Radiologists Actively Search Less than Half of Lung on CT

Radiologists appear to actively search less than half of the lung when looking for lung nodules on CT images, according to a study published in the journal Radiology.

Researchers from Duke University in Durham, NC, Stanford University School of Medicine in Stanford, CA, and the University of Toronto, in Ontario, Canada, undertook a study that used eye tracking to determine the effectiveness of radiologists’ search, recognition, and acceptance of lung nodules on CT images.

The study included 13 radiologists who interpreted 40 lung CT images with an average of 3.9 synthetic nodules (5-mm diameter) embedded randomly in the lung parenchyma. The researchers used a remote eye tracker to record time-varying gaze paths as the radiologists read the images. Gaze volumes (GVs) were defined as the portion of the lung parenchyma within 50 pixels (approximately 3 cm) of all gaze points.

The radiologists were instructed to identify all pulmonary nodules within each case and were informed that they could adjust window width and level if desired, although none did.[[{"type":"media","view_mode":"media_crop","fid":"29241","attributes":{"alt":"Example CT images.","class":"media-image media-image-right","id":"media_crop_9246234752760","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"3032","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"height: 200px; width: 200px; border-width: 0px; border-style: solid; margin: 1px; float: right;","title":"Example CT images show four of the 157 embedded targets. The targets are at the center of the red circles. The number in the upper right corner of each image indicates how many of the 13 readers detected the corresponding nodule. Image courtesy of Radiology. ©RSNA, 2014.","typeof":"foaf:Image"}}]]

The results showed that detected nodules were within 50 pixels of the nearest gaze point for 990 of 992 correct detections. It was found that, on average, the radiologists searched 26.7% of the lung parenchyma in three minutes and 16 seconds, and encompassed between 86 and 143 of 157 nodules within their GVs.

The average sensitivity of nodule recognition and acceptance ranged from 47 of 100 nodules to 103 of 124 nodules, once they were encompassed with the GB. Overall sensitivity ranged from 47 to 114 of 157 nodules and showed moderate correlation with the fraction of lung volume searched.

“The detection of lung nodules on CT images by experienced interpreters requires a combination of effective search, recognition, and decision making,” the authors wrote. “Our experiment shows that when the center of a radiologist’s gaze is never closer than 50 pixels from a lung nodule, there is a less than 1% likelihood that the nodule will be detected.”