Work

A systematic review on software tools for CT- and MRI-based body composition analysis

Year: 2025
Type: review
Study design: Systematic Review
Abstract: Body composition analysis (BCA) aims to quantify the quantity, distribution, and quality of different compartments of skeletal muscle and adipose tissue. This information is relevant when assessing patients for possible sources of generalized vulnerability, particularly older adults and cancer patients. Manual segmentation of muscle and adipose tissue measured with computed tomography (CT) and magnetic resonance imaging (MRI) is time-consuming and cumbersome and, therefore, several software tools for image segmentation have been developed. BCA software tools have varied features. Therefore, a systematic literature review, conducted according to PROSPERO principles, was performed to identify full-text publications that utilized the available BCA software tools in order to collect information on all of the identified tools. The review was combined with information from a structured questionnaire sent out to the developer of the software in question. Information collected included imaging modalities, areal/volumetric assessment, degree of automation, platform, system specification, license, open source, retraining of AI models using external data, and clinical certifications. The literature search identified 1857 search results. First, duplicates and papers for which full text was not available were excluded (n=642). Second, manual screenings of title, abstract, and full text were performed (n=1118) to determine whether publications identified relevant software. In total, 97 full-text publications were included. Forty-six BCA software tools were found in the systematic literature review. A web search was performed to find websites for the software tools, during which three additional software tools were identified. A website was found for 42 of the 49 identified software tools, resulting in the inclusion of a total of 42 publicly available software tools. The following features were captured for the 42 software tools. Imaging modality accepted by the software was CT only in 19, MRI only in 3, and both in 20. Areal only assessment (2D) was available in 15, volumetric (3D) in 7, and both in 20. The degree of automation of segmentation varied, the highest being automatic in 27, semiautomatic in 14, and manual in 1 case. Retraining based on external data features was available in 11. Downloadable software was available in 23, web application in 9, and combined in 4. Sixteen software tools were commercial and 26 open access, of which 22 were open source. CE-marking was granted for 3 and FDA approval for 5. There are multiple CT- and MRI-based BCA software tools available for research settings, but the selection of software tools for the clinical environment is smaller. Most software tools lack clinical certification. The software tools have their pros and cons that should be considered when selecting software to apply to research or clinical questions.
Author: Österlund, Aron
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W6996951524
FWCI:
Cites: 64
Cited by:
Related to: 10