269 Non-invasive imaging biomarkers 3 Uniting medical imaging with artificial intelligence Unlike tissue sampling procedures, medical imaging can provide information about the entire lesion, including intra- and interlesional heterogeneity [361], thereby circumventing the shortcoming of sampling error that may occur with FNAC. Visual interpretation of images consists of (qualitative) assessment of signal intensity (e.g., density, echogenicity, radiopharmaceutical uptake, apparent diffusion coefficient), location, size, shape, deformability (elastography), border (relation with surrounding tissues), patterns or vascularity (e.g., intravenous contrast enhancement, Doppler) of lesions. Medical imaging can stratify nodules before FNAC-procedures and thereby guide the choice of sampling location. Moreover, it can provide circumstantial evidence towards the nature of the nodule, such as suspicious cervical lymphadenopathy. Quantitative imaging Medical images contain much more information about the biology of the lesion hidden in the myriad of voxels of both lesions and healthy tissue than can be assessed visually by a human reader [362]. (Semi-)quantitative analysis of the images provides an objective complement to visual interpretation. The use of quantitative imaging in (multidevice) studies, and to a lesser extent in clinical management, requires adequate repeatability and reproducibility [363]. Repeatability refers to the likelihood of obtaining the same result in the same patient, when examined more than once on the same system. Reproducibility refers to the ability to yield the same results when that patient is examined on different systems and/or at different imaging sites. Ultimately, quantitative imaging enables the comparison of measurements in a single subject with normative values from a healthy population and permits the monitoring of subtle changes caused by the progression or remission of disease. PET, as no other, allows for (semi-)quantitative analysis [314]. The standardised uptake value (SUV, unit [g/mL]) expresses the ratio between the local activity concentration and the decay-corrected amount of injected radiotracer per unit of body mass. It indicates the radiotracer concentration factor in a specific region compared to homogeneous distribution of the radiotracer through the body. In case of the radiotracer [18F]FDG, the SUV is generally higher in malignant than in benign lesions. Nevertheless, the SUV is not only determined by tumour biology but also by preparative, procedural and postprocedural factors. The European Association of Nuclear Medicine (EANM) established guidelines for PET tumour imaging with the aim to achieve harmonisation in multicentre settings including accreditation programs (EARL) [363].
RkJQdWJsaXNoZXIy MTk4NDMw