ฌ ์ด ๊ฐ€์„ค์„ ์ฆ๋ช…ํ•˜๊ณ ์ž ํ•˜์˜€์œผ๋ฉฐ, ์ด๋ฅผ ์Œ๊ณก๊ณต๊ฐ„์— ์ž„๋ฒ ๋“œ(embed)ํ•จ์œผ๋กœ์จ ๊ธฐ๋Šฅ์  ๋‡Œ ๋„คํŠธ์›Œํฌ์˜ ํŠน์„ฑ์„ ์ƒˆ๋กœ์ด ์กฐ์‚ฌํ•˜๊ณ ์ž ํ•˜์˜€๋‹ค.
๋„คํŠธ์›Œํฌ์˜ ๊ผญ์ง€์ ์€ 274๊ฐœ์˜ ๋ฏธ๋ฆฌ ์ •์˜๋œ ๊ด€์‹ฌ์˜์—ญ(ROI) ํ˜น์€ 6mm ํฌ๊ธฐ์˜ ๋ณต์…€(voxel)์˜ ๋‘ ๊ฐ€์ง€ ์Šค์ผ€์ผ๋กœ ์ •์˜๋˜์—ˆ์œผ๋ฉฐ, ๊ผญ์ง€์  ์‚ฌ์ด์˜ ์—ฐ๊ฒฐ์„ฑ์€ ์ž๊ธฐ๊ณต๋ช… ์˜์ƒ์—์„œ ๊ฐ ์˜์—ญ์˜ ์‹œ๊ฐ„์— ๋”ฐ๋ฅธ BOLD ์‹ ํ˜ธ์˜ ์ƒ๊ด€๊ด€๊ณ„๋ฅผ ์ธก์ •ํ•˜๊ณ  ์ผ์ • ๋ฌธํ„ฑ๊ฐ’(threshold)์„ ์ ์šฉํ•จ์œผ๋กœ์„œ ๊ฒฐ์ •๋˜์—ˆ๋‹ค.
๋จผ์ € ์Œ๊ณก๊ธฐํ•˜ ๋„คํŠธ์›Œํฌ์˜ ํŠน์ง•์ธ ์Šค์ผ€์ผ-ํ”„๋ฆฌ(scale-free)๋ฅผ ๋งŒ์กฑํ•จ์„ ํ™•์ธํ•˜๊ธฐ ์œ„ํ•ด, ๋„คํŠธ์›Œํฌ์˜ ์ฐจ์ˆ˜(degree) ๋ถ„ํฌ์˜ ๊ธ‰์ˆ˜์„ฑ(power-law)์„ ํ‰๊ฐ€ํ•˜์˜€๋‹ค. ์ฐจ์ˆ˜์˜ ํ™•๋ฅ ๋ถ„ํฌ๊ณก์„ ์€ ๋กœ๊ทธ-๋กœ๊ทธ ์Šค์ผ€์ผ์˜ ๊ทธ๋ž˜ํ”„์—์„œ ์šฐํ•˜ํ–ฅํ•˜๋Š” ์ง์„  ๋ชจ์–‘์˜ ๋ถ„ํฌ๋ฅผ ๋ณด์˜€์œผ๋ฉฐ, ์ด๋Š” ์ฆ‰ ์ฐจ์ˆ˜ ๋ถ„ํฌ๊ฐ€ ์ฐจ์ˆ˜์˜ ์Œ์˜ ๊ธ‰์ˆ˜ํ•จ์ˆ˜์— ์˜ํ•ด ๋‚˜ํƒ€๋‚ด์–ด์ง์„ ์˜๋ฏธํ•œ๋‹ค.
์ด์–ด์„œ ๊ธฐ๋Šฅ์  ๋‡Œ ๋„คํŠธ์›Œํฌ์— ๊ฐ€์žฅ ์ ํ•ฉํ•œ ๊ธฐ์ € ๊ธฐํ•˜๋ฅผ ํ™•์ธํ•˜๊ธฐ ์œ„ํ•˜์—ฌ, ๊ทธ๋ž˜ํ”„๋ฅผ ์œ ํด๋ฆฌ๋“œ, ์Œ๊ณก, ๊ตฌ๋ฉด์  ํ‹์„ฑ์„ ๊ฐ€์ง„ ๋‹ค์–‘์ฒด๋“ค์— ์ž„๋ฒ ๋“œํ•˜์—ฌ ์ž„๋ฒ ๋”ฉ์˜ ์ถฉ์‹ค์„ฑ ์ฒ™๋„(fidelity measure)๋“ค์„ ๋น„๊ตํ•˜์˜€๋‹ค. ์ž„๋ฒ ๋“œ ๋Œ€์ƒ์ด ๋œ ์  ๋‹ค์–‘์ฒด๋“ค ์ค‘, 10์ฐจ์› ๋ฐ 2์ฐจ์› ์Œ๊ณก๊ณต๊ฐ„์˜ ํ‰๊ท  ๋’คํ‹€๋ฆผ(distortion)์ด ๋™์ผ ์ฐจ์›์˜ ์œ ํด๋ฆฌ๋“œ ๋‹ค์–‘์ฒด์™€ ๋น„๊ตํ•˜์—ฌ ๋” ๋‚ฎ์•˜๋‹ค.
์ด์–ด, ๋„คํŠธ์›Œํฌ๋ฅผ ๊ตฌ์ฒดํ™” ๋ฐ ์‹œ๊ฐํ™”ํ•˜๊ณ  ๊ทธ ํŠน์ง•์„ ํ™•์ธํ•˜๊ธฐ ์œ„ํ•˜์—ฌ, ๋„คํŠธ์›Œํฌ๋ฅผ ์ด์ฐจ์›์˜ ์Œ๊ณก ์›ํŒ์— ๐•Š1/โ„2 ๊ธฐํ•˜ํ•™์  ๋ชจ๋ธ์— ๋”ฐ๋ผ ์ž„๋ฒ ๋“œํ•˜์˜€๋‹ค. ์ด ์ด์ฐจ์›์˜ ๊ทน์ขŒํ‘œ ํ˜•ํƒœ์˜ ๋ชจ๋ธ์—์„œ ๋ฐ˜๊ฒฝ ๋ฐ ๊ฐ ์ฐจ์›์˜ ์ขŒํ‘œ๋Š” ๊ฐ๊ฐ ๊ผญ์ง€์ ์˜ ์—ฐ๊ฒฐ ์ธ๊ธฐ๋„ ๋ฐ ์œ ์‚ฌ๋„๋ฅผ ๋‚˜ํƒ€๋‚ธ๋‹ค. ROI ์ˆ˜์ค€์˜ ๋ถ„์„์—์„œ๋Š” ํŠน๋ณ„ํžˆ ๋†’์€ ์ธ๊ธฐ๋„๋ฅผ ๊ฐ–๋Š” ์˜์—ญ์€ ๊ด€์ฐฐ๋˜์ง€ ์•Š์•„ ์ž„๋ฒ ๋“œ๋œ ์›ํŒ์˜ ์ค‘์‹ฌ๋ถ€์— ๋นˆ ๊ณต๊ฐ„์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค. ํ•œํŽธ ๊ฐ™์€ ํ•ด๋ถ€ํ•™์  ์—ฝ(lobe)์— ์†ํ•œ ์˜์—ญ๋“ค์€ ๋น„์Šทํ•œ ๊ฐ๋„ ์˜์—ญ ๋‚ด์— ๋ฐ€์ง‘๋˜์—ˆ์œผ๋ฉฐ, ๋ฐ˜๋Œ€์ธก ๋™์ผ ์—ฝ์— ์†ํ•œ ์˜์—ญ๋“ค ์—ญ์‹œ ๊ทธ ๊ฐ์ขŒํ‘œ์˜ ๋ถ„ํฌ๊ฐ€ ๊ตฌ๋ถ„๋˜์ง€ ์•Š์•˜๋‹ค. ์ด๋Š” ๊ธฐ๋Šฅ์  ๋‡Œ ๋„คํŠธ์›Œํฌ์˜ ํ•ด๋ถ€ํ•™์  ์—ฐ๊ด€์„ฑ๊ณผ ๋ฐ˜๋Œ€์ธก ๋™์ผ ์—ฝ ๊ฐ„์˜ ๊ธฐ๋Šฅ์  ์—ฐ๊ด€์„ฑ์„ ๋‚˜ํƒ€๋‚ด๋Š” ๊ฒƒ์œผ๋กœ ๋ณผ ์ˆ˜ ์žˆ๋‹ค.
๋˜ํ•œ, ๋ณต์…€ ์ˆ˜์ค€์˜ ๋ถ„์„์—์„œ๋Š” ์†Œ๋‡Œ์— ์†ํ•œ ๋ณต์…€๋“ค ์ค‘ ๋‹ค์ˆ˜๊ฐ€ ๋„“์€ ๊ฐ์ขŒํ‘œ ์˜์—ญ์— ํฉ๋ฟŒ๋ ค์ง„ ํ˜„์ƒ์ด ๋‚˜ํƒ€๋‚ฌ์œผ๋ฉฐ, ์ด๋Š” ๊ฐœ๊ฐœ ๋ณต์…€์˜ ๊ธฐ๋Šฅ์  ์ด์งˆ์„ฑ์„ ์‹œ์‚ฌํ•œ๋‹ค. ๋˜ํ•œ, ์ „ ์˜์—ญ์— ๊ฑธ์ณ ๋งค์šฐ ์œ ์‚ฌํ•œ ๊ฐ์ขŒํ‘œ๋ฅผ ๊ฐ€์ง„ ๋ฐฉ์‚ฌํ˜•์˜ ๋ง‰๋Œ€ ๋ชจ์–‘์˜ ์ ์˜ ์ง‘ํ•ฉ์ด ๊ด€์ฐฐ๋˜์—ˆ์œผ๋ฉฐ, ๋†’์€ ๊ธฐ๋Šฅ์  ์œ ์‚ฌ์„ฑ์„ ๊ฐ€์ง„ ๋ณต์…€๋“ค๋กœ ๋ณผ ์ˆ˜ ์žˆ๋‹ค. ๋ณต์…€ ์ˆ˜์ค€์˜ ๋„คํŠธ์›Œํฌ์—์„œ ๋‡Œ์˜ ๋…๋ฆฝ์„ฑ๋ถ„ ๋ถ„์„(ICA) ์˜ ๊ฒฐ๊ณผ๋กœ ๋‚˜์˜จ ์„ฑ๋ถ„ ๋„คํŠธ์›Œํฌ๋“ค์„ ํ”Œ๋กœํŒ…ํ•œ ๊ฒฐ๊ณผ, ๊ฐ ๋„คํŠธ์›Œํฌ ์„ฑ๋ถ„์ด ๋†’์€ ๋ฐ€์ง‘๋„๋ฅผ ๋ณด์—ฌ ๋‘ ๋ฐฉ๋ฒ•๋ก  ๊ฐ„ ๊ฒฐ๊ณผ์˜ ์œ ์‚ฌ์„ฑ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค.
์žํ์ŠคํŽ™ํŠธ๋Ÿผ์žฅ์• ์˜ ABIDE II ์˜คํ”ˆ ๋ฐ์ดํ„ฐ์…‹์„ ์ด์šฉํ•˜์—ฌ ๐•Š1/โ„2 ๋ชจ๋ธ์— ๊ทผ๊ฑฐํ•˜์—ฌ, ๋Œ€์กฐ๊ตฐ ํ™˜์ž ๊ทธ๋ฃน๊ณผ ์งˆ๋ณ‘๊ตฐ ํ™˜์ž ๊ฐœ์ธ์˜ ๋„คํŠธ์›Œํฌ๋ฅผ ๋น„๊ตํ•˜๋Š” ๋ถ„์„์„ ์‹œํ–‰ํ•œ ๊ฒฐ๊ณผ, ์งˆ๋ณ‘๊ตฐ์—์„œ ๋‹ค์–‘ํ•œ ํŒจํ„ด์„ ๋ณด์˜€์œผ๋‚˜, ๊ทธ ์ค‘ ์žํ์ฆ ์ง„๋‹จ์„ ๋ฐ›์€ ํ™˜์ž์—์„œ ํ”ผ์งˆ-์„ ์กฐ์ฒด ๊ฒฝ๋กœ์˜ ์ด์ƒ์ด, ์•„์Šคํผ๊ฑฐ์ฆํ›„๊ตฐ ์ง„๋‹จ์„ ๋ฐ›์€ ํ™˜์ž์—์„œ ํ›„์œ„๊ด€์ž๊ณ ๋ž‘ (posterior superior temporal sulcus) ์„ ํฌํ•จํ•˜๋Š” ๊ฒฝ๋กœ์˜ ์ด์ƒ์„ ๋ฐœ๊ฒฌํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค.
๋ถ„์„์˜ ์žฌํ˜„์„ฑ์„ ํ™•์ธํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ๊ฐ™์€ ๋„คํŠธ์›Œํฌ๋ฅผ ๋Œ€์ƒ์œผ๋กœ ์ž„๋ฒ ๋”ฉ ๊ณผ์ •์„ ๋ฐ˜๋ณต ์‹œํ–‰ํ•˜์˜€์„ ๋•Œ, ๋„คํŠธ์›Œํฌ ๋ง๋‹จ์˜ ์ผ๋ถ€ ๊ผญ์ง€์ ์„ ์ œ์™ธํ•˜๋ฉด ๋†’์€ ์žฌํ˜„์„ฑ์„ ๋ณด์˜€๋‹ค. ์˜์ƒ์˜ ์‹œ๊ณ„์—ด(time series) ๋‚ด ์ผ๊ด€์„ฑ์„ ํ™•์ธํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ์˜์ƒ์„ ์‹œ๊ฐ„ ๊ตฌ๊ฐ„์— ๋”ฐ๋ผ ๋ถ„๋ฆฌํ•˜์—ฌ ๋ถ„์„ํ•˜์˜€์„ ๋•Œ, 4๊ตฌ๊ฐ„์œผ๋กœ ๋‚˜๋ˆˆ ์‹œ๊ณ„์—ด ์˜์ƒ์—์„œ๋Š” ์œ ์‚ฌํ•œ ๊ฒฐ๊ณผ๋ฅผ ์–ป์—ˆ์œผ๋‚˜ 30์ดˆ ๊ธธ์ด์˜ 30๊ตฌ๊ฐ„์œผ๋กœ ๋‚˜๋‰˜์—ˆ์„ ๋•Œ๋Š” ์ผ๊ด€์ ์ธ ๊ฒฐ๊ณผ๊ฐ€ ๊ด€์ฐฐ๋˜์ง€ ์•Š์•˜๋‹ค.
์ด ์—ฐ๊ตฌ๋Š” ๋‡Œ ๊ธฐ๋Šฅ์  ๋„คํŠธ์›Œํฌ์— ๋Œ€ํ•œ ๋ถ„์„ ์ค‘ ์ตœ์ดˆ๋กœ ๊ธฐํ•˜ํ•™์  ๊ด€์ ์—์„œ ์ง„ํ–‰๋œ ๊ฒƒ์ด๋ฉฐ, ์ด๋Ÿฌํ•œ ์ƒˆ๋กœ์šด ๊ด€์  ๋ฐ ์งˆ๋ณ‘๊ตฐ ๋Œ€์ƒ์—์„œ ๋‡Œ ๋„คํŠธ์›Œํฌ์˜ ์ด์ƒ์„ ์ฐพ๊ธฐ ์œ„ํ•œ ์ƒˆ๋กœ์šด ๋ฐฉ๋ฒ•๋ก ์„ ์ œ์‹œํ•œ๋‹ค๋Š” ์˜์˜๊ฐ€ ์žˆ๋‹ค.
-
dc.description.abstractFor most of the real-world networks, geometry plays an important role in organizing the network, and recent works have revealed that the geometry in the structural brain network is most likely to be hyperbolic. Therefore, it can be assumed that the geometry of the functional brain network would also be hyperbolic. In this study, we analyzed the functional connectomes from functional magnetic resonance imaging (fMRI) to prove this hypothesis and investigate the characteristics of the network by embedding it into the hyperbolic space, by utilizing human connectome project (HCP) dataset for healthy young adults and Autism Brain Imaging Data Exchange II (ABIDE II) dataset for diseased autism subject and control group.
Nodes of the network were defined at two different scales: by 274 predefined ROIs and 6mm-sized voxels. The adjacency between the nodes was determined by computing the correlation of the time-series of the BOLD signal of brain regions and binarized by adopting threshold value.
First, we aimed to find out whether the network was scale-free by investigating the degree distribution of the functional brain network. The probability distribution function (PDF) versus degree was plotted as a straight line at a log-log scale graph versus the degree of nodes. This indicates that degree distribution is roughly proportional to a power function of degree, or scale-free.
To clarify the most fitting underlying geometry of the network, we then embedded the graph into manifolds of Euclidean, hyperbolic, or spherical spaces and compared the fidelity measures of embeddings. The embedding to the hyperbolic spaces yielded a better fidelity measure compared to other manifolds.
To get a discrete and visible map and investigate the characteristics of the network, we embedded the network in a two-dimensional hyperbolic disc by the ๐•Š1/โ„2 model. The radial and angular dimensions in the embedding is interpreted as popularity and similarity dimensions, respectively. The ROI-wise analysis revealed that no nodes with particularly high popularity were found, which was revealed by a vacant area in the center of the disk. Nodes in the same lobe were more likely to be clustered in narrow similarity dimensions, and the nodes from the homotopic lobes were also functionally clustered. The results indicate the anatomic relevance of the functional brain network and the strong functional coherence of the homotopic area of the cerebral cortex.
The voxel-wise analysis revealed additional features. A large number of voxels from the cerebellum were scattered in the whole angular position, which might reflect the functional heterogeneity of the cerebellum in the sub-ROI level. Additionally, multiple rod-shaped substructures of radial direction were found, which indicates sets of voxels with functional similarity. When compared with independent component analysis (ICA)-driven results, each large-scale component of the brain acquired by ICA showed a consistent pattern of embedding between the subjects.
To find the abnormality of the network in the diseased patient, we utilized the autistic spectrum disorder (ASD) dataset. The two groups of ASD and the control group were found to be comparable in means of the quality of embedding. We calculated the hyperbolic distance between all edges of the network and searched for the alteration of the distance of the individual brain network. Among the variable results among the networks of ASD group subjects, the alteration of the cortico-striatal pathway in an autism patient and posterior superior temporal sulcus (pSTS) in an Aspergers syndrome patient were present, respectively.
The two different anatomically-scaled layers of the network showed a certain degree of correspondence in terms of degree-degree correlation and spreading pattern of network. But anatomically parcellated ROI did not guarantee the functional similarity between the voxels composing it.
Finally, to investigate the reproducibility of the embedding process, we repeatedly performed the embedding process and computed the variance of distance matrices. The result was stable except for end-positioned non-popular nodes. Furthermore, to investigate consistency along time-series of fMRI, we compared network yielded by segments of the time series. The segmented networks showed similar results when divided into four frames, but the result lost consistency when divided into 30 frames of 30 seconds each.
This study is the first to investigate the characteristics of the functional brain network on the basis of hyperbolic geometry. We suggest a new method applicable for assessing the network alteration in subjects with a neuropsychiatric disease, and these approaches grant us a new understanding in analyzing the functional brain network with a geometric perspective.
-
dc.description.tableofcontents1. Introduction 1
1.1. Human brain networks 1
1.1.1. Geometry of human brain networks 2
1.2. Scale-free network 3
1.2.1. Definition of a scale-free network 4
1.3. Embedding of the network in hyperbolic space 5
1.3.1. Hyperbolic spaces and Poincarรฉ disk 5
1.3.2. Geometric model of ๐•Š1/โ„2 9
1.4. The aim of the present study 10
2. Methods 12
2.1. Subjects and image acquisition 12
2.1.1. Human connectome project (HCP) dataset 12
2.1.2. Autism Brain Imaging Data Exchange II (ABIDE II) dataset 12
2.2. Preprocessing for resting-state fMRI 15
2.3. Resting-state networks and functional connectivity analysis 16
2.3.1. Analyzing degree distribution 18
2.4. Assessing underlying geometry 18
2.4.1. The three component spaces 18
2.4.2. Embedding into spaces 20
2.5. Embedding of the network in the ๐•Š1/โ„2 model 22
2.6. Comparison with ICA-driven method 23
2.7. Assessing the quality of embedding 23
2.8. Abnormality detection in the diseased subject 24
2.9. Assessing variability of analysis 27
3. Results 29
3.1. Global characteristics of the network 29
3.1.1. The degree distribution 31
3.1.2. Determining the threshold value of network 34
3.2. Graph embedding into spaces 36
3.3. ๐•Š1/โ„2 model analysis 39
3.4. Quality of the embedding 58
3.5. Alteration of the network in the diseased subject 61
3.6. Variability of results 63
3.6.1. Reproducibility of Mercator 63
3.6.2. Time variance of results 67
4. Discussion 70
4.1. Composition of the network 70
4.2. Scale-freeness of brain network 71
4.3. The underlying geometry of brain network 73
4.4. Hyperbolic plane representation 75
4.4.1. Voxelwise approach 78
4.4.2. Compatibility with ICA 80
4.5. Alteration of the network in ASD subjects 81
4.6. Variability and reproducibility of methods 83
4.7. Further applications 85
5. Conclusion 87
References 89
๊ตญ๋ฌธ ์ดˆ๋ก 106
-
dc.format.extentxi, 109-
dc.language.isoeng-
dc.publisher์„œ์šธ๋Œ€ํ•™๊ต ๋Œ€ํ•™์›-
dc.subjectfunctional brain network-
dc.subjectscale-free network-
dc.subjecthyperbolic geometry-
dc.subjectnetwork embedding-
dc.subjectautistic spectrum disorder-
dc.subject๐•Š1/โ„2 model-
dc.subject๊ธฐ๋Šฅ์  ๋‡Œ ๋„คํŠธ์›Œํฌ-
dc.subject์Šค์ผ€์ผ-ํ”„๋ฆฌ ๋„คํŠธ์›Œํฌ-
dc.subject์Œ๊ณก๊ธฐ ํ•˜-
dc.subject๋„คํŠธ์›Œํฌ ์ž„๋ฒ ๋”ฉ-
dc.subject์žํ์ŠคํŽ™ํŠธ๋Ÿผ์žฅ์• -
dc.subject.ddc610.28-
dc.titleHyperbolic embedding of functional brain network and its applications-
dc.title.alternative๋‡Œ ๊ธฐ๋Šฅ์  ๋„คํŠธ์›Œํฌ์˜ ์Œ๊ณก๊ธฐํ•˜ ๊ณต๊ฐ„์—์˜ ์ž„๋ฒ ๋”ฉ ๋ฐ ๊ทธ ์ ์šฉ: ํœด์‹์ƒํƒœ ๋‡Œ fMRI์— ๊ทผ๊ฑฐํ•˜์—ฌ-
dc.typeThesis-
dc.typeDissertation-
dc.contributor.AlternativeAuthorWonseok Whi-
dc.contributor.department์œตํ•ฉ๊ณผํ•™๊ธฐ์ˆ ๋Œ€ํ•™์› ๋ถ„์ž์˜ํ•™ ๋ฐ ๋ฐ”์ด์˜ค์ œ์•ฝํ•™๊ณผ-
dc.description.degree๋ฐ•์‚ฌ-
dc.date.awarded2021-08-
dc.title.subtitlebased on resting state fMRI-
dc.identifier.uciI804:11032-000000166411-
dc.identifier.holdings000000000046โ–ฒ000000000053โ–ฒ000000166411โ–ฒ-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share