The use of the orbitofrontal H-sulcus as a reference frame for value signals

Abstract

Understanding the factors that drive organization and function of the brain is an enduring question in neuroscience. Using functional magnetic resonance imaging (fMRI), structure and function have been mapped in primary sensory cortices based on knowledge of the organizational principles that likely drive a given region (e.g., aspects of visual form in primary visual cortex and sound frequency in primary auditory cortex) and knowledge of underlying cytoarchitecture. The organizing principles of higher-order brain areas that encode more complex signals, such as the orbitofrontal cortex (OFC), are less well understood. One fundamental component that underlies the many functions of the OFC is the ability to compute the reward or value of a given object. There is evidence of variability in the spatial location of responses to specific categories of objects (or value of said objects) within the OFC, and several reference frames have been proposed to explain this variability, including topographic spatial gradients that correspond to axes of primary versus secondary rewards and positive versus negative reinforcers. One potentially useful structural morphometric reference frame in the OFC is the “H-sulcus,” a pattern formed by medial orbital, lateral orbital and transverse orbital sulci. In 48 human subjects, we use a structural morphometric tracing procedure to localize functional activation along the H-sulcus for face and food stimuli. We report the novel finding that food-selective responses are consistently found within the caudal portion of the medial orbital sulcus, but no consistency within the H-sulcus for response to face stimuli. These results suggest that sulcogyral anatomy of the H-sulcus may be an important morphological metric that contributes to the organizing principles of the OFC response to certain stimulus categories, including food.

1 |. INTRODUCTION

Estimating the value of objects in our environment is thought to be important for learning, facilitating decision-making processes and promoting behaviors that lead to increased evolutionary fitness. Processing the value of objects is associated with activation in the mesolimbic reward system, including the nucleus accumbens, amygdala and orbitofrontal cortex (OFC), amongst other regions (Gottfried, O’Doherty, & Dolan, 2003; Holland & Gallagher, 2004; Knutson, Adams, Fong, & Hommer, 2001; Knutson, Rick, Wimmer, Prelec, & Loewenstein, 2007; O’Doherty, Buchanan, Seymour, & Dolan, 2006; Walton, Devlin, & Rushworth, 2004). The OFC, more specifically, has been implicated in a wide range of processes related to goal-directed behavior, value estimation and reward responsivity (for review, see Kringelbach, 2005).

Response in the OFC to stimuli with value or rewarding properties has been studied using functional magnetic resonance imaging (fMRI). This work has identified greater magnitude of responses in the OFC to categories of stimuli that are high in value, including food, sexual stimuli and human faces (Aharon et al., 2001; Costumero et al., 2014; Frank et al., 2010; Ishai, 2007; Siep et al., 2009; Troiani, Dougherty, Michael, & Olson, 2016). There is evidence that the OFC computes a domain-general value computation across multiple categories of rewarding stimuli (Bartra, McGuire, & Kable, 2013; Chib, Rangel, Shimojo, & O’Doherty, 2009; Kim, Shimojo, & O’Doherty, 2011; Levy & Glimcher, 2011, 2012), including primary rewards such as erotic stimuli and secondary or learned rewards like money. However, there is also evidence of localized, domain-specific activation within the OFC in both humans and non-human primates for items with value. For example, Tsao, Moeller, and Freiwald (2008) assessed fMRI response in monkeys while they viewed various types of objects and found face-selective activation in frontal cortex, including the OFC. We previously characterized these OFC “face-selective patches” in humans and showed that they are primarily located in medial OFC, near the midline of the brain (Troiani et al., 2016).

Although the bulk of human neuroimaging work in the domain of reward or stimulus value has focused on the functional response within the OFC, there is also substantial structural variation in the sulcogyral anatomy in this region of cortex (Chiavaras & Petrides, 2000; Ono, Kubik, & Abernathey, 1990). For example, there is tremendous variation between subjects in the spatial location and length of the medial, lateral, intermediate, transverse and posterior sulci that comprise the OFC. Further, there is variation in the numbers of sulci present in the OFC, with some people having 1–2 intermediate and/or 1–2 posterior sulci (Bartholomeusz et al., 2013). Variability in the number of sulci in the OFC necessitates manual characterization and labeling procedures, in which specific OFC sulci can only be identified based on their position relative to other sulci. Thus, the spatial location of the majority of OFC sulci and potential overlap with functional activation can only be assessed by considering structural anatomy on a subject by subject basis.

It is widely accepted that structural variation throughout the brain can impact functional organization. This has been found using functional and structural magnetic resonance imaging in many regions of cortex, including auditory (Peelle, Troiani, Grossman, & Wingfield, 2011), early visual (Benson et al., 2012) and higher-level visual cortices (Bi, Chen, Zhou, He, & Fang, 2014; Saygin et al., 2012; Weiner, Natu, & Grill-Spector, 2018). In the field of human visual neuroscience, the large differences in spatial location of activation across subjects for specific stimulus categories have driven fMRI analysis methods to be largely completed using functional localizers, based on individual subject’s functional responses that are constricted by structural anatomical boundaries (Nieto-Castañón & Fedorenko, 2012; Saxe, Brett, & Kanwisher, 2006).

Although functional localizers are not typically done in the OFC, the concept has been borrowed for regions outside of visual cortex for domains such as language (Fedorenko, Hsieh, Nieto-Castañón, Whitfield-Gabrieli, & Kanwisher, 2010). Thus, there is precedent for using individual functional nodes and assessing their relationship to subject-specific anatomy. It has even been suggested that using an individual’s own anatomy should be the gold standard (Devlin & Poldrack, 2007). Here, we assess whether individually localized functional activations can be described in a useful way using individual OFC sulcogyral anatomy as a reference frame. One previous paper found that sulcogyral labeling of the H-sulcus of the OFC allowed for more precise localization of value signals for erotic and monetary rewards (Li, Sescousse, Amiez, & Dreher, 2015) along the medial orbital sulcus. The motivation for previous work by Li et al. was to assess whether the spatial location of sexual and monetary rewards along a rostral–caudal gradient was consistent with a theoretically proposed rostral–caudal gradient for secondary–primary rewards. Li et al. found evidence for this gradient, with secondary rewards located within the rostral MOS and more primary rewards located within the caudal MOS. We and others have identified face value signals (frequently referred to as “face patches”) primarily along the midline of the OFC and thus not conforming to a rostral–caudal axis along the MOS (Troiani et al., 2016), although no previous work has specifically localized activations of faces and food along the MOS and other OFC sulci that comprise the H-sulcus.

To date, no one has spatially localized food-selective responses at the individual subject level within the OFC, although recent work has identified unique portions of the OFC that respond to various macronutrients within the same category of stimuli at the group level (i.e., food; Suzuki, Cross, & O’Doherty, 2017). Thus, the question of whether food value is represented in the same or different subregions of the OFC remains open. Therefore, we sought to establish whether there is a consistent location of food signals within the OFC sulci and to assess whether this differs from the location of face value signals (or “face patches”). Our theoretical motivation was not to establish or confirm a rostral–caudal gradient for these value signals, because existing evidence suggests that face signals are more consistently located along the midline of the OFC. Rather, we seek to establish whether there are consistencies in the sulcogyral location of food value signals in a typical population, in order to lay the groundwork for future studies of patients with brain disorders that may disrupt value of various object categories (e.g., autism, social anxiety, and eating disorders).

2 |. MATERIALS AND METHODS

This is a re-analysis of functional imaging data reported previously (Adamson & Troiani, 2018). Structural morphological characterizations and analyses and subject-level localization of functional data within the H-sulcus of the OFC have not been reported elsewhere.

2.1 |. Participants

A group of 48 healthy subjects (24 females) were included in this analysis. Twenty of the individuals were collected at Temple University Medical Center, and 28 were collected at the Autism & Developmental Medicine Institute at Geisinger Medical Center. Ages ranged from 19 to 31 years old (22.3 + 3.18 years). Education levels ranged from 13 to 18 years of education (16.0 + 1.4 years). None of the subjects had a history of self-reported neurological or psychiatric disorders. Written informed consent was obtained according to the guidelines of the Institutional Review Board of Temple University and Geisinger Medical Center. Participants received monetary compensation for participation in the experiment. One additional subject (not included in N = 48 group analysis) completed a longer version of the task during two scanner sessions, separated by 1 week. This subject was used to assess the reliability of the spatial location of value signals across time.

This analysis is a segment of a larger study that collected a broad neuroimaging battery, in-person phenotyping and online questionnaires. Our main interest and hypothesis was surrounding the identification of food value signal localization. However, we include summary phenotypic metrics in order to provide a more complete description of these subjects to allow for comparisons to new cohorts in any potential replications of this finding and/or to facilitate meta-analyses.

2.2 |. Phenotype metrics

1. Body mass index (BMI) is a measure that reflects the ratio of metric weight to height. These values were self-reported during an in-person testing session and then converted to BMI using an online calculator (https://www.nhlbi.nih.gov/health/educational/lose_wt/BMI/bmicalc.htm). The average BMI for the cohort was 24.3 ± 3.9, with a range from 17.8 to 35.4. Average BMI of males in the cohort was 25.4 and average BMI of females was 23.1.

2. The Broad Autism Phenotype Questionnaire (BAPQ; Hurley, Losh, Parlier, Reznick, & Piven, 2007) was collected from all subjects as a measure of subclinical autism traits. Originally created to identify subclinical traits in parents of children with autism spectrum disorder, the BAP has since been used as a quantitative measure of identifying subclinical traits within a normally distributed population (DiCriscio & Troiani, 2017; Troiani et al., 2016). BAP Total scores ranged from 2 to 3.8 in this population, with a mean of 2.7. Using normative cutoff scores based on a large community sample (Sasson et al., 2013) male cutoff 3.47; female cutoff 3.19), four individuals scored above these cutoff values (two males, two females).

3. The Cambridge Face Memory Test (CFMT; Duchaine & Nakayama, 2006) is a freely available, 72-item assessment that measures face processing abilities with respect to face memory. CFMT was completed by all participants, with total scores ranging from 56.9% to 97.2%, and a mean of 79.52 (±SD = 11.6). Mean score for females was 78.56 (±SD = 11.9), and mean score for males was 80.44 (±SD = 11.6).

2.3 |. Image acquisition

MRI scanning was conducted at both Temple University Hospital and the Geisinger Autism & Developmental Medicine Institute. Scanning at Temple University was conducted on a 3.0 T Siemens Verio scanner using a Siemens 12-channel phased-array head coil. High-resolution anatomical images (T1-weighted 3D MPRAGE) were also collected for each participant with the following parameters: 160 axial slices, 1 mm slice thickness, TR = 1,900 ms, TE = 2.93 ms, inversion time = 900 ms, flip angle = 9°, FOV = 256 mm and voxel size = 1 × 1 × 1 mm. Scanning at Geisinger was conducted on a 3.0 T Siemens Magnetom scanner using a Siemens thirty-two-channel phased-array. Sequence parameters were identical at both sites.

Functional data consisted of one 8-min run of whole-brain T2* weighted BOLD echoplanar images with 142 volumes acquired per run (61 oblique axial slices, 2.5 mm slice thickness, voxel size = 3 × 3 × 2.5 mm; matrix size = 80 × 80; TR = 3,000 ms, TE = 20 ms, flip angle = 90 degrees and GRAPPA = 2. We optimized for signal in the OFC by tilting slice acquisition −30 degrees from the AC-PC plane reduce signal dropout (Deichmann, Gottfried, Hutton, & Turner, 2003).

2.4 |. Experimental task: Face food localizer

We used a modified object localizer task described in detail elsewhere (Adamson & Troiani, 2018; Troiani et al., 2016). Briefly, images of attractive faces (including famous and unknown identities), appetizing food, scenes and clocks were presented in a block design. Images appeared for 500 ms, with a 250 ms interstimulus interval, and were arranged into superblocks, with stimulus category order randomized along with rest trials over the course of 142 image acquisitions (426 s). Face stimuli included half male and half female faces. Participants responded to image repeats that appeared within each block, in order to insure participants were awake and paying attention to the images. Although the original dataset from Temple University included four runs of the experimental task, only the first run was included in this analysis, in order to match subjects from both sites on number of runs.

2.5 |. Data analysis

The data analysis consisted of three parts: (a) group analysis of face- and food-selective regions in the OFC, (b) subject-level identification and localization of individual face and food value signals within the OFC and (c) tracing of individual OFC sulcal pattern anatomy from subject-specific T1 images.

2.6 |. Functional analysis: Preprocessing

Images were first converted from dicom to nifti format using MRIConvert (http://lcni.uoregon.edu/downloads/mriconvert ), and preprocessing of the images was then completed using the FMRIB Software Library (FSL, www.fmrib.ox.ac.uk/fsl/). All runs were estimated for movement. Movement across participants averaged 0.21 mm (±0.14 mm).

Additional preprocessing steps included stripping nonbrain material using the Brain Extraction Tool (BET), motion correction, B0 unwarping and slice time correction with FSL FEAT (fMRI Expert Analysis Tool) version 5.0.8. Images were normalized to 2 mm space via MCFLIRT and smoothed using a 5 mm Gaussian kernel.

2.7 |. Functional analysis: Modeling

Whole-brain analyses were implemented using a standard linear modeling approach. These models included four categorical regressors indicating whether the stimulus for each block was a face, place, food or clock. Categorical regressors were boxcar functions at stimulus onset convolved with a canonical hemodynamic response function. Six estimated motion parameters were also included as nuisance regressors. We performed group-level analysis by submitting individual whole-brain maps for contrasts of interest into a second-level analysis of covariance (ANCOVA) for the two contrasts of interest: Faces compared to place and clock stimuli and food stimuli compared to place and clock stimuli. Because the data were collected from two different sites, using two different scanners and head coils, we also accounted for any differences in acquisition due to scanner type or head coil by including scanner site as a covariate in these analyses. The inclusion of scanner site as a covariate accounts for any large differences in temporal signal to noise ratio (tSNR) between the sites in group-level results. To correct for multiple comparisons, we used threshold-free cluster enhancement (TFCE) (Smith & Nichols, 2009), which determines statistical significance using permutation labeling, with the α level set at p < .001.

2.8 |. Functional analysis: Signal to noise ratio

We also computed the tSNR for each individual within the primary region of interest here (the OFC), in order to assess for the influence of tSNR on our ability to identify value patches. The tSNR was computed by taking the functional run for each participant and dividing the mean of the time series by the residual error standard deviation after preprocessing. tSNR values within the OFC were averaged by using a mask derived from OFC regions in the AA2 atlas (Rolls, Joliot, & Tzourio-Mazoyer, 2015). We did find that in the OFC, the tSNR was greater overall for subjects scanned at the Geisinger site using the 32-channel coil (160 ± 37) than for subjects scanned at the Temple site using the 12-channel coil (123 ± 25) (independent samples t test; t(46) = 3.92, p < .001). Although we do not anticipate that this difference will influence our results, we explore the influence of tSNR on our ability to characterize value signals in the Statistical Comparison of Value Signals section of the Data Analysis section, below.

2.9 |. Functional analysis: Value signal identification using subject’s anatomy

Despite the large degree of variability in the OFC sulci across subjects, a pattern formed by the intersection of the lateral, medial and transverse orbital sulci can be labeled according to the continuity of the rostral and caudal portions of the medial and lateral orbitofrontal sulci (MOS and LOS, respectively) and classified into one of three distinct pattern types, first described by Chiavaras and Petrides (2000). The most common pattern (Type I) consisted of a discontinuous MOS and continuous LOS, with fewer individuals having a Type II pattern that consisted of a continuous MOS and LOS. The pattern observed with the least frequency was the Type III pattern, which consisted of a discontinuous MOS and LOS. While Chiavaras and Petrides (2000) earliest characterizations were on a group of people free from psychiatric illness, subsequent studies have associated the less frequent pattern types (Type II and Type III) with psychiatric disorders (Bartholomeusz et al., 2013; Chakirova et al., 2010; Chye et al., 2017; Lavoie et al., 2014; Nakamura et al., 2007; Patti & Troiani, 2018; Takayanagi et al., 2010; Watanabe et al., 2013) and subclinical traits associated with psychiatric disorders (Whittle et al., 2012; Zhang, Harris, Split, Troiani, & Olson, 2016).

Although the OFC characterization procedure developed by Chiavaras and Petrides established a number of “global” patterns formed by the H-sulcus (i.e., Type I, Type II and Type III), the tracing procedure necessary to characterize these patterns can be useful merely for sulcal definition within the OFC. That is, given tremendous variability in the number, depth, length and precise spatial location of OFC sulci across individuals, tracing individual sulci within a given subject’s brain is necessary to appropriately be able to label OFC sulci. One previous paper by Li et al. summarized the global pattern type of the H-sulcus (Type I, Type II and Type III) across their sample, but did not describe any differences in the location of value signals based on H-sulcus pattern type. Therefore, we focus on utilizing the tracing procedure in order to accurately identify any correspondence between a value signal and a given subject’s sulcal or gyral location. We also report individual results in the supplement and designate each subject’s H-sulcal pattern type for completeness.

Preprocessed images from first-level analyses described above were used to estimate value signals. We examined whole-brain maps for each individual subject, in order to characterize the location of face and food value signals in the OFC utilizing the cluster function within FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Cluster). We chose to use the cluster tool (rather than identifying peak voxels, for instance) because an indicator of the extent of activation is meaningful for characterizing whether the activation truly lies within a sulcus. Further, we were only interested in identifying clusters within the OFC and not the rest of the brain; thus, we restricted boundaries to the orbitofrontal region for each subject by inclusively masking their statistical maps for the contrast of Faces > Places & Clocks and Food > Places & Clocks using on orbitofrontal mask by combining all orbital regions from the AAL2 atlas (medial orbital gyri (MOG), anterior orbital gyri (AOG), posterior orbital gyri (POG), lateral orbital gyri (LOG); (Rolls et al., 2015)). Images were then set to a minimum t-score threshold of 2.0, and clusters within the orbitofrontal region were identified using the cluster tool in FSL with a minimum cluster threshold of 10 contiguous voxels for each subject. Clusters surviving this threshold were then output into 3D maps for each subject, and relevant peak coordinate information was recorded. Clusters were then viewed on each individual’s T1 image, which had been transformed into standard space. Location of each cluster was then recorded with respect to their location within or surrounding the H-sulcus. This analysis on each subject’s map allows for individual clusters to be identified using cluster-wise inference.

All clusters within the OFC were identified and labeled according to sulcal or gyral location as (a) rostral medial orbital sulcus; MOSr, (b) caudal medial orbital sulcus; MOSc, (c) rostral lateral orbital sulcus; LOSr, (d) caudal lateral orbital sulcus; LOSc, (e) transverse sulcus; TOS, (f) posterior orbital sulcus; POS (g) medial orbital gyrus; MOG (h) lateral orbital gyrus; LOG and (i) posterior orbital gyrus; POG, for each hemisphere. Clusters along the midline of the brain were characterized as lying within the (j) gyrus rectus, (k) olfactory orbital sulcus, (l) supraorbital sulcus or (m) subcallosal sulcus. All sulci are labeled and identified for each subject and included in Supplementary Tables. To summarize data for the purposes of analysis, we include only those clusters lying in sulcal locations in Table 2. Because clusters within the olfactory orbital sulcus and supraorbital sulcus tended to span the midline of the brain, these were labeled as medial, rather than designating a specific hemisphere. We then characterized the value signals within the sulci of the H-sulcus (those in the MOS, LOS, and/or TOS). Cluster location was determined by consensus review by both VT and MP. The majority of clusters were quite easy to label as a particular sulcal or gyral location, but there were a few large clusters that extended over 100 voxels that extended across a sulcal–gyral boundary. When large clusters extended over a sulcal–gyral boundary, the cluster was labeled according to the location which described the majority of the activation cluster.

TABLE 2.

Number of subjects with face and food value signal clusters in each anatomical location based on MNI-152 template anatomy

Medial N (%)		H-sulcus N (%)
		MOSr		MOSc		LOSr		LOSc		TOS
Face	Food	Face	Food	Face	Food	Face	Food	Face	Food	Face	Food
36 (75)	20 (42)	1 (2)	2 (4)	6 (13)	11 (23)	0 (0)	0 (0)	4 (8)	3 (6)	4 (8)	9 (19)
χ2 = 10.97, p = .0009		χ2 = 0.334, p = .557		χ2 = 1.787, p = .181		N/A		χ2 = 0.154, p = .695		χ2 = 2.22, p = .135

Abbreviations: LOSc, caudal lateral orbital sulcus; LOSr, rostral lateral orbital sulcus; MOSc, caudal medial orbital sulcus; MOSr, rostral medial orbital sulcus; TOS, transverse orbital sulcus.

2.10 |. Functional analysis: Value signal identification using template anatomy

In order to address whether there is utility to a localization procedure using individual sulcogyral anatomy as compared to a group average brain, we also characterized each value signal using the sulcogyral anatomy depicted in the MNI-152 1 mm template brain (Mazziotta et al., 2001). To do this, the MNI-152 1 mm template brain was traced to identify orbitofrontal sulci using ITK-SNAP. Following, all signals identified using the procedure outlined above using individual subject anatomical scans were instead overlaid onto the group average brain in order to determine sulcogyral location. We then assessed whether the sulcogyral location would be characterized differently when using the individual subject’s anatomy versus the group average anatomy.

2.11 |. Sulcal tracing: Preprocessing

To analyze the OFC sulcal patterns, T1 images were first converted from dicom to nifti format using MRIConvert (http://lcni.uoregon.edu/downloads/mriconvert). Anatomical images were preprocessed by first stripping non-brain tissue using FMRIB Software Library ((FSL; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/) BET (Smith, 2002)), then aligned along the anterior commissure–posterior commissure plane to adjust for head tilt (using FMRIB Linear Image Registration Tool, FLIRT; (Jenkinson, Bannister, Brady, & Smith, 2002) after registration to the 1 mm MNI-152 template and resampled into 1 mm cubic voxels.

2.12 |. Sulcal tracing: Characterization

The OFC sulci were then identified from each subject’s T1 image using the software ITK-SNAP (Yushkevich et al., 2006). To identify each sulcus, we used a procedure published as part of OFC sulcogyral pattern characterization methods (Chiavaras & Petrides, 2000) and adapted by our group (Patti & Troiani, 2018; Zhang et al., 2016). In this procedure, the medial orbital sulcus is defined as the sulcus immediately lateral to the olfactory sulcus, which is the most medial orbital sulcus. The lateral orbital sulcus is defined as the most prominent sulcus in the lateral most position on the orbital surface of the frontal lobe. The transverse orbital sulcus is defined as the sulcus running horizontally between the medial and lateral orbital sulci, separating the rostral and caudal portions of these sulci. Other sulci that are present between the MOS and LOS are considered to be intermediate sulci and can vary substantially in size, location and number across individuals. Any additional sulci caudal to the transverse sulcus are considered posterior orbital sulci. Although we primarily used this labeling procedure to identify OFC sulci, we also provide details on sulcogyral pattern characterization of these subjects in order to confirm sulcogyral variability, consistent with previous work (Li et al., 2015). Although we do not assess Type I, Type II and Type III patterns here, we provide the frequency distributions in supplementary data (Table S1), which are consistent with published work. To characterize OFC sulcogyral patterns, a sulcus was determined to be discontinuous if the sulcus was absent for three adjacent coronal slices separating the rostral and caudal portions. Type I (discontinuous MOS and continuous LOS), Type II (continuous MOS and LOS) and Type III (discontinuous MOS and LOS) patterns were characterized by MP. If the very rare Type IV pattern (continuous medial and discontinuous lateral sulcus) was observed, this subject was characterized as Type III, consistent with previous work (Bartholomeusz et al., 2013). Each subject’s OFC sulcal pattern was evaluated by MP, and a subset (20 hemispheres) was evaluated by VT. Interclass inter-rater reliability was high κ = .90. MP additionally re-classified a different subset of 20 hemispheres to establish intra-rater reliability κ = .92.

2.13 |. Statistical comparisons of value signals

To assess whether there was a significant difference in the location of face and food value signals, signals were grouped according to their sulcogyral location. We characterized those as lying near the midline of the brain (including supraorbital sulcus, olfactory sulcus and subcallosal sulcus) as medial and those anywhere within the right or left H-sulcus (MOSr, MOSc, LOSr, LOSc and TOS) as lateral. If a value signal was not within a sulcus, but primarily in a gyral location (posterior orbital gyrus, medial orbital gyrus, etc.), these clusters were not included in summary data corresponding to sulcal locations, but are included in the individual data files in supplementary info (Tables S2–S7). Chi-square tests were then used to assess whether numbers of subjects with value signals were significantly different for faces and food within each sulcus. Presence of a value signal was not considered separately for left and right sides of the brain. That is, if a subject had at least one signal on the left or right within a sulcus of interest, that participant was considered to have a value signal in that sulcal location. If a subject had a signal within the same sulcus on the left and the right, this subject was also considered to have a value signal in that location.

Because the signal to noise ratio in the OFC differed between our two acquisition sites, we assessed whether the number of value signals differed between the subjects scanned at the two sites. This was accomplished by summing the total number of value signals observed for each individual, across the entire OFC. We tested whether the number of signals was significantly different between those scanned at the Temple location versus those scanned at Geisinger using an independent sample t test. In addition, we assessed whether there was a relationship between number of value signals observed and individual tSNR in the OFC with a Pearson correlation.

2.14 |. Variability of sulcus locations within the OFC

In an effort to visualize the variability for a given sulcus in this region, overlap maps were created of all subject’s left and right MOS as well as left and right LOS. As part of the standard tracing procedure, 3D maps are saved in which each subject’s sulci are traced using a different color, which corresponds to a different numerical value in the saved 3D file. Individual sulci were extracted for each sulcus and saved as a new 3D map for each subject. These 3D maps were then summed across all 48 subjects for the left and right MOS and LOS, separately, such that the resulting numerical value would reflect the number of subjects that have a portion of each sulcus in a given voxel. That is, if the resulting image contains a value of 10 that indicates that 10 subjects have a portion of their MOS that lies in that voxel location.

2.15 |. Reliability of spatial location of value signals

Finally, we assessed whether the location of face and food value signals within the OFC were present in a consistent spatial location within an individual across time. A single female participant, aged 22, that was not included in the analyses above completed a longer version of the object localizer (four runs) at two time points, separated by 1 week. Identical preprocessing, modeling and analysis steps were used as described above, except multiple runs within a scanning time point were combined using standard methods.

3 |. RESULTS

3.1 |. Group analysis: Whole-brain

We present group-level, whole-brain analyses to assess whether there is consistent activation within the H-sulcus using traditional group-level statistical parametric mapping methods. Within the OFC, clusters of activation were present bilaterally within the H-sulcal region and medial OFC, including the gyrus rectus, the olfactory sulcus and the supraorbital sulcus for the contrast of Faces > Places & Clocks (see Figure 1a). There was also activation within the bilateral TOS of the template brain and right lateral caudal MOS for the contrast of Faces > Places & Clocks (again, see Figure 1a). Activation for the contrast of Food > Places & Clocks was concentrated to the caudal portion of the left MOS of the template brain (see Figure 1b). Other regions outside of the OFC were also activated, including the amygdala, insula and ventral visual cortex for Faces > Places & Clocks and the left insula and visual cortex for Food > Places & Clocks (For peak coordinates from the group-level whole-brain analysis, see Table S2).

FIGURE 1.

Whole-brain group activation of Faces > Places and Clocks (Panel a) and Food > Places and Clocks (Panel b). Images are thresholded at a t-score of 2 and corrected for multiple comparisons using threshold-free cluster estimation with p values set to p < .001. Coordinates at the top of Panel (a) and (b) correspond to the coordinates of the rendered brain image, using the ch2better template in MRIcron

3.2 |. Location of face- and food-selective clusters

We next summarize cluster locations using each individual’s anatomy within the OFC and distinguish between (a) sulcogyral locations along the midline of the brain (primarily within the supraorbital sulcus) and (b) sulcogyral locations within the H-sulcus. Consistent with our previous work (Troiani et al., 2016), the majority of value signals near the midline of the OFC were face-selective (75% of subjects had face-selective value patches in medial OFC; See Table 2). More subjects had face value signals in sulcogyral locations along the midline of the brain than food value signals (face signals: 36/48 subjects; food signals: 20/48 subjects; χ²(1, N = 48) = 10.97, p = .0009). These were primarily located within the supraorbital gyrus immediately dorsal to the olfactory sulcus. See Table 1 for the number of subjects with clusters in each sulcogyral location and more detailed subject-level data in Tables S3–S5.

TABLE 1.

Number of subjects with face and food value signal clusters in each anatomical location based on individual anatomy

		H-sulcus N (%)
Medial N (%)		MOSr		MOSc		LOSr		LOSc		TOS
Face	Food	Face	Food	Face	Food	Face	Food	Face	Food	Face	Food
36 (75)	20 (42)	3 (6)	4 (8)	7 (15)	23 (48)	2 (4)	1 (2)	9 (19)	7 (15)	6 (13)	12 (25)
χ2 = 10.97, p = .0009		χ2 = 0.154, p = .695		χ2 = 12.41, p = .0004		χ2 = 0.334, p = .557		χ2 = 0.300, p = .583		χ2 = 2.46, p = .117

Abbreviations: LOSc, caudal lateral orbital sulcus; LOSr, rostral lateral orbital sulcus; MOSc, caudal medial orbital sulcus; MOSr, rostral medial orbital sulcus; TOS, transverse orbital sulcus.

Next, we assessed whether face and food value signals were in distinct locations within particular sulci of the H-sulcus (see Table 1). Although clusters can be found within most sulci of the H-sulcus, food value signals were consistently located within the MOS, and specifically within the caudal portion of this sulcus (23 subjects with at least 1 MOSc cluster within either hemisphere). There were significantly more subjects with MOSc clusters for food than faces (χ²(1, N = 48) = 12.41, p = .0004). The number of face and food clusters in any other sulcus within the H-sulcus was not significantly different (all χ² < 1.46, all p’s > .117; see Table 1). We also report the individual locations of H-sulcus clusters in Tables S6–S8. For examples of individual subject’s anatomy overlaid with activation clusters, please see Figure 2.

FIGURE 2.

Example patterns of anatomo-functional organization of the orbitofrontal cortex (OFC) observed in individual subjects. The face-specific activity is most consistently located along the midline of the brain, within the supraorbital sulcus (Top panel). In contrast, the food-specific activity is consistently located within the caudal portion of the medial orbital sulcus (MOS) (Bottom panel). On the axial plane of subjects’ anatomical image, sulci are color-coded with the MOS in red, lateral orbital sulcus (LOS) in blue, intermediate orbital sulcus (IOS) in green, and transverse orbital sulcus (TOS) in yellow; L, left hemisphere; R, right hemisphere. The z-coordinate of each image is displayed at the top right of each brain image

Given the difference in tSNR between the two acquisition sites, we assessed whether tSNR influenced our ability to detect signals between the groups. There was no difference between the overall number of signals detected between the Temple group and the Geisinger group (Geisinger, 3.0 ± 1.5; Temple, 3.7 ± 1.5; t(47) = 1.16, p = .126, N.S.), despite the higher tSNR in the Geisinger group. Also, overall the number of signals was higher in the group scanned at Temple, which overall had a significantly lower tSNR. We also examined whether overall signal in the OFC for each subject influenced the number of value signals detected across all participants. There does not appear to be a relationship between number of signals and tSNR (r = .164, n = 48, p = .265, N.S.). Thus, these analyses using number of value signals and tSNR indicate that value signals can be detected at both sites using different head coils and that tSNR differences between sites are unlikely to impact our ability to detect value signals overall.

To those unfamiliar with the sulcogyral anatomy of the OFC, the extent to which sulcal locations vary from subject to subject may not be immediately obvious. To better describe this variability, we also assessed whether the value signals were consistently localized when using a standard average brain rather than the anatomy of an individual subject. This was completed for face and food responses in both medial and lateral (H-sulcus) activations. Summary results across all subjects can be seen in Table 2. Individual-level results of this localization procedure for face and food activations in the H-sulcus can be viewed in Tables S10–S12 and summarized in Tables S13 and S14. Please note that for both face and food activations in the medial portion of the OFC (olfactory sulcus, supraorbital sulcus and/or gyrus rectus), all of the sulcogyral locations were identical for the individual subject’s anatomy and the group average anatomy. For this reason, we did not generate new tables that depicted any differences in localization using the two different reference frames. In contrast, for both face and food activations in the H-sulcus, a large proportion of activations would be incorrectly identified if the group average template brain was used. More specifically, over half (67/114 distinct activations across all subjects) of functional activations within the H-sulcus are mischaracterized when using the group average anatomical reference frame. This was mostly due to activations that would be characterized within a sulcus using individual subject’s anatomy appearing to be in gyral locations when overlaid on the group average brain. This mischaracterization was true for activations to faces in the H-sulcus (27/47 activations would have been mischaracterized; see Table S12 for summary) and food activations (40/67 activations would have been mischaracterized; see Table S13 for summary). Importantly, for our finding of consistent activations in the caudal MOS for food, over half of the functional activations that were localized to the MOS based on individual subject anatomy would be localized to nearby sulci or gyri, including the anterior orbital gyrus, medial orbital gyrus, posterior orbital gyrus and transverse orbital sulcus of the group average brain. A summary table of the number of activations in each region is summarized in Table 2 along with corresponding statistical tests. These results indicate that while medial OFC localization was identical using either individual anatomy or the template brain, the large number of MOSc food-specific nodes would not have been identified if using the template brain (See Table 2: MOSc result; χ²(1, N = 48) = 1.787, p = .181, N.S.). Thus, the use of individual subject’s sulcogyral anatomy as a reference frame allowed us to identify this novel result.

In addition to assessing whether sulcogyral locations would be different when using the group average template brain rather than individual subject anatomy, we also rendered images to illustrate the variability of a sulcus in this region. For example, subject’s left MOS tracing (red) and a functional activation node (yellow) are overlaid on the subject’s anatomical image in Figure 3a (left), while only the activation node is overlaid on the subject’s anatomical image in 3a (right). In 3b, the left MOS tracing and activation node from this subject is overlaid onto the group average MNI image. This example was chosen in order to illustrate several important points: (a) All OFC sulci must be traced in order to properly identify individual sulci. In this example, the rostral portion of the MOS does not appear in this plane. The sulcus that appears rostral to the traced caudal MOS is actually an intermediate sulcus. This could easily be mistaken for the rostral MOS if all sulci were not traced and identified. (b) If a group average template is used rather than individual anatomy for the purposes of localizing value signals, functional activation lying on the caudal portion of the individual’s MOS depicted in Figure 3a could be mistakenly identified as within the medial orbital gyrus, rather than a sulcus (Figure 3b). To illustrate this variability further, we averaged the tracings across all subjects for the left and right MOS as an example (Figure 3c,d; left MOS only depicted in figure). The average covers a large swathe of cortex that may generally provide a predictive inclusion boundary where the MOS lies for any given subject. However, there is very little overlap across subjects for the exact location of the trough of the sulcus that is traced as part of our procedure (See Figure 3c for overlap map of at least 20% of subjects and Figure 3d for more liberal overlap map of at least 10% of subjects). Thus, averaging sulcogyral anatomy discards important variability that establishes a potentially useful reference frame. Only one sulcus is depicted in Figure 3 for simplicity, but additional Figures S1 and S2 include group averages of bilateral sulci in the MOS and LOS at different thresholds to further emphasize variability in precise sulcal location between subjects.

FIGURE 3.

(Panel a) Left; Axial view of example subject’s T1 image overlaid with left medial orbital sulcus (MOS) outlined in red and a functional activation node in yellow. Right; Axial view of example subject’s T1 with functional activation node in yellow. Note overlap of functional activation node and outline of caudal MOS. (Panel b) Left; Axial view of MNI template average brain overlaid with example subject’s (same subject as in Panel a) left MOS outlined in red and functional activation node in yellow. Right; Axial view of MNI template average brain overlaid with functional activation node in yellow. Note that the functional activation node does not lie in the caudal MOS of the template brain and would be identified as medial orbital gyrus. (Panel c) Axial view with overlaid tracings of all 48 subjects left MOS in peach. Portion of left MOS where at least 10/48 subjects (20%) have overlapping voxels in deep red. (Panel d) Axial view with overlaid tracings of all 48 subjects left MOS in peach. Portion of left MOS where at least 5/48 subjects (10%) have overlapping voxels in deep red

3.3 |. Reliability of spatial location of face and food signals

One subject not included in previous analyses was asked to complete a longer version of the functional scan at two time points. We scanned this additional subject in order to better understand the reliability of these value signals across time. Activation in the OFC to faces was consistently located within the supraorbital sulcus, consistent with our findings here in the larger group. Activation in the OFC was consistently located within the MOS, at the juncture between the caudal MOS and the TOS (see Figure 4). These results indicate that the locations of these value signals are reliable across time points.

FIGURE 4.

Single subject test-retest reliability of face and food activation within the orbitofrontal cortex (OFC). A single subject completed a longer version of the face/food localizer task at two separate time points, separated by 1 week. Whole-brain t-statistic maps for this subject are depicted on the subject’s anatomical image with face-selective activation in Panel a (Faces > Places/Clocks) and food-selective activation in Panel b (Food > Places/Clocks). (Panel a) Time point 1 in red and Time point 2 in yellow. (Panel b) Time point 1 in blue and Time point 2 in green. T-statistic maps are presented with a threshold of 4. Coordinates at the top of Panel (a) and (b) correspond to x, z, and y coordinates of the subject’s brain image transformed into 1 mm MNI space

4 |. DISCUSSION

The majority of work assessing differences in functional activation does not consider sulcogyral anatomy. Thus, our work demonstrates the utility of considering sulcogyral location as a useful reference frame for activation nodes. The benefit of using a sulcogyral reference frame to describe activation can be helpful for differentiating the specific location of category-based responses (i.e., face but not food selectivity in medial OFC; Troiani et al., 2016). The individual anatomy for a given subject will always be a more accurate reflection of the underlying morphology; thus, we suggest using individual anatomical references in all cases of sulcogyral localization. The main reason that we suggest using individual anatomy in all cases is that it would be unclear whether a given individual’s anatomy differs substantially from the template brain, necessitating a direct comparison between the individual’s anatomy and template anatomy to establish the correspondence between the individual anatomy and the template brain. If one always uses the individual anatomy, this requires only referencing one anatomical schema rather than two. However, it is important to note that in some cases, the template brain anatomy would have been sufficient. For example, we did not show an advantage to using individual anatomy in the case of medial activations within the supraorbital sulcus in the current study. Future work should continue to assess which orbitofrontal sulci have the greatest overlap in order to determine analyses that would benefit from using individual anatomy versus using the template brain.

We report the novel finding that food value signals are consistently located in the caudal portion of the medial orbital sulcus, but only when using subject-specific sulcogyral anatomy. We also confirm our previous finding that face signals are most consistently located in more medial portions of the OFC, specifically the supraorbital gyrus, regardless of whether the sulcogyral reference frame is that of an individual subject or group average. Thus, using a sulcogyral reference frame appears to be particularly important when localizing functional activation to certain object categories, such as food. This finding further emphasizes the importance of considering individual anatomical variability on functional response.

Our main finding is that food value signals appear to be located most consistently within the caudal MOS, when using subject-specific sulcogyral reference frames. This finding is consistent with recent work by Suzuki et al. (2017), which found that certain nutrient aspects of food (vitamin content, carbohydrate content) was localized to portions of the caudal MOS. The location of food value signals is also consistent with the findings of Li et al. (2015), where erotic value signals were specifically at the intersection of caudal MOS and TOS, while monetary signals were in more rostral MOS. However, there are also several differences between the current study and the work by Li et al. (2015), including the experimental procedure, the participant population and the category of the stimuli. Li et al. (2015) used an explicit valuation procedure, included only heterosexual males who abstained from sexual activity, and used monetary and erotic (naked women) stimuli. Authors found that monetary rewards are located within the rostral portion of the MOS, and erotic rewards were localized to the junction of the caudal MOS and TOS. While we do find consistency for food signals in the caudal MOS and consistent face signals in the medial OFC (supraorbital gyrus), we also demonstrate a much larger degree of variability of the particular segment and sulcus for activation within the H-sulcus across individuals. That is, many individuals have activation signals in other portions of the H-sulcus for both faces and food. The larger variability in sulcogyral location of value signals in the current study may be due to the greater diversity of subjects (males and females) and reliance on participants’ spontaneous valuation of stimuli, rather than an explicit procedure. Namely, we used this design in an attempt to evoke automatic valuation, whereas the bulk of studies ask participants to assign explicit value to the stimulus while in the scanner. Thus, it may be that requiring participants to assign value across multiple stimulus categories evokes a comparison process that necessitates a common reference dimension, resulting in activation that has been interpreted as a “common currency.” It will be important for future work to explore the differences between automatic evaluation processes and those associated with a forced comparison.

It is unclear whether the subjects in the study by Li et al. (2015) also had value signals in more medial portions of the OFC for erotic value signals, similar to those described here for faces. Only signals in the MOS are reported, although group-level activation maps for erotic value response shows activation within the same region of cortex that we find for face value signals. The potential of medial value signals for erotic stimuli is relevant, as the full-body stimuli also included faces. Future work should explore whether these “face patches” are selective for face stimuli or more generally tuned to objects with social relevance.

Results from the more traditional group-level analyses (Figure 1) indicate there is some degree of spatial differentiation in response to faces versus food. Faces activated OFC along the midline of the brain, in the bilateral TOS and in the right MOS of the template brain. Food activated left-lateralized MOS on the template brain. These group-level results may lead one to erroneously infer that face signals are also primarily located in the TOS, even though TOS value signals were identified in equivalent numbers of subjects for faces and food. We completed an additional analysis to illustrate the utility of using the individual subject’s anatomy as a reference frame. Results from this analysis indicate that over half of the activations within the H-sulcus would be mischaracterized as a different sulcogyral location if using the group average template brain. Interestingly, this only impacted results in the lateral OFC (H-sulcus) but did not impact results in the medial OFC, including activations within the olfactory sulcus, gyrus rectus and supraorbital sulcus. This difference in localization between medial and lateral portions indicates a greater degree of overlap and homology in the most medial orbitalfrontal sulcogyral landmarks. Further, this difference emphasizes how we were able to identify consistencies in face signals within medial portions of the OFC in our previous work, despite using regions of interest based on group average templates (Troiani et al., 2016).

Thus, while traditional statistical parametric mapping group results are informative, an analysis that considers each individual’s sulcogyral anatomy may yield different results. Further, the differences between the group map results and our individual analyses suggest that identification of subject-specific signals may lend themselves to future analyses that probe subject-specific response properties of these regions, similar to the use of localizer scans to identify individualized functional regions of interest in other regions of cortex (Barch et al., 2013; Fedorenko et al., 2010; Fox, Iaria, & Barton, 2009). We and others have found that individual differences in sulcogyral anatomy are related to individual differences in behavior (Chye et al., 2017; Hirjak et al., 2017; Takahashi et al., 2016; Watanabe et al., 2013; Whittle et al., 2014; Zhang et al., 2016) and multiple psychiatric disorders (Patti & Troiani, 2018). Thus, it will be important for future research to identify how the interaction of functional response to stimulus value and sulcogyral anatomy is related to psychopathology.

The OFC is often described as a singular region, but the underlying cellular architecture indicates multiple subregions. Because multiple findings suggest that stimulus- or domain-specific responses in the OFC are sometimes identified within distinct spatial locations, these locations are thought to reflect the underlying organization of OFC. For example, response properties in the OFC have been described according to spatial gradients in the anterior–posterior and medial–lateral directions (Kringelbach, 2004, 2005). Value responses to primary rewards are thought to be located in more posterior portions of the OFC, while value responses to secondary rewards are located in more anterior regions. Carmichael and Price (1994) suggested a medial–lateral distinction based on cortical–cortical connections, with lateral regions comprising a sensory network with dense connections between olfactory, gustatory and visual cortex. The medial regions comprise the visceromotor network, with dense connections to nuclei important for autonomic control, including the hypothalamus and periaqueductal gray.

This differentiation might be best explained by the medial–lateral distinctions suggested by Carmichael and Price (1994), with food activating more lateral regions with connections to olfactory and gustatory cortex and faces most consistently activating medial portions. Medial regions associated with the visceromotor network project to brain nuclei important for autonomic control, which is thought to be important for appropriate social reciprocity (Porges & Furman, 2011). Thus, it is interesting that face value signals, a social stimulus, are most consistently located within medial OFC.

The consistent overlap in spatial location of both face and food value signals in our test–retest subject indicates that these value signals are reliably activated within a subject across time. These test–retest results also suggest that even though we use a passive viewing paradigm without an explicit valuation response, the spatial location of these activations is consistent. This suggests that passive viewing paradigms that lack an explicit valuation procedure can evoke reliable and consistent value signals within the OFC.

4.1 |. Limitations

There are several limitations in the current design that should be addressed in future studies. We did not manipulate or control the context of the participant, such as limiting their food intake or social interactions prior to the scan. Animal work indicates that OFC neurons can be tuned to specific stimuli that have inherent value, including olfactants, food, sexual signals and even the faces of conspecifics (Gottfried & Zelano, 2011; Watson & Platt, 2012). However, the response profile of neurons tuned to specific stimuli are also subject to satiation, with reduced or no response following satiation to the specific odorant or food (Critchley & Rolls, 1996; Rolls, Critchley, Browning, Hernadi, & Lenard, 1999; Rolls, Murzi, Yaxley, Thorpe, & Simpson, 1986; Rolls, Sienkiewicz, & Yaxley, 1989). Further, fMRI work in humans also suggests that specific reward identity responses in OFC were altered after satiation (Gottfried et al., 2003; Howard & Kahnt, 2017; Kringelbach, O’Doherty, Rolls, & Andrews, 2003), with connectivity impacted by dopamine receptor availability (Kahnt & Tobler, 2017). Thus, the context and homeostasis of the organism is an important consideration for understanding how manipulating homeostasis and/or context impacts value signal location.

We also did not assess or control for attractiveness between the various object categories used here. In previous work (Troiani et al., 2016), we assessed the hedonic value of this stimulus set in an independent group of adults. Subjects rated faces and food to have significantly higher value than places and clocks but also rated food to have greater hedonic value than faces. Other recent work has found food-specific macronutrient signals in the MOS (Suzuki et al., 2017), consistent with our finding of the majority of food-specific signals in the MOS. Further, another work that has controlled for attractiveness across image categories still identified category-specific and domain-general responses to faces and scenes in the OFC (Pegors, Kable, Chatterjee, & Epstein, 2015). Thus, our results converge with previous evidence found in other fMRI paradigms. However, future work should assess whether the location of value signals could be influenced by a difference in overall value of one stimulus category relative to another.

We used a block design study rather than an event-related design that would allow us to capture temporal differences in value signal activation and location. The different components of reward anticipation and consumption is well-documented in both animal and human work (Berridge, 2009; Berridge & Robinson, 1998; Berridge, Robinson, & Aldridge, 2009; Knutson, Adams, et al., 2001; Knutson, Fong, Adams, Varner, & Hommer, 2001) and temporal dynamics may be different based on reward category (Li, Vanni-Mercier, Isnard, Mauguière, & Dreher, 2016). Future studies should address whether the location of value signals identified through a block design paradigm is consistent with those identified through event-related paradigm.

We did not assess whether participants were more familiar with one category relative to another. For example, participants could have been more familiar with food stimuli based on their everyday experiences with food. Data from the subject that completed test–retest using the same stimuli indicate that the value signals are in identical locations at each time point, despite more familiarity with all the stimuli at the second time point. For this reason, we do not think that familiarity with an object category will necessarily change the location of the value signal, although as interest is diminished in category, overall activation in the region could certainly decrease with diminished attention. Future work may want to investigate whether the location of value signals change according to differences in stimulus category familiarity.

We adopted the term value signal based on previous work characterizing erotic- and monetary-specific activation when explicitly valuing a stimulus. However, we did not have participants specifically place a conscious value on each stimulus nor did we confirm that participants were attracted to these specific stimuli. Based on an independent group of participants rating, the images used in this fMRI paradigm, face and food stimuli were rated to have significantly higher value than places and clocks (Troiani et al., 2016). Previous fMRI studies have also found increased activation in the OFC and other regions of the limbic system to be more activated to pictures of faces (Aharon et al., 2001; Fusar-Poli et al., 2009; Mende-Siedlecki, Said, & Todorov, 2013) and food (for review, see Pursey et al., 2014), which is evident that the OFC activates more to overall object categories of high value, even within block design studies that do not require an explicit report of object value.

5 |. CONCLUSIONS

This study confirms that identification of sulci in the OFC is a useful reference frame for identifying individual differences in the spatial location of value signals in the human brain. We confirm the presence of face-selective signals within medial OFC. We also identify both face and food- value signals within the H-sulcus with the most consistent activation in the caudal MOS for food signals. This work adds to the growing body of evidence that sulcogyral anatomy and anatomo–functional relationships demonstrate variance within the typical population. Future work that addresses the disjunctions between anatomy and functional response properties within psychiatric and neurodevelopmental disorders may be important for understanding the organization of the OFC and its impact on brain disorders.

Abbreviations:

H-sulcus

lateral orbital sulcus and transverse orbital sulcus, used to refer to the combination of medial orbital sulcus

LOSc

caudal portion of the lateral orbital sulcus

LOSr

rostral portion of the lateral orbital sulcus

MOSc

caudal portion of the medial orbital sulcus

MOSr

rostral portion of the medial orbital sulcus

OFC

orbitofrontal cortex

TOS

transverse orbital sulcus

tSNR

temporal signal to noise ratio