Database of marks of the sulci on the endocasts and brains obtained on the same sample of 75 volunteers

75 patients (36 females, 39 males, age range: 18–75 years, Table S1) were recruited during the PaleoBRAIN project for complementary biological and behavioral experiments. MRI acquisitions took place in the Center for Neuroimaging Research, Brain Institute, Pitié-Salpêtrière Hospital, Paris, France in 2022 and 2023. The ‘Comité de Protection des Personnes Sud-Méditerranée II’ approved the research protocol for the imaging center that was used in this study (comity reference 221 B38, identification number 2021-A02404-37). All volunteers underwent T1 and ultrashort time-to-echo (UTE) sequences. Those data were used to reconstruct 3D models of the brain and endocast of each of the volunteers. The final step of the protocol is the identification and labelling of the sulci observed on the endocasts. To achieve this, the automatic identification functionality of brain sulci within the software BrainVISA (Anatomist) was used as a reference. In Blender, each tube corresponding to a specific type of sulcus was selected, separated and isolated from the others so a color code could be applied to it. The endocast of each volunteer was reconstructed using the image obtained with an ultrashort time-to-echo (UTE) MRI sequence following our previously published experience (Labra et al., 2024). The segmentation was made automatically using the Morphologist toolbox from BrainVISA (Cointepas et al., 2001; Rivière et al., 2009; Fischer et al., 2012). Nevertheless, the obtained models for 26 of the 75 specimens were not satisfactory because of the lack of contrast between the soft tissues and the bone and the difficulty encountered by the automatic process to reconstruct the precise interface. In those cases, we performed manual segmentation using avizo software. This consisted of selecting an intermediate threshold value between the values observed for the bone and those of the tissues in contact inside the cranial vault. This setting had to be modified according to the variations observed for these different tissues in order to obtain as much as possible the precise characterization of the interface. Slightly less detail is visible on the surface of manually obtained endocasts compared to those obtained automatically. Nevertheless, this allowed us to reconstruct all endocasts for a broad comparative analysis. In the same imaging session at the Brain Institute, a T1-MPRAGE image was acquired. It was used to segment the brain, to generate the pial surface mesh and to extract and automatically label the sulci (Labra et al., 2024). Those procedures were also conducted using the Morphologist toolbox from BrainVISA (Cointepas et al., 2001; Fischer et al., 2012; Labra et al., 2024). In order to facilitate our determination, we used the automatically assigned sulcal labels as used by Morphologist (Perrot et al., 2011). This nomenclature was simplified to facilitate the work of description of those traits on endocasts where the diversity of the sulci does not appear in its completeness. This allowed us to focus on the main imprints sought on endocasts, i.e. those which separate the different lobes, which border important functional areas or which have been regularly found to be visible on endocasts. The final step of the protocol was the identification and labelling of the sulci observed on the endocasts. To achieve this, the automatic identification functionality of brain sulci within the software BrainVISA (Anatomist) was used as a reference. In Blender, each tube corresponding to a specific type of sulcus was selected, separated and isolated from the others so a color code could be applied to it. It corresponds to 75 zbrush files including marks on the endocast related to brain sulci, MNAS (for endocranial Marks Not Associated with Sulci) and sulci on the brain of the same sample of volunteers. The corresponding 3D data of the endocasts and brains are available in another online database (https://doi.org/10.48579/PRO/KZMMLM)

75名受试者（女性36名，男性39名，年龄范围18~75岁，详见补充表S1）于PaleoBRAIN项目期间招募，用于配套的生物学与行为学实验。MRI数据采集于法国巴黎皮提耶-萨尔佩特里耶医院脑研究所神经影像研究中心，完成时间为2022年至2023年。本研究使用的成像方案已获“Sud-Méditerranée II人体研究保护委员会”批准（委员会编号221 B38，识别号2021-A02404-37）。所有志愿者接受了T1序列及超短回波时间（ultrashort time-to-echo, UTE）序列扫描，所得数据用于重建每位志愿者的大脑与颅内模（endocast）。本研究方案的初步环节为识别并标注颅内模上的脑沟：为此，我们以BrainVISA软件（Anatomist模块）内置的脑沟自动识别功能作为参照，在Blender中逐一选中、分离并独立出对应特定脑沟类型的管状模型，以便为其赋予专属颜色编码。每位志愿者的颅内模基于我们此前已发表的研究经验（Labra等，2024），利用超短回波时间（UTE）MRI序列所得图像重建而成。分割操作最初通过BrainVISA的Morphologist工具箱自动完成（Cointepas等，2001；Rivière等，2009；Fischer等，2012）。但75例样本中有26例的重建模型效果不佳，原因是软组织与颅骨间对比度不足，且自动重建流程难以精准重构两者的界面。针对此类情况，我们使用Avizo软件进行手动分割：选取颅骨与颅腔内接触组织的灰度阈值中间值，并根据不同组织的灰度差异调整该阈值，以尽可能精准地界定界面。手动重建的颅内模表面细节略少于自动重建的模型，但这使得我们得以完成全部颅内模的重建，用于后续的大规模比较分析。在脑研究所的同一次成像扫描时段，我们还采集了T1-MPRAGE图像，用于分割大脑、生成软脑膜表面网格，并自动提取并标注脑沟（Labra等，2024）。该流程同样通过BrainVISA的Morphologist工具箱完成（Cointepas等，2001；Fischer等，2012；Labra等，2024）。为便于后续标注工作，我们采用了Morphologist自动分配的脑沟命名体系（Perrot等，2011）。由于颅内模上的脑沟多样性无法完全展现，我们对该命名体系进行了简化，以聚焦于颅内模上的主要脑沟印记：即分隔不同脑叶、毗邻重要功能区，或在颅内模上可稳定观测到的脑沟。本研究方案的最终环节为识别并标注颅内模上的脑沟：为此，我们以BrainVISA软件（Anatomist模块）内置的脑沟自动识别功能作为参照，在Blender中逐一选中、分离并独立出对应特定脑沟类型的管状模型，以便为其赋予专属颜色编码。本数据集包含75个ZBrush文件，内容涵盖与脑沟相关的颅内模标记、MNAS（颅内非脑沟相关标记，endocranial Marks Not Associated with Sulci），以及同批次志愿者大脑的脑沟相关数据。颅内模与大脑的对应3D数据已上传至另一在线数据库（https://doi.org/10.48579/PRO/KZMMLM）