Home / Research Supervisor Connect / Opportunity details

Establishment of optimum viewing times for radiological images

Summary

The project will explore the optimum time period for identifying cancerous lesions on a radiographic chest
image.

Supervisor(s)

Professor Patrick Brennan

Research Location

Medical Imaging and Radiation Sciences Research Group

Program Type

Masters/PHD

Synopsis

Perception and interpretation factors are responsible for up to 60% of radiologic errors and approximately 10 million diagnostic radiological events will occur in Australia annually 1,2.  Whilst the emphasis of previous research and quality assurance procedures in radiology has been to consistently produce a high quality image, this is of little use, if important details relating to disease are not perceived or interpreted.   Critical to this interpretation is time spent studying the image.  Early work of Kundel and Nodine demonstrated that Az values of 0.71 could be achieved after only viewing the image for 200 milliseconds using global recognition strategies, whilst more recent work by Nodine, showed that in the longer search and identify phase, performance particularly for non-experts deteriorated with increasing time 3,4. Optimum time and associated observer search patterns are currently unknown for specific pathologies. The current work will employ five expert radiologists and five non-radiology doctors, who will examine 50 postero-anterior radiographic chest images, and in line with Australian Government Health Priorities 25 of these images will have 1, 2 or 3 cancer-type nodules.  Using software already developed for this study by our team (see: http://www.ucd.ie/diagnosticimaging/html/johnryan/sydney/Time%20Study%20-%20Chest.zip)we will present all 50 images for 6 different time-periods: 200, 500 and 100 milliseconds, 5, 25 and 40 seconds4.  Receiver operating characteristic analyses will be performed and area under the curve, sensitivity and specificity for each viewer at each time point will be calculated and compared. Also, using an eye tracker device, important parameters such as overall search patterns, dwell time, total observer time, fixation clusters and detection v decision events will be recorded and quantified for all viewers.  This University of Sydney research project will be run by a leading Imaging Group with 150+ publications in image optimisation and will provide important perceptual pilot data.  The student will be offered expert supervision and the opportunity to work with world leaders in the field from Sydney, US and Europe. 

  1. Beam CA.  The place of medical image perception in 21st-century health care.  J Am Coll Radiol. 2006 3:409-12. Review.
  2. United Nations Scientific Committee on the effects of Atomic Radiation (UNSCEAR) . Report of the United Nations Scientific Committee on the effects of Atomic Radiation to the General Assembly. 2001
  3. Kundel HL, Nodine CF.  Interpreting chest radiographs without visual search.  Radiology 1975 116: 527-532
  4. Nodine CF, Mello-Thoms C, Kundel HL, Weinstein SP.  Time course of perception and decision making during mammographic interpretation.  AJR 2002 179: 917-923

Additional Information

This project is scaleable and therefore available to graduate and undergraduate students interested in imaging research.  The candidate does NOT have to be an expert in the area of the study and training in research methods, statistical analyses and image evaluation techniques will be provided.  It is expected that the work will contribute to a publication in an important journal.

Want to find out more?

Contact us to find out what’s involved in applying for a PhD.

Contact Research Expert to find out more about participating in this opportunity.

Browse for other opportunities within the Medical Imaging and Radiation Sciences Research Group .

Keywords

visual perception, chest nodules, ROC

Opportunity ID

The opportunity ID for this research opportunity is: 865

Other opportunities with Professor Patrick Brennan