Radiologists’ experience allows them to unconsciously detect lung nodules on CT, from RSNA 2016.
Unlike radiologists, medical students do not show unconscious detection of lung nodules, according to a study presented at RSNA 2016.
Researchers from Massachusetts studied the unconscious detection of lung nodules among radiologists and evaluated eye movement metrics in untrained readers (1st year medical students).
Twenty-four subjects participated in the study, 12 radiologists and 12 medical students. They interpreted nine normal and nine abnormal axial chest CT scans. The abnormal scans contained 16 lung nodules. Eye tracking was followed for location and duration of gaze, while visual dwell time on healthy tissue versus on a lung nodule, the number of total eye movements (saccades), and the total number of images viewed were used to evaluate the efficiency of visual search patterns by both groups.
The results showed that both radiologists and students dwelled longer on the nodules, compared with healthy lung tissue. The radiologists also dwelled longer on the nodules when they were not consciously detected. However, the medical students did not fixate longer on a lung nodule versus healthy lung tissue when not consciously detected.
The radiologists scrolled through the image set 2.5 times more than the students and made significantly more saccades than the student, with an average of 376 versus 215, respectively. However, radiologists were significantly more efficient, making on average 0.46 saccades per image while the students made 0.62. The students also bounced from one location to another across the entire image set before moving on from that image, and only rarely returned to an image they looked at previously.
The researchers concluded that unlike radiologists, medical student do not show unconscious detection of lung nodules, with a search pattern and efficiency of search significantly worse for the students, compared with the radiologists. “These data suggest that during the process of radiological training, both conscious and unconscious learning is developed that influence the success of the search, the efficiency of the search, and the pattern in which the search is undertaken,” they wrote. “Although some component of radiological learning is the result of specific training and conscious processes, additional unconscious learning likely occurs that influences radiological performance.”
What is the Best Use of AI in CT Lung Cancer Screening?
April 18th 2025In comparison to radiologist assessment, the use of AI to pre-screen patients with low-dose CT lung cancer screening provided a 12 percent reduction in mean interpretation time with a slight increase in specificity and a slight decrease in the recall rate, according to new research.
The Reading Room: Artificial Intelligence: What RSNA 2020 Offered, and What 2021 Could Bring
December 5th 2020Nina Kottler, M.D., chief medical officer of AI at Radiology Partners, discusses, during RSNA 2020, what new developments the annual meeting provided about these technologies, sessions to access, and what to expect in the coming year.
Can CT-Based AI Radiomics Enhance Prediction of Recurrence-Free Survival for Non-Metastatic ccRCC?
April 14th 2025In comparison to a model based on clinicopathological risk factors, a CT radiomics-based machine learning model offered greater than a 10 percent higher AUC for predicting five-year recurrence-free survival in patients with non-metastatic clear cell renal cell carcinoma (ccRCC).
Could Lymph Node Distribution Patterns on CT Improve Staging for Colon Cancer?
April 11th 2025For patients with microsatellite instability-high colon cancer, distribution-based clinical lymph node staging (dCN) with computed tomography (CT) offered nearly double the accuracy rate of clinical lymph node staging in a recent study.