Jan Derrfuss, 1 * Marcel Brass, 2 D. Yves von Cramon, 3,4 Gabriele Lohmann, 3 and Katrin Amunts 1,5

Transcription

1 Human Bain Mapping 30: (2009) Neual Activations at the Junction of the Infeio Fontal Sulcus and the Infeio Pecental Sulcus: Inteindividual Vaiability, Reliability, and Association With Sulcal Mophology Jan Defuss, 1 * Macel Bass, 2 D. Yves von Camon, 3,4 Gabiele Lohmann, 3 and Katin Amunts 1,5 1 Medicine (INB3), Institute of Neuosciences and Biophysics - Medicine (INB-3), Reseach Cente Juelich, Juelich, Gemany 2 Depatment of Expeimental Psychology, Ghent Univesity, Ghent, Belgium 3 Max Planck Institute fo Human Cognitive and Bain Sciences, Leipzig, Gemany 4 Max Planck Institute fo Neuological Reseach, Cologne, Gemany 5 Depatment of Psychiaty and Psychotheapy, Univesity Hospital, RWTH Aachen, Gemany Abstact: The sulcal mophology of the human fontal lobe is highly vaiable. Although the stuctual images usually acquied in functional magnetic esonance imaging studies povide infomation about this inteindividual vaiability, this infomation is only aely used to elate stuctue and function. Hee, we investigated the spatial elationship between posteio fontolateal activations in a taskswitching paadigm and the junction of the infeio fontal sulcus and the infeio pecental sulcus (infeio fontal junction, IFJ) on an individual-subject basis. Results show that, although vaiable in tems of steeotaxic coodinates, the posteio fontolateal activations obseved in task-switching ae consistently and eliably located at the IFJ in the bains of individual paticipants. The IFJ shaes such consistent localization with othe nonpimay aeas as motion-sensitive aea V5/MT and the fontal eye field. Building on tension-based models of mophogenesis, this stuctue function coespondence might indicate that the cytoachitectonic aea undelying activations of the IFJ develops at ealy stages of cotical folding. Hum Bain Mapp 30: , VC 2007 Wiley-Liss, Inc. Key wods: fontal lobe; pefontal cotex; cognitive contol; magnetic esonance imaging Additional Suppoting Infomation may be found in the online vesion of this aticle. Contact gant sponso: Deutsche Foschungsgemeinschaft; Contact gant numbe: AM 118/1-1; Contact gant sponso: Bain Imaging Cente West (Bundesministeium fü Bildung und Foschung); Contact gant numbe: 01GO0204; Contact gant sponso: Helmholtz-Gemeinschaft; Contact gant numbe: VH-NG-012; Contact gant sponsos: National Institute of Mental Health, National Institute of Neuological Disodes and Stoke, National Institute on Dug Abuse, and National Cance Cente; Contact gant numbe: 5 P01EB *Coespondence to: Jan Defuss, Institute of Medicine, Reseach Cente Juelich, Juelich, Gemany. j.defuss@fz-juelich.de Received fo publication 4 July 2007; Revised 19 Septembe 2007; Accepted 21 Septembe 2007 DOI: /hbm Published online 10 Decembe 2007 in Wiley InteScience (www. intescience.wiley.com). VC 2007 Wiley-Liss, Inc.

2 Defuss et al. INTRODUCTION The anatomy of the fontal lobe of the human bain is highly vaiable between subjects. This is tue in tems of sulcal mophology [Cunningham, 1892; Ebestalle, 1890; Gemann et al., 2005; Juch et al., 2005; Lohmann et al., 1999; Ono et al., 1990] and cytoachitectue [Amunts et al., 1999; Rajkowska and Goldman-Rakic, 1995]. While cytoachitectonic featues ae usually not identifiable on anatomical magnetic esonance images acquied in functional imaging studies (but see, e.g., Waltes et al. [2003, 2007] fo anatomical high-esolution studies), these images povide us with infomation about the goss mophology of individual bains. The impotance of systematically analyzing this infomation was ecently emphasized by Devlin and Poldack [2007], Fadiga [2007], and Tzouio-Mazoye et al. [2007]. Howeve, in the lage majoity of imaging studies activation maps ae aveaged acoss bains and displayed on magnetic esonance images of a single-subject template bain (e.g., the Colin27 ; [Holmes et al., 1998]), o an aveage of multiple bains (e.g., the ICBM152 [ Toga et al., 2006], o the aveaged bains of the subjects who paticipated in the study). This pocedue, howeve, pecludes an investigation of functional anatomical elationships on an individual-subject basis. The consideable sulcal vaiability of the fontal lobe was aleady noted by Cunningham [1892], who was able to identify 27 vaiations of the patten of the intemediate fontal sulcus in 69 hemisphees. 1 Moe ecently, Ono et al. [1990] published a compehensive atlas of sulcal mophology. These authos descibed fou pattens of the infeio pecental sulcus (acuate, amified, bayonet, y- shaped) and fou pattens of the infeio fontal sulcus (continuous, one to thee inteuptions). Ono et al. found a junction of the infeio pecental sulcus and the infeio fontal sulcus in 90% of cases. 2 In addition to vaiations in the sulcal patten, thee is also a high degee of vaiability in the position of sulci when bains ae bought into a common efeence space. Fo example, Gemann et al. [2005] have shown that the location of the infeio pecental sulcus may vay up to 2 cm in anteio posteio diection. Given this vaiability, the question aises how individualsubject functional imaging activations ae elated to sulcal pattens. It is well known fom cytoachitectonic studies that thee is a consistent elationship between sulci and 1 Note, howeve, that Cunningham may have been somewhat biased to oveestimate the vaiability of the intemediate fontal sulcus. He viewed this vaiability as suppoting his notion that the intemediate fontal sulcus is not homologous to the pincipal sulcus of macaques as poposed 2 yeas ealie by Ebestalle. Fo a moe ecent investigation of the intemediate fontal sulcus focusing on sulcal basins, see Huttne et al. [2005]. 2 Note, howeve, that it has ecently been suggested that in many cases the most posteio aspect of the infeio fontal sulcus might be a pat of the infeio pecental sulcus [Gemann et al., 2005]. cytoachitectonic aeas fo a numbe of pimay aeas [e.g., Amunts et al., 2000; Geye et al., 1996, 1999; Rademache et al., 1993]. Accodingly, pimay visual activations in functional imaging studies have been found in the calcaine sulcus [e.g., Bidge et al., 2005; DeYoe et al., 1996; Seeno et al., 1995; Tootell et al., 1998] and pimay moto activations in the cental sulcus [e.g., Dechent and Fahm, 2003; Geye et al., 1996; Yousy et al., 1997]. Thus, fo these pimay aeas thee is a stong coespondence between sulcal mophology and functional imaging activations. Howeve, it is less clea whethe thee might be a simila coespondence fo nonpimay aeas. In one of the few functional imaging studies investigating sulcal-functional elationships in highe-ode aeas, Watson et al. [1993] employing positon emission tomogaphy and stuctual magnetic esonance imaging (MRI) showed that while peak activations of the motion-sensitive aea V5 vaied up to 27 mm between paticipants, the location of these activations was consistent in elation to the individual sulcal pattens: activations wee most fequently located just posteio to the junction of the ascending limb of the infeio tempoal sulcus and the lateal occipital sulcus. In a subsequent functional magnetic esonance imaging (fmri) study, Dumoulin et al. [2000] showed that the junction of the ascending limb of the infeio tempoal sulcus and the infeio tempoal sulcus might be a moe eliable landmak fo the location of V5. Dumoulin et al. found that the individual activations wee on aveage less than 10 mm fom the junction of these sulci. Anothe study investigating the elationship of sulcus mophology and functional imaging activations was conducted by Amiez et al. [2006]. In this study, Amiez et al. showed that the patten of the supeio pecental sulcus povided infomation about the location of activations in a saccade task and an abitay visuomoto mapping task: eye movement activations wee consistently located in the vental pat of the supeio pecental sulcus, wheeas the abitay visuomoto mapping activations wee located in the dosal pat of the supeio pecental sulcus. Thus, at least some nonpimay aeas appea to be consistently associated with paticula sulci. Cuently it is not clea whethe a simila coespondence exists fo activations located at the junction of the infeio fontal sulcus and the infeio pecental sulcus (the so-called IFJ; [Bass and von Camon, 2002]). Activations of this egion wee epeatedly found in pevious imaging studies that employed task-switching paadigms o the Stoop task (fo a meta-analysis, [see Defuss et al., 2005]). Employing vaiants of the task-switching paadigm, Bass and von Camon have shown that activity of the IFJ is coelated with efficient task pepaation [Bass and von Camon, 2002] and that this activity cannot be explained solely by cue-encoding pocesses [Bass and von Camon, 2004]. The studies by Bass and von Camon [2002, 2004] indicated that the IFJ plays a ole in the updating of task epesentations. Howeve, it is cuently not known to which degee the IFJ is specifically involved in updating pocesses o 300

3 Single-Subject IFJ Activations whethe it might play a moe geneal ole in the integation of woking memoy, phonological, and moto-elated infomation [Bass et al., 2005]. Based on studies that employed aveaging acoss paticipants, Defuss et al. [2004] have suggested that activations of the IFJ ae appoximately located in the following coodinate ange: x-coodinates between 630 and 647, y-coodinates 3 between 21 and 10, and z-coodinates between 27 and 40. This egion encompasses pats of Bodmann aeas 6, 9, and 44 accoding to the atlas of Talaiach and Tounoux [1988]. Hee, data fom Bass and von Camon [2004] wee eanalyzed to investigate activations of the IFJ on an individualsubject level. In paticula, we wee inteested in the question whethe individual sulcal mophologies would be pedictive of functional activations. Futhemoe, a numbe of subjects who paticipated in the study of Bass and von Camon [2004] also paticipated in the studies by Defuss et al. [2004] and Zysset et al. [2006]. Fo these subjects, also the data fom the latte two studies wee e-analyzed to investigate the stability of IFJ activations ove time. MATERIALS AND METHODS Paticipants Fouteen voluntees (8 females, 6 males) took pat in the study by Bass and von Camon [2004]. All wee ight-handed as assessed by the Edinbugh Inventoy [Oldfield, 1971] and had nomal o coected-to-nomal vision. Mean age was 24.4 yeas (SD 1.9). Witten consent in accodance with the egulations of the local ethics committee was obtained fom all paticipants. Five of the 14 paticipants also paticipated in a study by Defuss et al. [2004], thee voluntees also paticipated in a study by Zysset et al. [2006]. Behavioal Paadigm 3 Please note that thee is a pinting eo in ou 2004 publication. The y-coodinates should have ead 21 to 10, not 1 to 10. The task employed was a modified vesion of the taskswitching paadigm intoduced by Sudevan and Taylo [1987]. Hee, we give only a bief account of the task; fo a moe detailed desciption and a figue of the sequence of events in a tial, see Bass and von Camon [2004]. Tagets wee numbes between 20 and 40 (excluding 30). On each tial, paticipants had to execute one of two tasks: eithe they had to decide whethe the numbe was odd o even (paity task); o they had to decide whethe the numbe was smalle o geate than 30 (magnitude task). Which task to execute was indicated by cues: the paity task was indicated by a diamond o a tiangle, and the magnitude task was indicated by a squae o an inveted tiangle. In double-cue tials, two cues wee pesented consecutively befoe the taget appeaed. Thee wee thee diffeent double-cue conditions: a condition in which the same cue was pesented twice (cue-epetition condition), a condition in which the cue changed, but both cues indicated the same task (cue-switch condition) and a condition in which the cue switched to indicate the othe task (meaning-switch condition). As the goup-level analysis by Bass and von Camon [2004] was based on the contast of meaning-switch vs. cue-switch tials, only these conditions wee consideed fo ou single-subject analysis. The expeiment consisted of two blocks of 192 tials each; 192 double-cue tials (64 cue epetition/64 cue switch/64 meaning switch) wee andomly intemixed with 128 single-cue tials (not epoted hee) and 64 null events. Tial length was 6 s. MRI Data Acquisition Imaging was pefomed using a 3-T scanne (Medspec 30/100, Buke, Ettlingen, Gemany). Twenty axial slices (19.2 cm FOV, matix, 4 mm thickness, 1 mm spacing) paallel to the AC-PC plane wee acquied using a single-shot gadient-ecalled EPI sequence (TR 5 2,000 ms, TE 5 30 ms, flip angle 5 908). Befoe the functional uns, 20 coesponding anatomical 2-D MDEFT and EPI-T1 slices wee acquied. In a sepaate scanning session, highesolution T1-weighted efeence bain images wee acquied fom each paticipant. Stimuli wee pesented using a head-mounted display with a esolution of 1, and a efesh ate of 60 Hz. Identification of Sulci Othogonal sections of the high-esolution T1-weighted bain scans, 3-D endeings, and white-matte segmentations wee used to identify the cental sulcus, the anteio ascending amus of the Sylvian fissue, the infeio pecental sulcus, and the infeio fontal sulcus in all bains. The cental sulcus was identified using the citeia of Yousy et al. [1997] and the anteio ascending amus of the Sylvian fissue as the lateal continuation of the cicula sulcus of the insula. The infeio pecental sulcus was defined as the fist majo sulcus anteio and appoximately paallel to the infeio pat of the cental sulcus [Ebeling et al., 1989]. The infeio fontal sulcus was identified as the fist majo sulcus unning in a caudoostal diection and lying immediately dosal to the anteio ascending amus of the Sylvian fissue. Imaging Analysis The functional magnetic esonance imaging (fmri) data wee pocessed with LIPSIA softwae [Lohmann et al., 2001]. This softwae package contains tools fo pepocessing, egistation, statistical evaluation, and pesentation of fmri data. Fist, the functional data wee coected fo movement atifacts. Then, the tempoal offset diffeences between slices acquied in one scan wee coected using a 301

4 Defuss et al. spline intepolation algoithm. A tempoal high-pass filte with a cutoff fequency of 80 s was used fo baseline coection of the signal. Anatomical egistation fo each paticipant was done in thee steps: fist, we computed a tansfomation matix containing thee otational and thee tanslational paametes that optimized the match between the 2-D MDEFT/ EPI-T1 slices and the 3-D efeence data set. Then, this tansfomation matix was nomalized to standad Talaiach bain size [Talaiach and Tounoux, 1988] by applying linea scaling. Finally, the nomalized tansfomation matix was applied to the functional data using tilinea intepolation. Slice gaps wee intepolated to geneate output data with a spatial esolution of mm 3. The statistical evaluation was based on a least-squaes estimation using the geneal linea model fo seially autocoelated obsevations [Aguie et al., 1997; Fiston et al., 1995; Wosley and Fiston, 1995; Zaahn et al., 1997]. The design matix was geneated utilizing a synthetic hemodynamic esponse function with a vaiable delay [Fiston et al., 1998]. The model equation, including the obsevation data, the design matix, and the eo tem, was convolved with a Gaussian kenel with a dispesion of 4 s FWHM to account fo the tempoal autocoelation [Wosley and Fiston, 1995]. Subsequently, z maps fo the contast of meaning-switch vs. null-event tials wee computed fo each paticipant. In the next analysis step, the T1-weighted individual MR data wee bought into Talaiach space and white-matte segmentations wee pefomed. We used these white-matte segmentations and the T1 volumes to identify the left infeio pecental sulcus, left infeio fontal sulcus, and, if pesent, thei junction in evey bain. Then, we ovelaid the individual z maps on the coesponding anatomical data and tied to localize an activation that was located at o close to the junction of the infeio pecental sulcus and the infeio fontal sulcus. When such an activation was identified, we adjusted the significance theshold in such a way that this activation had a volume of 2,000 mm 3. That is, instead of choosing the same significance theshold fo evey paticipant, we decided to choose a significance theshold that esulted in a pedefined activation volume aound the IFJ peak. In the pesent study, this appoach was supeio to thesholding at a common significance level since we wanted to compae the localization of these activations acoss paticipants. Thus, we wee less inteested in a common significance theshold constant acoss paticipants, but athe in a constant volume of activations. In addition, the volumebased theshold assued that each paticipant contibuted the same volume to a map displaying the ovelap of IFJ activations. The 2,000 mm 3 volume was chosen based on an appoximation of the volume of the IFJ as esulting fom the coodinate ange descibed above (17 mm 3 11 mm 3 13 mm 5 2,431 mm 3 ). The selection of this volume does not imply that the cytoachitectonic aea undelying activations of the IFJ has appoximately the same volume. Instead, the chosen volume was a pagmatic decision given that we cuently do not know what the exact volume of this aea might be. It was selected lage enough to likely include the putative undelying cytoachitectonic aea, but small enough not to include majo pats of neighboing functional aeas. As we identified IFJ activations based on a null-event contast, we also computed time couses to ascetain diffeences between the meaning-switch and the cue-switch condition. These tial-aveaged time couses fo the meaningswitch, the cue-switch, and the null-event tials wee extacted fom the peak coodinate in the IFJ and expessed in pecent signal change units. The activations wee displayed in thee ways on the individual bains. Fist, activations wee shown on endeings of white-matte segmentations of T1-weighted MRI scans. These figues allow an identification of the sulci of the fontolateal cotex and give a bette idea of the localization of the activation volume. Second, we taced the outline of the activation volume and the depths of the infeio pecental sulcus and the posteio pat of the infeio fontal sulcus on the white-matte segmentations and displayed these in schematic line dawings. Thid, we show an axial and a sagittal slice though the peak activation in the IFJ. These figues enable the identification of the peak location elative to the suounding sulci and the localization of the activations in tems of Talaiach space. Lateality Analysis In a meta-analysis of switching studies [Defuss et al., 2005], we found significant clusteing of activations at the left and the ight IFJ. Howeve, in the study by Bass and von Camon [2004] significant IFJ activity was found only in the left hemisphee. Thus, we decided to also inspect ight-hemispheic single-subject IFJ activations to gain a bette undestanding of lateality effects in the double-cueing paadigm. As Subject 12 showed eliable IFJ activation only in the ight hemisphee, this subject was not included in the lateality analysis. In a fist step, ight IFJ activations wee identified analogue to those on the left. The mean ight IFJ peak and the distance to the left IFJ peak (absolute x values) wee computed. Then, pecent signal change values wee extacted fom the left and ight IFJ peak voxels fo the meaningswitch condition and the cue-switch condition. In a thid step, the maximum diffeence between these two conditions in a time window between 5 and 7 s was identified fo each paticipant. These maximum pecent signal change diffeences wee then tested fo significance. Reliability Analysis A subset of six subjects who paticipated in the study by Bass and von Camon [2004] also paticipated in the study by Defuss et al. [2004]. In the latte study, paticipants pefomed a task-switching paadigm, an n-back 302

5 Single-Subject IFJ Activations Figue 1. Left fontolateal views of white-matte segmentations of six subjects with functional imaging maps ovelaid in ed. Relevant sulci ae highlighted by tiangles. Note the consistent activation at the junction of the infeio pecental sulcus and the infeio fontal sulcus. task, and a Stoop task (fo details, [see Defuss et al., 2004]). Thee futhe subjects also paticipated in a study by Zysset et al. [2006]. This study employed a matching vesion of the Stoop task (fo details, [see Zysset et al., 2006]). The fmri measuements wee pefomed between August 2002 and Mach To investigate the eliability of IFJ activations, we eanalyzed the data fom the paticipants who took pat in two o moe of the above studies. This was done with the same paametes employed fo the analysis of the data fom Bass and von Camon [2004]. Again, the condition of inteest was contasted with null events. Subsequently, the IFJ peak was identified and the volume aound this peak was adjusted to coespond as closely as possible to 2,000 mm 3 (paticipant 13 was excluded fom this analysis because his IFJ activation was continuous with othe activations and, thus, it was not possible to define a volume coesponding to 2,000 mm 3 ). RESULTS A junction of the infeio pecental sulcus and the infeio fontal sulcus (IFJ) was identified in 13 of 14 hemisphees. In 5 hemisphees, the two sulci wee connected at thei fundi (Subjects 1, 2, 3, 9, 10), in 7 hemisphees the infeio pecental sulcus was slightly deepe than the infeio fontal sulcus (2 3 mm; Subjects 4, 7, 8, 11, 12, 13, 14), and in one hemisphee the infeio pecental sulcus was substantially deepe than the infeio fontal sulcus (Subject 5). Subject 6 had a small gyal bidge between the infeio pecental sulcus and the infeio fontal sulcus and, thus, no sulcal junction. When the z maps wee ovelaid on the coesponding anatomical data sets, we wee able to locate an activation at the left IFJ in 11 of 13 paticipants (Subjects 1 10, 13, 14). Two paticipants did not show an activation of the left IFJ. One of them (Subject 11) still had an activation in the left infeio pecental sulcus; it was, howeve, located futhe supeio, at the dosal-most pat of the infeio pecental sulcus. Anothe paticipant (Subject 12) had no activation of the left IFJ, but of the ight IFJ. Figue 1 and Supplementay Figue 1 display the IFJ activations on white-matte segmentations of individual bains. Figue 2 displays schematic dawings of these activations and Figue 3 depicts the individual peak locations and the outline of the 2,000 mm 3 activation volumes on 303

6 Defuss et al. with the ight-latealized activation, was 239, 2, 32. The standad deviations of the peak locations in x-, y-, and z- diection wee 66, 64, and 65 mm, espectively. The anges in x-, y-, and z-diection wee 22 mm, 15 mm, and 21 mm (12 mm in z-diection without Subject 11). An ovelap analysis showed that maximally five out of 13 z maps ovelapped in a given voxel (Fig. 4). Lateality Analysis Figue 2. Schematic views of IFJ activations. IFJ activations ae shown in gey; black lines indicate the fundi of the infeio pecental and infeio fontal sulci. Note: Double lines indicate tuncations of the infeio fontal sulcus. What is maked as infeio fontal sulcus in Subject 4 would be consideed as a pat of the infeio pecental sulcus following the scheme of Gemann et al. [2005]. The cicle at the junction of the infeio pecental sulcus and the infeio fontal sulcus in Subject 5 denotes a clea in-depth sepaation of these sulci. An activation of the ight IFJ could be identified in all of the 13 subjects included in the lateality analysis. The mean ight IFJ peak was located at 42, 2, 31. Fo absolute x values, the distance between the left and the ight IFJ peak was on aveage 8 mm. At the left IFJ, the pecent signal change diffeence between the meaning-switch condition and the cue-switch condition was on aveage 0.15% (SD 5 0.1%). A t test showed that the diffeences between these two conditions wee significantly diffeent fom zeo (t(12) 5 5.5, P < ). At the ight IFJ, the diffeence between the meaning-switch condition and the cue-switch condition was on aveage 0.07% (SD %). The pecent signal change diffeences at the ight IFJ wee not significantly diffeent fom zeo (t(12) 5 1.4, P < 0.18). A paied t test showed that the signal change diffeences on the left tended to be highe than on the ight (t(12) 5 2.1, P < 0.06). An inteesting finding was that the vaiability was consideably highe on the ight. In fact, fou paticipants (Subjects 4, 8, 10, and 14) showed highe pecent signal changes fo the cue-switch condition in the ight hemisphee. When these paticipants wee excluded, the mean pecent signal change diffeence on the ight was vey simila to the one on the left (M %, SD %). Reliability Analysis the individual anatomies of six paticipants (Table I gives the peak locations fo all paticipants). Oveall, the peak was located in the infeio pecental sulcus in 86% of paticipants. The peak was located in the most posteio pat of the infeio fontal sulcus in the emaining two paticipants (14%). Because the activations descibed ealie wee identified in contasts against null events, we also analyzed the timecouses of the meaning-switch, the cue-switch, and the null-event condition fo the IFJ peaks (Fig. 3). The meaning-switch condition showed a highe pecent signal change in all paticipants. The compaison of the Talaiach coodinates of the peak locations with the IFJ ange as oiginally defined based on goup studies (i.e., x between 230 and 247, y between 21 and 10, and z between 27 and 40) showed that the individual IFJ peaks wee located within this ange in 50% of paticipants. In six of the seven emaining paticipants, the IFJ peak was 4 mm o less outside this ange. The peak of Subject 11 was located 8 mm above the uppe z limit. The mean of the peak activations, excluding the paticipant Table II and Figue 5 display the esults of the eliability analysis. Mean IFJ peak distance between tasks aveaged ove all subjects was 7.4 mm. Fou subjects showed high eliability of the IFJ peak location with distances below 5.4 mm. Such distances appoximately coespond to a shift of 1 voxel diagonally (5.2 mm at a esolution). Two subjects (Subjects 1 and 2) had peak distances of 11.2 and 14.1 mm, espectively. A moe detailed inspection of these latte two subjects evealed that Subject 1 had a sub-peak within the IFJ activation volume in the taskswitching paadigm of Study 2 located at 232, 1, 27. This peak had a z scoe of 4.2 (in compaison to 4.5 of the main peak at 244, 1, 33). This sub-peak was vey simila to the peaks obseved in the othe two tasks. Using this peak fo the computation of the mean peak distance educes this distance fo Subject 1 to 5.2 mm. Similaly, Subject 2 had a sub-peak in the Stoop task of Study 3 at 233, 4, 33. This peak had a z scoe of 5.3 (in compaison to 6.1 of the peak at 242, 13, 27). The location of the sub-peak was identical to the location of the peak in Study 1. No futhe sub-peaks could be identified in Subjects 1 and

7 Single-Subject IFJ Activations Figue 3. Sagittal and axial slices though the IFJ peak fo six individuals. The activation peak is maked by a yellow squae, the volume of activation by a ed outline. On the ight, time-couse analyses fo the activation peaks ae shown. 305

8 Defuss et al. Figue

9 Single-Subject IFJ Activations TABLE I. Infeio fontal junction peaks and z thesholds fo all paticipants Subject IFJ peak z theshold 1 235, 22, , 4, , 22, , 2, , 23, , 6, , 0, , 1, , 1, , 24, , 0, a 37, 5, , 11, , 6, Mean 239, 2, 32 Note: The z thesholds wee adjusted such that the IFJ activation peaks wee suounded by a 2,000 mm 3 activation volume (see Methods). a Not included in the computation of the mean. suggest that the cytoachitectonic aeas esponsible fo the functional activations in the tasks investigated also bea consistent elationships to the sulci. In a cytoachitectonic study, Malikovic et al. [2006] found that hoc5/v5 had the highest pobability (in 65% of the hemisphees) of being located in the anteio occipital sulcus (the ascending limb of the infeio tempoal sulcus in the study of Dumoulin et al.). This localization coesponds well with the esults of Dumoulin et al. who found the anteio occipital sulcus most consistently activated (in 53% of the hemisphees). In both the cytoachitectonic study by Malikovic et al. and the imaging study by Dumoulin et al. V5 was less likely located in the sulci adjacent to the anteio occipital sulcus. Futhemoe, evidence has been povided that the fontal eye field as defined in high-esolution fmri [Rosano et al., 2002], chemoachitectonics [Rosano et al., 2003], and cytoachitectonics [Schmitt et al., 2005] is located in the supeio pecental sulcus. Consideation of the sub-peaks fo Subjects 1 and 2 educes the mean peak distance aveaged acoss subjects and tasks to 4 mm. DISCUSSION The pesent study investigated the elationship of fmri activations and sulcal mophology in the lateal fontal lobe on an individual-subject basis. Data fom a taskswitching study [Bass and von Camon, 2004] wee eanalyzed to assess the consistency of activations in the posteio fontolateal cotex and to detemine the elationship of these activations to individual mophologies of the infeio pecental sulcus and the infeio fontal sulcus. We found a junction of the infeio fontal sulcus and the infeio pecental sulcus (IFJ) in 13 of 14 bains. Twelve of these 13 paticipants (86%) had an activation that was located at the individually identified IFJ. Additional analyses fo a subset of these paticipants showed that IFJ peak activations wee stable acoss time. Taken togethe, in the paticipants included in ou study IFJ activations can be eliably identified based on the analysis of sulcal mophology. Stuctual Functional Relationships The esults of the pesent study indicate that posteio fontolateal activations in the task-switching paadigm in individual subjects bea a consistent elationship to individual sulcal mophology. These data supplement esults fom ealie studies in othe bain egions that have shown a similaly consistent stuctual functional coespondence fo tempoooccipital (V5; [Dumoulin et al., 2000; Watson et al., 1993]) and dosal fontal aeas (fontal eye fields and dosal pemoto cotex; [Amiez et al., 2006]). These esults Figue 4. Uppe panel: IFJ peak activations of all individuals (yellow squaes) ovelaid onto an individual bain in Talaiach space. Anatomical slices wee chosen to coespond to the mean peak location (oange squae). Lowe panel: Ovelap of IFJ activation volumes. The colo ba denotes the numbe of z maps ovelapping at a given voxel. Thee was a maximum ovelap of 5 z maps. 307

10 Defuss et al. TABLE II. IFJ activation peaks fo the six subjects who paticipated in multiple studies Subject Study 1 (task switching) Study 2 (task switching) Study 2 (Stoop) Study 3 (Stoop) Mean peak distance 1 235, 22, 30 44, 1, , 1, , 4, , 13, , 22, , 22, , 1, , 2, , 5, , 5, , 5, , 6, , 3, , 3, , 0, , 23, , 23, , 0, Note: Study 1 is Bass and von Camon [2004], Study 2 is Defuss et al. [2004], and Study 3 is Zysset et al. [2006]. These esults indicate that not only pimay (e.g., pimay visual o moto cotex), but also some non-pimay aeas like V5 and the fontal eye field ae consistently associated with paticula sulci. Such an association between cytoachitectue and sulcal mophology could be explained by tension-based models of mophogenesis [Van Essen, 1997; Hilgetag and Babas, 2006] which assume that white-matte development is a citical facto in detemining cotical folding. Following these models, it could be agued that aeas which fom o eceive connections ealy duing ontogeny tend to be those aeas whee the fist sulci will develop. Thus, a consistent colocalization of cytoachitectonic aeas and sulci might emege. It is inteesting to note in this context that the sulci found to include pimay aeas ae among the fist to develop in ontogeny. Fo example, Chi et al. [1977] found that the calcaine sulcus develops aound gestational week 16 and the cental sulcus aound week 20. Also, the infeio and supeio pecental sulci develop elatively ealy (week 24; [Chi et al., 1977]). In compaison, sulci delineating seconday fontal gyi develop aound weeks 32 35, and sulci delineating tetiay fontal gyi develop aound weeks [Chi et al., 1977]. Likely, thee will be less consistent associations with cytoachitectonic aeas fo the late-developing seconday and tetiay sulci (e.g., see the esults of Amunts et al. [1999] fo the diagonal sulcus). In the pesent study, a consistent elationship between sulcal mophology and functional imaging activations was obseved fo the junction of the infeio fontal sulcus and the infeio pecental sulcus. While the activation peaks wee vaiable in tems of thei Talaiach coodinates, they wee consistently found at the IFJ of individual paticipants. Activation volumes wee mostly confined to the dosal pat of the infeio pecental sulcus, but sometimes eached the most posteio pat of the infeio fontal sulcus and moe vental pats of the infeio pecental sulcus. This esult suggests that the cytoachitectonic aea undelying activations of the IFJ might be mainly located in the pat of the infeio pecental sulcus above the junction with the infeio fontal sulcus, with an extension into the infeio fontal sulcus o the vental pat of the infeio pecental sulcus in some bains. This is eminiscent of the esults fo V5 which was most consistently located in the anteio occipital sulcus but could also be found in sulci adjacent to the anteio occipital sulcus. Lateality Figue 5. IFJ activations fo the six subjects who paticipated in multiple studies. Volumes of IFJ activations ae indicated by coloed outlines and ae ovelaid onto the individual anatomies. Fo each subject, on the left a sagittal slice is shown and on the ight an axial slice. The sulcus abbeviations ae explained in Figue 3. A numbe of switching studies have epoted IFJ activations in the ight hemsiphee (fo eview, [see Defuss et al., 2005]). Howeve, the study by Bass and von Camon [2004] found only left-latealized IFJ activity. The pesent study shows that this esult was most likely due to a sub-goup of fou paticipants who had highe pe- 308

11 Single-Subject IFJ Activations cent signal change values fo the cue-switch condition in compaison with the meaning-switch condition in the ight hemisphee. It is cuently unclea whethe this might be a esult of the specific stategy employed by these paticipants o might eflect a diffeent cotical oganization in these paticipants. Also, it is an open question whethe this patten of esults would be eplicable. Thus, the cuent esults offe some insights into why a left-latealization was found by Bass and von Camon [2004], but clealy futhe studies will be necessay to bette undestand the functional ole of the ight IFJ. Reliability The eliability of IFJ activations was good. When subpeaks wee consideed in two subjects who showed less consistent peak locations, mean peak distance could be educed to 4 mm. This esult is emakable consideing that measuements wee sepaated by at least 4 months (up to 19 months fo the subjects who paticipated in Study 1 and Study 3) and that diffeent tasks wee employed (vaiants of the task-switching paadigm and the Stoop task). Thus, ou esults ae in line with anothe study that povided evidence fo a high long-tem eliability of fmri data [Aon et al., 2006]. A limitation of ou analysis is that conditions wee contasted against null events. We chose this way to analyze the data as the event-elated designs employed in the studies analyzed hee did not have enough powe to show between-condition diffeences fo all paticipants (in paticula, Study 2 employed event-elated designs with task lengths of only aound 15 min). Howeve, we inspected timelines to ascetain that the IFJ activations wee geneally not solely due to low-level pocesses but wee modulated by expeimental conditions. Nevetheless, futue single-subject studies pefeably employing blocked designs might be useful to povide evidence fo the eliability of IFJ activations in contasts against highe-level baselines. Limits of Localization When one consides the localization of activations in fmri studies in elation to anatomical stuctues, the question of localization accuacy aises. The within-plane esolution of the EPI sequence in the pesent study was 3 mm, epesenting the uppe limit of spatial esolution. In addition, thee ae othe factos which ae likely to have had an influence on the accuacy of the localization of activations. One such facto is that with gadient-echo sequences the stongest signal changes ae often obseved in daining veins, not in neual tissue [e.g., Lai et al., 1993]. Tune [2002: p. 1062] estimated that an activated cotical aea of 100 mm 2 will geneate an oxygenation change in venous blood that extends without dilution along the vein no moe than 4.2 mm beyond the edge of the activated aea. At geate distances along the daining vein this oxygenation change will be diluted. In a study compaing gadient-echo BOLD at 1.5 T with electophysiological ecodings in the macaque, Disbow et al. [2000] found that the maximal mean displacement of the BOLD cente of volume was 9.5 mm in a given diection of space. In addition, one has to conside that the egistation of the functional data to the anatomical data will be less than pefect and that this will intoduce additional small eos in localization. Taken togethe, in the pesent study, the active neual tissue might be up to 2 o 3 voxels away fom the obseved activation peak. Given this spatial esolution, it appeas save to conclude that the activation volumes in ou study wee located at the IFJ. Howeve, we would geneally conside it poblematic to daw conclusions about whethe the posteio o the anteio bank of the pecental sulcus was activated in a given paticipant. CONCLUSION Based on goup studies, we have peviously agued that the junction egion of the infeio fontal sulcus and the infeio pecental sulcus, the IFJ, is consistently involved in updating pocesses [Bass and von Camon, 2002, 2004; Defuss et al., 2004]. Hee, we have shown that posteio fontolateal activations in a task-switching paadigm ae consistently and eliably located at the IFJ of individual paticipants. Thus, the IFJ is suggested to be simila to the supeio pecental sulcus in that it indicates the location of a functionally defined aea. This stuctue function coespondence could be explained by tension-based models of mophogenesis. In addition, peliminay esults fom cytoachitectonic analyses [Amunts and von Camon, 2006] suggest that the functionally defined IFJ egion might have a micostuctual coespondent not labeled in classical cytoachitectonic maps like that of Bodmann [1909]. Stuctue function elationships like the one obseved in the pesent study become inceasingly elevant in the context of tanscanial magnetic stimulation studies that use stuctual MRI scans to identify stimulation sites. ACKNOWLEDGMENTS We would like to thank Stefan Zysset and Deek Ott fo thei help in the ealization of this study. REFERENCES Aguie GK, Zaahn E, D Esposito M (1997): Empiical analyses of BOLD fmri statistics. II. Spatially smoothed data collected unde null-hypothesis and expeimental conditions. Neuoimage 5(3): Amiez C, Kostopoulos P, Champod AS, Petides M (2006): Local mophology pedicts functional oganization of the dosal pemoto egion in the human bain. J Neuosci 26(10): Amunts K, von Camon DY (2006): The anatomical segegation of the fontal cotex: What does it mean fo function? Cotex 42(4):

12 Defuss et al. Amunts K, Schleiche A, Bugel U, Mohlbeg H, Uylings HB, Zilles K (1999): Boca s egion evisited: Cytoachitectue and intesubject vaiability. J Comp Neuol 412(2): Amunts K, Malikovic A, Mohlbeg H, Schomann T, Zilles K (2000): Bodmann s aeas 17 and 18 bought into steeotaxic space-whee and how vaiable? Neuoimage 11(1): Aon AR, Gluck MA, Poldack RA (2006): Long-tem test-etest eliability of functional MRI in a classification leaning task. Neuoimage 29(3): Bass M, von Camon DY (2002): The ole of the fontal cotex in task pepaation. Ceeb Cotex 12(9): Bass M, von Camon DY (2004): Decomposing components of task pepaation with functional MRI. J Cogn Neuosci 16(4): Bass M, Defuss J, Fostmann B, von Camon DY (2005): The ole of the infeio fontal junction aea in cognitive contol. Tends Cogn Sci 9(7): Bidge H, Clae S, Jenkinson M, Jezzad P, Pake AJ, Matthews PM (2005): Independent anatomical and functional measues of the V1/V2 bounday in human visual cotex. J Vis 5(2): Bodmann K (1909): Vegleichende Lokalisationslehe de Goßhininde in ihen Pinzipien dagestellt auf Gund des Zellenbaues. Leipzig: Bath. Chi JG, Dooling EC, Gilles FH (1977): Gyal development of the human bain. Ann Neuol 1(1): Cunningham DJ (1892): Contibution to the Suface Anatomy of the Ceebal Hemisphees (Cunningham Memois, No. VII). Dublin: Royal Iish Academy. Dechent P, Fahm J (2003): Functional somatotopy of finge epesentations in human pimay moto cotex. Hum Bain Mapp 18(4): Defuss J, Bass M, von Camon DY (2004): Cognitive contol in the posteio fontolateal cotex: Evidence fom common activations in task coodination, intefeence contol, and woking memoy. Neuoimage 23(2): Defuss J, Bass M, Neumann J, von Camon DY (2005): Involvement of the infeio fontal junction in cognitive contol: Metaanalyses of switching and Stoop studies. Hum Bain Mapp 25: Devlin JT, Poldack RA (2007): In paise of tedious neuoanatomy. Neuoimage 37(4): DeYoe EA, Caman GJ, Bandettini P, Glickman S, Wiese J, Cox R, Mille D, Neitz J (1996): Mapping stiate and extastiate visual aeas in human ceebal cotex. Poc Natl Acad Sci USA 93(6): Disbow EA, Slutsky DA, Robets TP, Kubitze LA (2000): Functional MRI at 1.5 tesla: A compaison of the blood oxygenation level-dependent signal and electophysiology. Poc Natl Acad Sci USA 97(17): Dumoulin SO, Bitta RG, Kabani NJ, Bake CL, Le Goualhe G, Pike GB, Evans AC (2000): A new anatomical landmak fo eliable identification of human aea V5/MT: A quantitative analysis of sulcal pattening. Ceeb Cotex 10(5): Ebeling U, Steinmetz H, Huang YX, Kahn T (1989): Topogaphy and identification of the infeio pecental sulcus in MR imaging. Am J Roentgenol 153(5): Ebestalle O (1890): Das Stinhin. Ein Beitag zu Anatomie de Obefläche des Gosshins. Wien & Leipzig: Uban & Schwazenbeg. Fadiga L (2007): Functional magnetic esonance imaging: Measuing vesus estimating. Neuoimage 37(4): Fiston KJ, Holmes AP, Wosley KJ, Poline J-P, Fith CD, Fackowiak RSJ (1995): Statistical paametic maps in functional imaging: A geneal linea appoach. Hum Bain Mapp 2: Fiston KJ, Fletche P, Josephs O, Holmes A, Rugg MD, Tune R (1998): Event-elated fmri: chaacteizing diffeential esponses. Neuoimage 7(1): Gemann J, Robbins S, Halsband U, Petides M (2005): Pecental sulcal complex of the human bain: Mophology and statistical pobability maps. J Comp Neuol 493(3): Geye S, Ledbeg A, Schleiche A, Kinomua S, Schomann T, Bugel U, Klingbeg T, Lasson J, Zilles K, Roland PE (1996): Two diffeent aeas within the pimay moto cotex of man. Natue 382(6594): Geye S, Schleiche A, Zilles K (1999): Aeas 3a, 3b, and 1 of human pimay somatosensoy cotex. Neuoimage 10(1): Hilgetag CC, Babas H (2006): Role of mechanical factos in the mophology of the pimate ceebal cotex. PLoS Comput Biol 2(3):e22. doi: /jounal.pcbi Holmes CJ, Hoge R, Collins L, Woods R, Toga AW, Evans AC. (1998): Enhancement of MR images using egistation fo signal aveaging. J Comput Assist Tomog 22(2): Huttne HB, Lohmann G, von Camon DY (2005): Magnetic esonance imaging of the human fontal cotex eveals diffeential anteio-posteio vaiability of sulcal basins. Neuoimage 25(2): Juch H, Zimine I, Seghie ML, Lazeyas F, Fasel JH (2005): Anatomical vaiability of the lateal fontal lobe suface: Implication fo intesubject vaiability in language neuoimaging. Neuoimage 24(2): Lai S, Hopkins AL, Haacke EM, Li D, Wasseman BA, Buckley P, Fiedman L, Meltze H, Hedea P, Fiedland R (1993): Identification of vascula stuctues as a majo souce of signal contast in high esolution 2D and 3D functional activation imaging of the moto cotex at 1.5T: Peliminay esults. Magn Reson Med 30(3): Lohmann G, von Camon DY, Steinmetz H (1999): Sulcal vaiability of twins. Ceeb Cotex 9(7): Lohmann G, Mulle K, Bosch V, Mentzel H, Hessle S, Chen L, Zysset S, von Camon DY (2001): LIPSIA A new softwae system fo the evaluation of functional magnetic esonance images of the human bain. Comput Med Imaging Gaph 25(6): Malikovic A, Amunts K, Schleiche A, Mohlbeg H, Eickhoff SB, Wilms M, Palomeo-Gallaghe N, Amstong E, Zilles K (2006): Cytoachitectonic analysis of the human extastiate cotex in the egion of V5/MT1: A pobabilistic, steeotaxic map of aea hoc5. Ceeb Cotex 17(3): Oldfield RC (1971): Assessment and analysis of handedness Edinbugh Inventoy. Neuopsychologia 9(1): Ono M, Kubik S, Abenathey CD (1990): Atlas of the Ceebal Sulci. Stuttgat: Geog Thieme Velag. Rademache J, Caviness VS J, Steinmetz H, Galabuda AM (1993): Topogaphical vaiation of the human pimay cotices: Implications fo neuoimaging, bain mapping, and neuobiology. Ceeb Cotex 3(4): Rajkowska G, Goldman-Rakic PS (1995): Cytoachitectonic definition of pefontal aeas in the nomal human cotex. II. Vaiability in locations of aeas 9 and 46 and elationship to the Talaiach Coodinate System. Ceeb Cotex 5(4): Rosano C, Kisky CM, Welling JS, Eddy WF, Luna B, Thulbon KR, Sweeney JA (2002): Pusuit and saccadic eye movement subegions in human fontal eye field: A high-esolution fmri investigation. Ceeb Cotex 12(2): Rosano C, Sweeney JA, Melchitzky DS, Lewis DA (2003): The human pecental sulcus: chemoachitectue of a egion coesponding to the fontal eye fields. Bain Res 972(1 2):

13 Single-Subject IFJ Activations Schmitt O, Modesitzki J, Heldmann S, Witz S, Hömke L, Heide W, Kömpf D, Wee A (2005): Thee-dimensional cytoachitectonic analysis of the posteio bank of the human pecental sulcus. Anat Embyol 210(5 6): Seeno MI, Dale AM, Reppas JB, Kwong KK, Belliveau JW, Bady TJ, Rosen BR, Tootell RB (1995): Bodes of multiple visual aeas in humans evealed by functional magnetic esonance imaging. Science 268(5212): Sudevan P, Taylo DA (1987): The cueing and piming of cognitive opeations. J Exp Psychol Hum Pecept Pefom 13(1): Talaiach J, Tounoux P (1988): Co-Plana Steeotaxic Atlas of the Human Bain. Stuttgat: Geog Thieme Velag. Toga AW, Thompson PM, Moi S, Amunts K, Zilles K (2006): Towads multimodal atlases of the human bain. Nat Rev Neuosci 7(12): Tootell RB, Hadjikhani NK, Vanduffel W, Liu AK, Mendola JD, Seeno MI, Dale AM (1998): Functional analysis of pimay visual cotex (V1) in humans. Poc Natl Acad Sci USA 95(3): Tune R (2002): How much cotex can a vein dain? Downsteam dilution of activation-elated ceebal blood oxygenation changes. Neuoimage 16(4): Tzouio-Mazoye N, Hevé PY, Mazoye B (2007): Neuoanatomy: Tool fo functional localization, key to bain oganisation. Neuoimage 37(4): Van Essen DC (1997): A tension-based theoy of mophogenesis and compact wiing in the cental nevous system. Natue 385(6614): Waltes NB, Egan GF, Kil JJ, Kean M, Waley P, Jenkinson M, Watson JD (2003): In vivo identification of human cotical aeas using high-esolution MRI: An appoach to ceebal stuctuefunction coelation. Poc Natl Acad Sci USA 100(5): Waltes NB, Eickhoff SB, Schleiche A, Zilles K, Amunts K, Egan GF, Watson JD (2007): Obseve-independent analysis of highesolution MR images of the human ceebal cotex: In vivo delineation of cotical aeas. Hum Bain Mapp 28(1):1 8. Watson JD, Myes R, Fackowiak RS, Hajnal JV, Woods RP, Mazziotta JC, Shipp S, Zeki S (1993): Aea V5 of the human bain: Evidence fom a combined study using positon emission tomogaphy and magnetic esonance imaging. Ceeb Cotex 3(2): Wosley KJ, Fiston KJ (1995): Analysis of fmri time-seies evisited Again. Neuoimage 2(3): Yousy TA, Schmid UD, Alkadhi H, Schmidt D, Peaud A, Buettne A, Winkle P (1997): Localization of the moto hand aea to a knob on the pecental gyus. A new landmak. Bain 120(Pat 1): Zaahn E, Aguie GK, D Esposito M (1997): Empiical analyses of BOLD fmri statistics. I. Spatially unsmoothed data collected unde null-hypothesis conditions. Neuoimage 5(3): Zysset S, Schoete ML, Neumann J, von Camon DY (2006): Stoop intefeence, hemodynamic esponse and aging: An event-elated fmri study. Neuobiol Aging 28(6):