Approval by the local Independent Ethics Committee was not necessary due to the retrospective study design. The legal guardians of all patients provided written informed consent for the clinically indicated MRI examinations. The study population consisted of all consecutive patients who underwent at least one MRI examination during the first four months of life in our tertiary care University Medical Center from July 2009 until August 2022. We collected patient-specific data, including gender, body height and weight, prematurity, chronological and corrected age at the time of the MRI examination from the medical records. We applied corrected age for preterm (i.e., < 37 weeks of gestation) patients, calculated as chronological age minus the difference between the estimated and actual birth date, for the statistical analyses.

MRI examinations were performed using a 1.5 T scanner (Magnetom Avanto®; Siemens Healthineers) or a 3 T scanner (Magnetom Skyra® or Magnetom Vida®, Siemens Healthineers). Appropriate coils and standard protocols, depending on the indication and body region, were used as determined by the responsible radiologist. We retrospectively documented field strength, indication, examined body region, technique (deep sedation versus FWT), total number of sequences, use of contrast agent, and total examination length for further analysis. Total examination length was defined as the time from acquisition of the first localiser sequence until the end of the last performed sequence, including possible breaks due to awakening of the patient when using the FWT.

Examinations in the deep sedation group (shortened as “sedation” group throughout the further manuscript for better readability) were performed using intravenous propofol or inhalational sevoflurane (applicated via laryngeal mask) as determined by the anaesthesiologist [17]. Endotracheal intubation, if necessary, was carried out at the paediatric intensive care unit before the child was transported to the MRI scanner. Children were kept fasting prior to the examination according to national anaesthesiology guidelines [18]. In our institution, dedicated paediatric radiologists radiologically supervised neonatal MRI under sedation since 2009. For a larger sample size in the sedation group, we also included cases examined between 2009 and 2015 in our analysis, as the respective MRI protocols were equivalent to those in the FWT group.

The FWT was introduced in 2015 in our institution. The responsible radiologist decided whether to perform the examination by using the FWT or under sedation in consensus with clinical and anaesthesiology staff. A predetermined need for contrast agent application did not exclude an FWT attempt. Patients planned for MRI using the FWT were prepared according to a standardised scheme whenever possible. A radiologist explained the whole examination setup to the parents the day before the planned time slot. We instructed the parents to keep the child awake 3 to 4 hours before the examination. Approximately 30 minutes before the examination, the children were fed, diapers were changed, all metal objects were removed, and metal-free clothes were put on. Then, we wrapped/swaddled the child into blankets with a special focus on avoiding skin-to-skin contacts (e.g., between the arms and thorax, or between the legs). We used an MR-compatible vacuum mattress (MedVac VMR433X01N, Kohlbrat & Bunz GmbH) for immobilisation in the MR unit. MR-compatible earplugs and noise attenuators (“Mini Muffs”, Natus Medical Inc) were applied for noise reduction. An MR-compatible pacifier in combination with oral glucose (5%) was offered to calm the child [19]. Vital parameters were monitored using an MR-compatible pulse oximetry system (Nonin Sensors 7500 FO, Nonin Medical Inc) throughout the examination. Figure 1 demonstrates a typical setup in the MR preparation room and scanner.

immobilisation of a premature patient in the MR preparation room. The child is transferred from the incubator into the vacuum mattress, MR-compatible monitoring is applied (a) and a pacifier (b, red arrow) is used. After vacuum has been applied, the child is taken into the MR scanner (c)

All sequences were retrospectively analysed by two of the authors (one paediatric radiologist with more than 10 years of experience in paediatric MRI reading, AL, and one last-year medical student, KSF) in consensus. A 4-point semi-quantitative scale was used to assess image quality, with 4 being the highest score: A score of 1 indicated severe motion artefacts, resulting in nondiagnostic images. A score of 2 indicated an image quality with marked artefacts, partially limiting diagnostics. A score of 3 was given in the case of mostly good image quality with only minor artefacts in the area of interest, predominantly not limiting diagnostics. Excellent images without artefacts or diagnostic limitations were rated with a score of 4. Figure 2 demonstrates examples of the four categories. For multivariate analyses, a sequence rated with a score of 1 or 2 was called a sequence with diagnostic limitations. Mean scores were calculated for each examination.

Examples of the different quality scores based on a T2-weighted turbo spin echo sequence of the brain at the level of the basal ganglia. a Score of 4 with excellent image quality. b Score of 3 with slight motion artefacts, primarily affecting the frontal lobe. c Score of 2 with marked artefacts. Anatomic landmarks are still visible, but details, such as cortico-medullary differentiation, are only partially assessable. d Nondiagnostic image with severe artefacts, corresponding to a score of 1

In addition, we assessed further factors, indicating examination quality: number of repeated sequences (including reasons for repetition), whether the main clinical question was answered, discontinuation of the examination (including reasons for termination), and if the whole examination was repeated.

Statistical evaluations were performed with SPSS Statistics (Statistical Package for the Social Sciences, version 27.0; IBM) or R (version 4.3, The R Foundation of Statistical Computing). A minority of patients had multiple examinations in the determined age span. As none of these examinations were performed on the same day, they were modelled as stochastically independent. Descriptive statistics were computed for all clinical and radiological parameters. Group differences were assessed using the Mann-Whitney U test, or Fisher’s exact test, as appropriate. Spearman’s rho was calculated to correlate sequence quality with corrected age. We compared both techniques with regard to the occurrence of sequences with diagnostic limitations. To this end, we fitted a negative binomial regression model (for count data) to estimate the incidence rate ratio, adjusting for age, gender and body weight as well as for the examined body region and the field strength used. The incidence rate was defined as the proportion of sequences with diagnostic limitations (i.e. sequences with a quality score of 1 or 2) per examination. A p value < 0.05 was considered significant in all statistical analyses, which were exploratory. Therefore, the presented p values were descriptive.