An Investigation of Radiologists' Perception of Lesion Similarity

摘要: Rationale and Objectives We conducted an observer study to investigate whether radiologists can judge similarities in pairs of breast masses lung nodules consistently reproducibly. Materials Methods Institutional review board approval informed consent were obtained. This was compliant with the Health Insurance Portability Accountability Act. used eight on mammograms computed tomographic images, for which subjective similarity ratings ranging from 0 1 determined our previous studies. From these, four sets image created (ie, a set mass pairs, nodule two mixed pairs). Eight radiologists, including chest compared all combinations each using two-alternative forced-choice (2AFC) method determine ranking scores by identifying pair more similar than other based overall impression diagnosis. Results In set, relationship between average 2AFC indicated very high correlations (r = 0.91 0.88). Moreover, sets, also 0.90 0.98). Thus, able compare lesions consistently, even unusual comparison completely different types lesions. Conclusion The medical images be quantified group radiologists. concept subjected rigorous scientific research investigation future.