Fabio Maggi MD, Presenter: Nothing to Disclose

Anna Rita Larici MD, Abstract Co-Author: Nothing to Disclose

Andrea Caulo MD, Abstract Co-Author: Nothing to Disclose

Lucio Calandriello, Abstract Co-Author: Nothing to Disclose

Ferdinando Ferrara, Abstract Co-Author: Nothing to Disclose

Lorenzo Bonomo MD, Abstract Co-Author: Nothing to Disclose

To assess detection and reading time of solid lung nodules on MDCT chest screening examinations using a CAD integrated in a PACS system as concurrent reader

50 chest MDCT examinations of asymptomatic smokers (age range: 50-70 years) were performed with a 16-row scanner (collimation: 1.25 mm). Two 10-years experienced chest radiologists identified solid uncalcified nodules, with diameter ranging between 2 and 30 mm, in consensus without CAD (reference standard). Exams were randomized on the basis of patient characteristics (age, gender and BMI) and nodule number in two groups (A and B). One radiology resident (reader 1) analyzed group A exams using as concurrent reader a CAD system (LMS, Median Technologies) integrated in the PACS workstation (Carestream Health) and group B exams without CAD. A second radiology resident (reader 2) examined group A without CAD and group B with the same integrated CAD-PACS system as concurrent reader. Readers recorded slice number and location of each nodule identified. For each exam reading time was reported. Performance in lung nodule detection was then compared to the reference standard. Reading time with CAD was compared to reading time without CAD for each exam.  

150 nodules were identified by the reference standard (Group A: 82; Group B: 68). Reader 1 and reader 2 identified correctly 123 nodules with CAD (Group A: 71; Group B: 52) and 113 nodules without CAD (Group A: 60; Group B: 53) with a sensitivity of 82% and 75% respectively, without a statistically significant difference (Mc Nemar, p = 0.14). Reading time resulted 9 minutes and 25 seconds per exam with CAD and 10 minutes and 54 seconds per exam without CAD with a statistically significant difference (Student’s T test; p = 0.004).

A CAD system used as concurrent reader significantly reduces reading time in lung nodule detection. In unexperienced readers a slight increase in sensitivity may be obtained.

In a screening population use of CAD as concurrent reader in lung nodule detection is helpful, particularly in unexperienced readers, reducing reading time and moderately increasing sensitivity.

Maggi, F, Larici, A, Caulo, A, Calandriello, L, Ferrara, F, Bonomo, L, Performance and Reading Time in Detection of Solid Lung Nodules Using a Concurrent Reader a CAD System Integrated in a PACS.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8015446.html