6.30 VA Reading - Smith 2006 NeuroImage (cont.)

Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE, Mackay CE, Watkins KE, Ciccarelli O, Cader MZ, Matthews PM, Behrens TEJ (2006). Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data. NeuroImage; 31: 1487-505.

Preprocessing

• The prealignment is both to correct for head motion during the session and to reduce the effects of gradient coil eddy currents.
• Head motion mostly causes rigid-body image motion, eddy currents appear as a (slightly more general) linear image transformation, to first order.
• Use FLIRT to apply full affine (linear) alignment of each image to the no-diffusion-weighting image, using the mutual information cost function.
• Diffusion tensor is calculated after data prealignment, then calculate FA
• Apply BET brain extraction to the B0 image to exclude non-brain voxels from further consideration.

Nonlinear Alignment

• Keep the general tract structure intact, align the images sufficiently well that the second stage (projection of data onto a tract skeleton) functions correctly.
• Requires intermediate degrees of freedom (DoF).
• Low DoF won't guarantee alignment of major tracts
• High DoF will overwarp the original images that one may not have preserved the overall structure
• Use a nonlinear registration approach based on free-form deformations and B-Splines (a package called "Image Registration Toolkit").
• free-form deformation: deform an image by moving the control points of an underlying mesh.
• Running IRTK takes approximately 20 min on a modern desktop computer to align a single FA image to a different FA target.

Identifying the target for alignment

• Need the subject to be the "most typical" of the entire group
• Register every subject to every other subject, summarise each warp field by its mean displacement, and choose the target subject as being the one with the minimum mean distance to all other subjects.
• An alternative approach (faster) would be to choose an initial target at random. But the search strategy is complex. Still is safer to take the full search strategy described above.

6.30 四犬

四犬 is just the two Chinese characters represent the ten white matter regions shown on the brain pinned near my desk.

Another 4.0 Semester

cGPA is close to 3.9. Fantastic! The hard work finally paid off.

Next step is this master thesis. With a strong research background, how come I can't get another A on this (and the most important one)?

6.29 TBSS manual

预处理 -> 校准 -> 后校准 -> 预统计
1-tbss_1_preproc: make directory -> put files into directory -> scale FA values -> check scaling -> create an exploratory webpage for QC (5 steps);
2-tbss_2_reg: non-linear registration, through either: 1) pre-defined target (FMRIB58_FA); 2) self-selected target; 3) study-specific target (MNI152 standard space). Can be time consuming;
3-tbss_3_postreg: non-linear transformation -> 4D image file -> mean FA -> check threshold -> skeletonization (5 steps);
4-tbss_4_prestats: threshold the mean FA skeleton -> create a distance map -> create a 4D image file containing the skeletonized FA data

• (Smith 2006) distance map: all voxels in the image are filled with a value encoding the distance to the nearest skeleton point.
  There are two limits placed on this perpendicular search within a given subject's FA image. The first is that we constrain the search to remain closer to the starting section of skeleton than to any other section of skeleton; where two separate sections of the skelton lie close to each other, the space in between is divided into two, and each skeleton section can only search voxels within its part of that space.
  Secondly, there is a further constraint placed on the maximum search distance via a soft distance limit. A wide Gaussian function (FWHM 20 mm) is applied as a multiplicative weighting to FA values when carrying out the search for maximum FA (not: this is a weighting function in the search, not a smoothing). This deweights the most distant voxels in a smooth, controlled manner. Once the optimal voxel has been found, its FA value (not weighted by the distance function) is placed into the current skeleton voxel. I am not fully understand this part of the algorithm.

6.28 Pride Parade

 

Though we are already on different track and pursuing different ambitions, I will still show my support to you. It was a great day, and I feel honored to be part of the SF community.

6.27 VA Reading - Segobin 2015 HBM

Segobin S, Ritz L, Lannuzel C, Boudehent C, Vabret F, Eustache F, Beaunieux H, Pitel AL (2015). Integrity of White Matter Microstructure in Alcoholics With and Without Korsakoff's Syndrome. Human Brain Mapping; in progress.

Results: UA(uncomplicated alcoholism) v.s. HC (healthy control)

• Nonparametric permutation tests
• Reported value: T_max, k, n^2
• Lower FA values found in: corpus callosum, anterior limb of the internal capsule, anterior corona radiata, fornix, cingulum, middle cerebellar peduncle, superior cerebellar peduncle (I still don't know where these are!)

KS (Korsakoff's Syndrome) v.s. HC

• Similar to that between UA and HC
• FA was lower in KS
• T and k values are higher
• Regions: corpus callosum, anterior corona radiata, anterior limb of the internal capsule, cingulum, fornix, middle cerebellar peduncle, superior cerebellar peduncle

UA v.s. KS

• Differences mainly found in the corpus callosum and anterior corona radiata
• No gender effect
• In control group, Mann-Whitney U test shows differences between men and women for the cingulum

 Discussion The use of a voxel-based approach may have enabled the observation of the latter finding since a ROI approach, which would average the FA values within a region defined a priori, would not reveal any localized impairments within the fibers. Thus, WM abnormalities in the middle and superior cerebellar peduncles revealed by TBSS suggest that disruption may be more localized than spread-throughout in those fibers.
The study provides consolidating evidence of WM disruptions in the PC that is linked to episodic memory deficits and therefore amnesia [Kessels and Kopelman, 2012 for review). Damaged FCC has been hypothesized to be involved in working memory and executive dysfunction in KS [Wijnia and Goossensen, 2010].
Still difficult to evaluate the cascade of events (neurotransmission dysfunction - local or global network disruption - cellular damage/atrophy) that effectively governs the pathophysiological mechanism of KS when using a cross-sectional paradigm. Longitudinal studies are required to concretely establish this mechanism.
The abnormalities in the microstructural integrity of the corpus callosum have also been hypothesized to be due to thiamine deficiency as observed in a study with rats [He et al., 2007].
Significance: The use of DTI may be particularly relevant as a structural biomarker toward the early identification of UA patients at risk of developing KS. The early identification is important for clinicians to apply the correct and optimal treatment with the aim of preventing severe, debilitating and irreversible neurological complications.
Multi-modalneuroimaging, combined with biological and neuropsychological analyses will enable researchers to explore the characteristics of these clinical forms in terms of detailed microstructure, regional volume, function, enzyme metabolism, and cognitive deficits.

6.26 JHU DTI-based white-matter atlases

http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases

Two white-matter atlases provided by Dr.Susumu Mori, Laboratory of Brain Anatomical MRI, JHU.

• ICBM-DTI-81 white matter labels atlas
• 48 white matter tract labels
• hand segmentation
• standard-space average of diffusion MRI tensor maps
• 81 subjects
• mean age: 39
• JHU white-matter tractography atlas
• 20 structures
• identified probablistically
• average the deterministic tractography
• 28 normal subjects
• mean age: 29

Donna: When get to the TBSS stage, should talk about the WM atlas in more detail. Comparison of the pros/cons for each atlas.
MNI152 standard-space T1-weighted average structural template image (Smith, 2006)

• 152 structural images
• averaged after high-dimensional nonlinear registration into the common MNI152 co-ordinate system.
• 153 T1-volumetric brain MRI scans from the Oxford Project to Investigate Memory and Ageing (OPTIMA) were used. 1.5 Tesla dataset, spatial resolution: 0.86 * 0.86 * 2mm.

Read the reference: Grabner G, Janke AL, Budge MM, Smith D, Pruessner J, Collins DL (2006). Symmetric Atlasing and Model Based Segmentation: An Application to the Hippocampus in Older Adults. MICCAI; 4191: 58-66.
The eventual goal of all these methods is to produce an average model that can accurately both show the similarities in the group and exclude the effect of anatomic variance.
The technique presented here is different in that at each stage images are matched to the model in both their original and (Left-Right) mirrored orientation. The aim of this methodology were to:

1. Build a high resolution symmetric atlas from a population;
2. Automatically extract sub-cortical structures in a population;
3. Allow left-right volumetric comparisons to be made without bias;
4. Eliminate subjective error.

This paper has presented a method whereby automated model based segmentation of symmetric structures is made possible. Done by registering complementary original and flipped images to a target simultaneously, which is important to ensure bias is not introduced at any stage of the model generation. During iterations, registration begins with each of the images in their initial space, such constraint ensures that no bias is introduced towards a particular intermediate model during the construction of the final model.
This work embodies a small extension to the work done by Guimond on iterative atlasing but with two differences: 1) after each iteration of model building, the inverse of the average transformation is applied to the model, which controls for possible bias in the chosen method of registration. 2) in this case we are building a symmetric model.
Due to the nature and extent of processing required to build such a model, the computing time taken to generate this model is large. Approximately 2 weeks of dedicated processing time on a cluster of 16 Linus machines. But, once a model is built, the time taken to then match a novel subjects anatomy to the model is on the order of minutes.
One of the problems is the issue of symmetric bias when comparing structures left-right. The methodologies presented in this paper are intended to enhance the existing methods whereby models are built using non-linear registration with no corresponding impact on accuracy of automated segmentation. The most obvious benefit is the ability to perform comparisons of structures that are by nature assymmetric but posses enough symmetry to allow left-right comparisons to be made.

Dieter: has age been considered when the choosing between the two atlases?
(7.6) Dr.Joseph Schacht's response: the actual ROIs were the conjunction of the common white matter skeleton from TBSS and the JHU ROI because it was the best age match to my (younger) subjects and because it's nicely integrated with the FSL package. I do like that it's probablistic (I like the Harvard-Oxford gray matter atlas for the same reason), but this wasn't a big factor in my decision.

6.26 VA Reading - Smith 2006 NeuroImage

Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE, Mackay CE, Watkins KE, Ciccarelli O, Cader MZ, Matthews PM, Behrens TEJ (2006). Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data. NeuroImage; 31: 1487-505.

Challenge: how can one guarantee that any given standard space voxel contains data from the same part of the same white matter (WM) tract from each and every subject?
We project individual subjects' FA data into a common space in a way that is not dependent on perfect nonlinear registration. This is achieved through:

1. an initial approximate nonlinear registration;
2. projection onto an alignment-invariant tract representation (a.k.a. mean FA skeleton)

Does not require smoothing in the image processing.

VBM is most commonly carried out using the SPM software package. It allows one, subject to interpretation caveats, to find changes anywhere in the brain - it is not necessary to prespecify regions or features of interest.
Alignment issues: one must be very careful not to misinterpret residual misalignments. If one cannot guarantee that alignment is "correct", then it must be assumed that sensitivity to true differences is suboptimal.
Smoothing issues: it is not generally known in advance if the extent of smoothing is matched to the spatial extent of the structure of interest. Smoothing increases the partial voluming problem.
Alternative: specify an ROI. More sophisticated approaches use tractography; the relevant tracts are usually identified by initialising/constraining tractography using hand-drawn ROIs.
What is a MNI152 space? - Kindly supplied by Andrew Janke; derived from 152 structural images, averaged together after high-dimensional nonlinear registration into the common MNI152 co-ordinate system. It corresponds to the "152 nonlinear 6th generation" atlas. (http://www.ncbi.nlm.nih.gov/pubmed/17354756 )

In TBSS, we attempt to bring together the strengths of each approach (VBM-style analysis and Tractography-based approach). We aim to solve the alignment and smoothing issues, while being fully automated, investigating the "whole" brain - not requiring prespecification of tracts of interest.
This is achieved by estimating a "group mean FA skeleton", which represents the centres of all fibre bundles that are generally common to the subjects involved in a study. Each subject's FA data is then projected onto the mean FA skeleton in such a way that each skeleton voxel takes the FA value from the local centre of the nearest relevant tract, thus hopefully resolving issues of alignment and correspondence.
TBSS Steps:

• Preprocessing
• Nonlinear alignment
• Identifying the target for alignment
• Creating the mean FA image and its skeleton
• Statistics and thresholding

6.25 知乎 - 高考填志愿，生物技术如何？ - 我的回答

既是一个回答的帖子，也是一个回顾，过往的努力一定是值得的。

http://www.zhihu.com/question/31509142/answer/52731559
本科生物技术，期间有一年出国交流， 目前即将硕士毕业，有意在生物医学领域深造。从个人的角度谈谈对生物技术专业的感受。

专业选择：首 先，专业的选择， 不仅关乎个人兴趣爱好、职业规划，往往也考量一个家庭的综合实力。父母对我的教育非常重视。凭借他们的努力，家里的经济基础逐渐改善，再加上有他们的支 持，我才可以相对自由地选择探索自己喜欢的学科。我高中时主修化学，当初选择去香港读生物，主要是想换一个环境，学一门新的学科，掌握一门技术，“走遍天 下都不怕”。那个时候，没有过多考虑过将来的就业问题，但既然这么选了，不如就踏踏实实地学下去。

学习过程：当时我们系的本科是三年制，分环境科学和生物技术两个分支，头两年上同样的专业课，有生物化学，微生物学，细胞生物学，动植物学，生态学，遗传 学，以及相应的实验课；第三年上分子生物学，实验课，以及环境科学分支的选修课。这些专业课有一个共性：需要大量的记忆，要背许多生化反应，细胞机理，动 植物分类，基因表达的原理等等，坦白讲，许多知识我现在也不记得了，但是当年就是为了应付考试，也是蛮拼的！实验课都是分组进行，学习最基本的实验技能， 辨认不同的动植物样本，培养严格遵循步骤操作的意识，然后学着写篇幅不大的实验报告。这些实验课确实可以了解具体的生物实验技术，但谈不上掌握。我自己真 正意义上熟悉实验技术并感受到生物研究的快乐，是在大学的最后一年，通过完成毕业论文，反复地推敲和设计每一个实验，不厌其烦地重复实验技术，其间的过程 是漫长甚至枯燥的，但是每当做出漂亮的结果，那种喜悦无以复加。
在生物技术专业的学习中，我最大的感受是，通过独立设计实验，学习实验技术，我的耐心，批判思维，独立性，时间管理，和精细管理的能力都获得了极大的锻 炼。可能这种想法比较理想化，但相信这些能力放在各行各业，都是可贵的品质。当然，隔行如隔山，生物实验做得好，不代表其他事也能做得好；但一个能够理解 并解释复杂的生物原理的人，学习和接受新事物的能力会差到哪里呢？假以时日，一样可以在其他行业干得出色。

大学四年，确实看到身边一些聪明和有天赋的朋友选择了转系，或是毕业后转投其它行业。读研或就业，做学术或去企业，归根结底还是在面对自己最想要的东西的时候做怎样的现实取舍。每个人要面对的现实都不一样，自己选的路，还要自己一步步地走下去。

我对题主的具体情况不太了解，但希望能有一点帮助。

6.25 VA Reading - Smith 2004 Neuroimage

Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TEJ, Johansen-Berg H, Bannister PR, De Luca M, Drobnjak I, Flitney DE, Niazy RK, Saunders J, Vickers J, Zhang Y, De Stefano N, Brady JM, Matthews PM (2004). Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage; 23: S208-S219.

The fundamental challenge in the analysis of functional MRI experiments is to identify voxels that show signal changes varying with changing brain states. Difficult in three ways:

1. SNR is generally poor, with the activation signal being often no larger than the noise level;
2. The neurophysiology that couples the underlying brain activity to the measured response in FMRI is complex and generally poorly understood;
3. The noise consists of a complex blend of spatiotemporal deterministic and stochastic components due to physiological and scanner-based artefacts.

Different analyses techniques for functional MRI research:

1. FILM - voxel wise series analysis
2. FLAME - multilevel modeling for group analysis
3. FEAT - a complete tool for model-based FMRI analysis
4. Bayesian inference on constrained linear basis sets for HRF models
5. MELODIC - probablistic independent component analysis for FMRI
6. Inference via spatial mixture modeling

Different analyses techniques for structural MRI research:

1. BET - brain-nonbrain segmentation
2. FAST - tissue-type segmentation and bias field correction
3. FLIRT - affine intermodal image registration
4. MCFLIRT and FORCE - head motion correction
5. SIENA - brain change analysis

FDT - diffusion and white matter connectivity analysis: the sensitivity of this "diffusion tractography" process to, for example, image noise, partial volume effects, and incomplete signal modeling, has meant that, in general, tractography has been limited to major white matter pathways that are easily found in postmortem dissection. Estimate the probablity density functions (pdfs) on fiber direction leads to being able to quantify belief in the existence of axonal connections between brain regions. By removing the need to make a deterministic decision at every step in the tractography process, we are able to trace beyond regions of low diffusion anisotropy and deep into grey matter structures.

FSL - FMRIB's software library: FMRIB's Software Library. Available as both source code and as self-contained binary distributions for many platforms. Freely available for academic (noncommercial) use. Additional help is available via the FSL email list at www.jiscmail.ac.uk/lists/fsl.html

6.19 FSL and TBSS

FSL is a comprehensive library of analysis tools for FMRI, MRI, and DTI brain imaging data.

Steps in TBSS:

1. Use medium-DoF nonlinear reg (FNIRT) to pre-align all subjects' FA;
2. "Skeletonise" Mean FA;
3. Threshold Mean FA Skeleton, giving "objective" tract map;
4. For each subject's warped FA, fill each point on the mean-space skeleton with nearest maximum FA value (i.e., from the centre of the subject's nearby tract);
5. Do cross-subject voxelwise stats on skeleton-projected FA and Threshold, (e.g., permutation testing, invluding multiple comparison correction).

TBSS Conclusion: Attempting to solve correspondence/smoothing problems; less ambiguity of interpretation / spurious results than VBM; easier to test whole brain than ROI / tractography. Limitations and Dangers: interpretation of partial volume tracts still an issue. (What is partial volume's definition?) Interpretation of crossing tracts. Future work: use full tensor (for registration and test statistics); use other test statistics (MD, PDD, width); multivariate stats (across voxels and/or different diffusion measures) & discriminant (ICA, SVM).
*Partial volume (Wiki): the loss of apparent activity in small objects or regions because of the limited resolution of the imaging system. e.g. If the object or region to be imaged is less than twice the full width at half maximum resolution in x, y, and z dimension of the imaging system, the resultant activity in the object or region is underestimated. A higher resolution decreases this effect, as it better resolves the tissue. Partial volume loss alone occurs only when the surrounding activity of the object or region is zero (Remember the example Roland Henry mentioned in the research defense). And the loss of activity in the object generally involves an increase in activity in adjacent regions, which are considered outside the object (i.e. spillover). For a small object (e.g. a voxel) or an object of size comparable to the spatial resolution of the imaging system, the observed activity is the sum of activity due to partial volume loss plus spillover from adjacent regions. The method to correct for the partial volume effect is referred to as partial volume correction.
Spillover (increase in activity) is the opposite of partial volume loss.

6.15 VA Reading - Trivedy 2013 Behavioural Brain Res

Trivedi R, Bagga D, Bhattacharya D, Kaur P, Kumar P, Khushu S, Tripathi RP, Singh N (2013). White matter damage is associated with memory decline in chronic alcoholics: A quantitative diffusion tensor tractography study. Behavioural Brain Research; 250: 192-198.

10 abstinent chronic alcoholic, 10 demographically equivalent control men

PGI-memory scale (PGIMS) test

Spearman's rank correlation coefficient

Significantly reduced FA in corpus callosum (CC), fornix (FX), and right hemispheric arcuate fasciculus (AF), anterior thalamic radiation (ATR), inferior longitudinal fasciculus (ILF). (where are these regions?)

significant inverse correlation with memory dysfunction score was observed with right cingulum, right uncinate fasciculus, right ILF, and left ILF.

Main findings: white matter deficit in these white matter fibers may contribute to underlying dysfunction in memory in alcoholism.

Alcoholism-impaired effects: cognitive functioning, attention, working memory, speed of processing, visuospatial abilities, executive functions, impulsivity, learning, memory and verbal fluency.

The damaging effects of alcoholism has been focused on the frontal lobes toward the toxicity of ethanol. Inhibition, categorization, flexibility, deduction of rules, organization and planning have mostly been found to be impaired in alcoholism.

Neuroimaging studies have reported an association between white matter integrity and cognitive performance in normal aging and alcoholics.

Diffusion tensor tractography (DTT) offers an overall view of individual fiber bundle in 3D spaces. It also helps in delineation of specific white matter tracts for quantitative analysis.

Significance: many studies have shown brain defects in relation to neurocognitive dysfunction in alcoholics, but very few have correlated DTI measures and memory assessment at the same time.

Statistical Analysis: Student's independent t test was performed to determine the changes in memory dysfunction scores and DTI measures (FA and MD) among controls and alcoholic groups. In alcoholic patients as well as healthy controls, Spearman's rank correlation coefficient was computed to study the relationship between the white matter tract specific DTI measures and memory dysfunction scores.

The mesocorticolimbic reward circuit is viewed as important in encoding maintenance, and retrieval of information. The mesocorticolimbic reward system consists of several brain regions that include amygdata, hippocampus, ventral striatum, ventral diencephalon, and cortical areas such as dorsolateral-prefrontal, orbitofrontal, temporal pole, subcallosal, and cingulate cortices, parahippocampal gyri, and the insula.
CNG (cingulum) is a white matter fiber bundle which connects cingulated gyrus with entorhinal cortex. The main function of entorhinal cortex is to relay messages to and from the hippocampus, which is viewed as the epicenter of long-term memory and spatial navigation.

UNC (uncinate fasciculous) is a ventral associative white matter fiber tract that connects the anterior temporal lobe with the medial and lateral orbitofrontal cortex. The UNC has been found to be associated with memory and language impairments in patients with temporal lobe epilepsy.

6.10 After Research Defense

• For longitudinal study, average 3 timepoints and compare each to the average. choose the one has the least variation. How come I can't remember that?!

6.8 Major Fiber Tracts & Neuroanatomy

Alcohol

• (Segobin, 2015): Frontocerebellar circuit (FCC): middle cerebellar peduncle, superior cerebellar peduncle. Papez circuit (PC): cingulum, fornix
• These are the tracts that are believed to offer less variability in terms of specificity to the PC and FCC, respectively, having a lower involvement in fibers connecting other parts of the brain or involved in other circuitry.
• (Durkee, 2013): fornix, anterior commissure, genu of the corpus callosum, splenium of the corpus callosum, right inferior fronto-occipital fasciculus (IFO), bilateral dorsal cingulum bundle.

Cocaine: corpus callosum (reference)

Meth: (Alicata, 2009) right frontal WM, basal ganglia

Heroin: right orbito-frontal WM, bilateral temporal WM, right parietal WM (reference?). 

(Wang, 2011) reduced FA in left genu of CC, increased MD in left splenium of CC

Nicotine (Smoking):

• (Yu, 2015) increased FA, increased AD, and decreased RD in right posterior limb of the internal capsule, the right external capsule, the right superior corona radiata.
• (Wang, 2009) frontal WM (together with decreased NAA), (Kochunov, 2013) genu of corpus callosum
• Basal ganglia is not a WM, but an aggregated gray matter region.

Cocaine: (Ma, 2015) genu, body, and splenium of corpus callosum, (Bell, 2011) decreased FA in left anterior callosal fibers, left genu of CC, right superior longitudinal fasciculus, right callosal fibers, and bilateral superior corona radiata.

Neuroanatomy

• Fornix: carry signals from hippocampus to mammiliary bodies to anterior nuclei of thalamus; part of limbic system; involve in long-term memmory.
• Cingulum bundle: cingulate gyrus to entorhinal cortex; part of limbic system; involve in learning and correcting mistakes.
• Medial forebrain bundle: a tract containing fibers from basal olfactory region, periamygdaloid region, septal nuclei, brainstem, ventral tegmental area, mesolimbic pathway.