Evaluating a decision support system for patient image pre-fetching: An experimental study

When reading images from a newly taken radiological examination, a radiologist often needs to reference relevant prior images of the same patient to confirm a preliminary diagnosis, compare suspicious radiographic signs, or evaluate the progression of a known underlying pathological process, injury, or abnormality. To mitigate the stress and time requirements for the reading radiologist's image searches, some healthcare organizations have taken a pre-fetching approach to make relevant patient prior images conveniently accessible. Motivated by the importance of patient-image pre-fetching to radiologists' examination readings, as well as by the limited scope and ad hoc evaluation of most previously reported systems, we develop the Image Retrieval Expert System (IRES) and experimentally evaluate its effect on the radiologist's examination-reading efficiency, service quality, and satisfaction using the current pre-fetching practice of the studied organization as a benchmark. Our overall analysis suggests that image pre-fetching has an important effect on radiologists' examination readings and that radiologists (including residents) can become more efficient, effective, and satisfied when supported by IRES pre-fetching than by the benchmark practice. However, the exact magnitude or statistical significance of the IRES-induced improvements may vary with examination category and/or radiologists' experience. Our findings have several important implications for research and patient image management practices, which also are discussed.