American Society of Functional Neuroradiology–Recommended fMRI Paradigm Algorithms for Presurgical Language Assessment

INTRODUCTION:Functional MR imaging is increasingly being used for presurgical language assessment in the treatment of patients with brain tumors, epilepsy, vascular malformations, and other conditions. The inherent complexity of fMRI, which includes numerous processing steps and selective analyses, is compounded by institution-unique approaches to patient training, paradigm choice, and an eclectic array of postprocessing options from various vendors. Consequently, institutions perform fMRI in such markedly different manners that data sharing, comparison, and generalization of results are difficult. The American Society of Functional Neuroradiology proposes widespread adoption of common fMRI language paradigms as the first step in countering this lost opportunity to advance our knowledge and improve patient care.LANGUAGE PARADIGM REVIEW PROCESS:A taskforce of American Society of Functional Neuroradiology members from multiple institutions used a broad literature review, member polls, and expert opinion to converge on 2 sets of standard language paradigms that strike a balance between ease of application and clinical usefulness.ASFNR RECOMMENDATIONS:The taskforce generated an adult language paradigm algorithm for presurgical language assessment including the following tasks: Sentence Completion, Silent Word Generation, Rhyming, Object Naming, and/or Passive Story Listening. The pediatric algorithm includes the following tasks: Sentence Completion, Rhyming, Antonym Generation, or Passive Story Listening.DISCUSSION:Convergence of fMRI language paradigms across institutions offers the first step in providing a “Rosetta Stone” that provides a common reference point with which to compare and contrast the usefulness and reliability of fMRI data. From this common language task battery, future refinements and improvements are anticipated, particularly as objective measures of reliability become available. Some commonality of practice is a necessary first step to develop a foundation on which to improve the clinical utility of this field.

The use of fMRI in the presurgical assessment of language function, especially in patients with brain tumors, vascular malformations, or epilepsy, has become standard throughout numerous institutions in North America, Europe, and other parts of the world, and the reliance on this technology is increasing.¹ fMRI offers a valuable noninvasive means of assessing language function lateralization and localization, which complements, or in some cases, obviates intraoperative electrocortical stimulation (ECS) mapping.^2,3 The Organization of Human Brain Mapping Committee on Best Practice in Data Analysis and Sharing has recently published a white paper that addresses fMRI research reproducibility through transparency of trial design, tools used for data manipulation, and reporting.⁴Beyond research reporting standards, the American Society of Functional Neuroradiology (ASFNR) perceives the need for increased standardization in clinical practice in an attempt to enhance the communicability of how we assess our patients and the meaningfulness of our imaging findings and reports. Functional MR imaging involves numerous processing steps, which vary among manufacturers of fMRI systems, and this complexity is compounded by institution-unique methods for patient training, paradigm choice, and postprocessing. Data sharing is necessary for research integrity and scientific transparency, but current practice variability obscures reproducibility and hinders adequate interinstitutional sharing of information. This widespread variability thereby limits collective progression, particularly in the presurgical assessment of language function—one of the most important applications of fMRI in clinical practice.Since April 2013, the ASFNR has hosted a monthly teleconference in which clinical fMRI practitioners from across North America have presented presurgical mapping cases from their various practices for educational purposes. This recurring teleconference has highlighted the existing practice variability between institutions. However, it has also offered a regularly occurring, accessible line of communication that has provided impetus for converging practice parameters with a view toward enhanced validation of imaging methodologies, data sharing, and knowledge growth. Consequently, the ASFNR Clinical Practice Committee decided that to strengthen the value of preoperative fMRI language assessment, an ASFNR-approved set of standardized language paradigms should be developed.Because language can be represented across phonologic, orthographic, semantic, pragmatic, and discourse dimensions and 1 task cannot simultaneously activate all of these aspects, multiple tasks are recommended to provide a more sensitive and specific map of language function that will aid in surgical planning.^5,6 There are a number of desirable features for a standardized fMRI task battery: 1) The ideal fMRI language task paradigm set needs to be appropriately challenging for the patient to produce ample activation without being so difficult as to overwhelm a neurologically compromised or otherwise challenged individual; 2) the tasks need to provide an appropriate balance of sensitivity and specificity for language-related activation; and 3) there needs to be the ability to provide both reliable interhemispheric language lateralization as well as intrahemispheric localization of both expressive and receptive language sites, such as the Broca or Wernicke area, with respect to intracerebral lesions.Language Paradigm Review ProcessMembers of an ASFNR Language Paradigm Taskforce were invited to join this project following discussions with the ASFNR Clinical Practice Committee and Executive Committee in an attempt to form a group of interested participants with expertise in the field, a historical perspective of how fMRI has evolved to its current state, and insight into how fMRI must continue to change to further the field.As a first step, the members of the taskforce discussed their own institution''s paradigm choices, practice parameters, and the value of combining results from multiple paradigms. From these discussions, project goals and scope were distilled, including the concept of deriving standardized recommendations that all institutions would hopefully adopt. Initially, the group decided to limit the first iteration of standard paradigms to visually presented paradigms for adult patients with a seventh-grade-equivalent or greater reading ability. Subsequently, it became clear that there was a clinical need to address the pediatric population and adults with limited reading ability, and recommendations for these populations were assessed and developed.The entire ASFNR membership was polled as to how many language-assessment fMRIs per month were performed at their various institutions to determine the most commonly used language paradigms. The poll attempted to gauge the willingness of ASFNR members to adopt a common set of language paradigms. Poll results were assessed by the taskforce, and an attempt was made to balance current practice preferences across the nation with evidence-based data regarding the variable strengths and weaknesses of various language paradigms. The goal was to choose complementary paradigms that would be reasonably applicable to the greatest number of patients. By balancing current practice and scientific evidence, we hoped to motivate adoption of the standardized task battery by making the required change as simple as possible for the practicing members of the society while at the same time being guided by the scientific evidence. This goal necessitated a literature review for each of the most commonly used language paradigms.Two-hundred fifty-nine requests for surveys were sent out with 6 e-mail addresses failing. Fifty-three of 253 responded (21%). Figure 1 depicts the number of fMRIs performed per month for language assessment at different institutions. Fifty-seven percent (30/53) of the responders reported being very likely to adopt language paradigm algorithms, and only 8% (4/53) reported being unlikely to adopt ASFNR recommendations (Fig 2).Open in a separate windowFig 1.fMRIs performed per month for language assessment at different institutions (about one-third of responders do 6+ per month).Open in a separate windowFig 2.Fifty-seven percent (30/53) of responders reported being very likely to adopt language paradigm algorithms, and only 8% (4/53) reported being unlikely to adopt ASFNR recommendations.Results of the ASFNR poll produced 3 tiers of commonly used paradigms across the nation (Fig 3). In the poll, Silent Word Generation (SWG) and Sentence Completion (SC) stood out as the most frequently employed tasks while Verb Generation, Object Naming, Rhyming, and Reading Comprehension seemed to coalesce in a second-tier group. Because SWG and SC were the most commonly used paradigms and also had support in the literature for being reliable and useful, these tasks were favored to form the core of the standard language task battery. However, because a combination of language paradigms has been shown to increase sensitivity and specificity, a third task was considered desirable.^7,8 Nevertheless, converging on a single choice for that third task proved to be more difficult because the taskforce wanted to maintain flexibility for the radiologist to tailor an examination to the patient''s specific clinical scenario.^9,10Open in a separate windowFig 3.The most commonly used language paradigms were sorted into 3 tiers with the total number of imagers by using the paradigms represented on the y-axis. The first tier included SWG and Sentence Completion. The second tier included Verb Generation, Object Naming, Rhyming, and Reading Comprehension. The third tier included Antonym Generation, Semantic Decision, Noun-Verb Semantic Association, Category Naming, and other.Once the paradigm recommendations were decided upon, the taskforce then converged on the specific scanning parameters and stimuli for the paradigms such that the tasks would be vendor-neutral.ASFNR RecommendationsFollowing an analysis of the poll results as well as literature review regarding each of the paradigm tasks mentioned in the poll, the ASFNR taskforce developed language algorithms for both adults and pediatric subjects (Figs 4 and ​and55).Open in a separate windowFig 4.Adult algorithm for presurgical language fMRI.Open in a separate windowFig 5.Pediatric algorithm for presurgical language fMRI.The default adult algorithm includes SC, SWG, and Rhyming. However, to satisfy the need to customize the analysis to the specific clinical scenario, the radiologist may choose to drop Rhyming and repeat the SC or SWG task as a means of confirming and correlating activations between time courses. SC most often offers more robust language area activation than SWG; thus, it would be the most appropriate task to repeat. In patients who may have difficulty adequately performing the SC or SWG tasks, the radiologist may choose either the Object Naming (ON) or Passive Story Listening (PSL) tasks. This also allows customization in that ON is primarily an expressive task and PSL is primarily receptive.The default pediatric algorithm includes SC, Rhyming, and Antonym Generation (AG), but for those patients unable to adequately perform Rhyming, the PSL task should be used instead. In general, this pediatric algorithm should apply to most patients 5–11 years of age, but each patient''s ability to adequately perform tasks should be assessed before the scan.In addition to the language-specific tasks, breath hold fMRI may enhance the fMRI language task battery by identifying areas of potentially false-negative activation. By elevating blood carbon dioxide levels, a breath hold task measures cerebrovascular reactivity throughout the brain. It can therefore help demonstrate potential neurovascular uncoupling or confounding susceptibility artifacts as regions of absent or reduced blood oxygen level–dependent (BOLD) signal in response to breath holding, thereby clarifying regions of false-negative language activation. Due to atypical postprocessing requirements for this task, inclusion of the breath holding task is not considered to be a required component of the language task battery but is nonetheless recommended if available at an institution, to assess potential loss of sensitivity to the BOLD signal, which would influence the confidence level of the interpretation.