Phantom studies in medical imaging (PSMI): a guide with recommendations and checklist.

Phantom studies are essential in medical imaging, offering a controlled and reproducible framework for evaluating imaging technologies across all modalities. Phantoms, whether physical (synthetic, biological, or mixed) or computational, simulate human tissues or anatomical structures and serve roles in technology validation, performance benchmarking, protocol optimization, quality assurance, and artificial intelligence development. We provide recommendations for designing and conducting phantom studies in medical imaging (PSMI). Key aspects include phantom selection, image acquisition protocols, and analysis strategies, particularly when image quality is evaluated in relation to radiation dose or contrast agent optimization. Quantitative image analysis is considered with emphasis on signal-to-noise ratio, contrast-to-noise ratio, and spatial resolution (e.g., modulation transfer function). Qualitative assessment is addressed considering reader selection and training, blinding, randomization, and use of absolute or relative Likert scales. Brief recommendations for sample size calculation, data reporting, and statistical analysis are provided, covering continuous/ordinal data, inter-rater agreement, and group comparisons. A checklist is provided to allow authors to document adherence to these recommendations and to identify shortcomings, limitations, and weaknesses in their phantom studies. The PSMI checklist is proposed to promote transparency, reproducibility, and critical appraisal, containing 25 items regarding: title/abstract (1, 2); background/introduction (3); methods/study design (4); methods/phantom description (5-7); methods/imaging protocol (8, 9); methods/image analysis (10, 11); methods/statistics (12-15); results/quantitative analysis (16, 17); results/qualitative analysis (18); results/tables and figures (19); discussion (20-23); and conclusions (24, 25). Finally, the importance of maintaining a clinical perspective is underscored, highlighting how well-designed phantom studies can inform, but not replace, clinical validation. RELEVANCE STATEMENT: This paper provides comprehensive recommendations for designing and conducting PSMI. The use of the PSMI checklist may contribute to increasing the quality of phantom studies. KEY POINTS: Phantom studies provide controlled, reproducible evaluation of imaging technologies. Phantoms simulate human tissues for validation, optimization, and AI development. Good design includes proper phantom selection and analysis strategies. Clinical relevance must guide interpretation; phantoms cannot replace clinical validation. The proposed 25-item PSMI checklist supports transparent and reproducible phantom study reporting.