Language fMRI has become an integral part of the planning process in brain surgery. However, fMRI may suffer from confounding factors both on the patient side, as well as on the provider side. In this study, we investigate how patient-related confounds affect the ability of the patient to perform language fMRI tasks (feasibility), the task sensitivity from an image contrast point of view, and the anatomical specificity of expressive and receptive language fMRI protocols. 104 patients were referred for language fMRI in the context of presurgical procedures for epilepsy and brain tumor surgery. Four tasks were used: (1) a verbal fluency (VF) task to map vocabulary use, (2) a semantic description (SD) task to map sentence formation/semantic integration skills, (3) a reading comprehension (RC) task and (4) a listening comprehension (LC) task. Feasibility was excellent in the LC task (100%), but in the acceptable to mediocre range for the rest of the tasks (SD: 87.50%, RC: 85.57%, VF: 67.30%). Feasibility was significantly confounded by age (p = 0.020) and education level (p = 0.003) in VF, by education level (p = 0.004) and lesion laterality (p = 0.019) in SD and by age (p = 0.001), lesion laterality (p = 0.007) and lesion severity (p = 0.048) in RC. All tasks were comparable regarding sensitivity in generating statistically significant image contrast (VF: 90.00%, SD: 92.30%, RC: 93.25%, LC: 88.46%). The lobe of the lesion (p = 0.005) and the age (p = 0.009) confounded contrast sensitivity in the VF and SD tasks respectively. Both VF and LC tasks demonstrated unilateral lateralization of posterior language areas; only the LC task showed unilateral lateralization of anterior language areas. Our study highlights the effects of patient-related confounding factors on language fMRI and proposes LC as the most feasible, less confounded, and efficiently lateralizing task in the clinical presurgical context.

语言功能磁共振成像已成为脑外科手术计划过程中不可或缺的一部分。但是，fMRI可能在患者端以及提供者端都受到混杂因素的困扰。在这项研究中，我们调查了与患者相关的混杂因素如何影响患者执行语言fMRI任务的能力（可行性），从图像对比的角度看任务敏感性以及表达性和接受性语言fMRI协议的解剖学特异性。在癫痫和脑肿瘤手术的术前程序中，有104例患者接受了语言功能磁共振成像检查。使用了四个任务：（1）口语流利（VF）任务来映射词汇使用情况；（2）语义描述（SD）任务来映射句子形成/语义整合技巧；（3）阅读理解（RC）任务；以及（4）听力理解（LC）任务。LC任务的可行性极好（100％），但其余任务的可接受性却处于中等水平（SD：87.50％，RC：85.57％，VF：67.30％）。VF的年龄（p = 0.020）和受教育程度（p = 0.003），SD的受教育程度（p = 0.004）和病灶偏侧性（p = 0.019）和年龄（p = 0.001）受病灶显着地混淆了可行性RC中的偏侧性（p = 0.007）和病变严重程度（p = 0.048）。就产生统计学上显着的图像对比度的敏感性而言，所有任务都是可比的（VF：90.00％，SD：92.30％，RC：93.25％，LC：88.46％）。病变的叶（p = 0.005）和年龄（p = 0.009）分别在VF和SD任务中混淆了对比敏感度。VF和LC任务均显示了后方语言区域的单侧偏侧。仅LC任务显示前语言区域的单侧偏侧化。我们的研究强调了与患者相关的混杂因素对语言功能磁共振成像的影响，并提出在临床术前背景下，LC是最可行，混杂程度最低且有效的分支任务。