SHARM: Segmented Head Anatomical Reference Models

Essam A. Rashed*, Mohammad Al-Shatouri, Ilkka Laakso, Sachiko Kodera, Akimasa Hirata

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Reliable segmentation of anatomical tissues in the human head is a crucial step in various clinical applications, including brain mapping, surgery planning, and computational simulation studies. Segmentation involves identifying different anatomical structures by labeling various tissues using different medical imaging modalities. While the segmentation of brain structures has seen significant progress, non-brain tissues receive less attention due to anatomical complexity and difficulties in observation using standard medical imaging protocols. The lack of comprehensive head segmentation methods and large segmented datasets limits variability studies, particularly in computational evaluations of electrical brain stimulation (neuromodulation), protection from electromagnetic fields, and electroencephalography, where non-brain tissues are essential. To address this gap, this study introduces the open-access Segmented Head Anatomical Reference Models (SHARM), comprising 196 subjects. These models are segmented into 15 different tissues: skin, fat, muscle, skull cancellous bone, skull cortical bone, brain white matter, brain gray matter, cerebellum white matter, cerebellum gray matter, cerebrospinal fluid, dura, vitreous humor, lens, mucous tissue, and blood vessels. The segmented head models are generated using the open-access IXI MRI dataset and a convolutional neural network structure named ForkNet+. Results indicate high consistency in the statistical characteristics of different tissue distributions across ages compared to real measurements. Electromagnetic exposure studies demonstrate variability in specific absorption rate (SAR) values among subjects. SHARM is expected to be a valuable benchmark for electromagnetic dosimetry studies and various human head segmentation applications.

Original languageEnglish
Article number107481
Number of pages10
JournalBiomedical Signal Processing and Control
Volume104
DOIs
Publication statusPublished - Jun 2025
MoE publication typeA1 Journal article-refereed

Keywords

  • Brain segmentation
  • Convolutional neural networks
  • Human head models
  • MRI

Fingerprint

Dive into the research topics of 'SHARM: Segmented Head Anatomical Reference Models'. Together they form a unique fingerprint.

Cite this