Health Imaging News | May 2, 2008 | Clinical Studies
The interpretation of chest radiographs for lung nodules can be improved using an automated computer-aided detection (CAD) nodule detection system, according to a prospective study published in the May issue of Academic Radiology.
Edwin J. R. van Beek, MD, PhD, and colleagues from the department of radiology at Carver College of Medicine in Iowa City, Iowa, sought to assess the performance of a real-time interactive pulmonary nodule analysis system for evaluation of chest digital radiographic (DR) images in a routine clinical environment.
The researchers used a real-time interactive pulmonary nodule analysis system for chest DR image softcopy reading—IQQA®-Chest from EDDA Technology in Princeton Junction, N.J.—with a PACS in a National Cancer Institute-designated cancer teaching hospital. Patients referred for follow-up of known cancer underwent chest DR, the authors wrote.
The investigators noted that resident radiologists, along with experienced chest radiologists, read the posteroanterior and lateral DR images using a PACS workstation. Subsequently, they applied the CAD program to the posteroanterior DR images, and changes (if any) in diagnosis were recorded.
Van Beek and colleagues performed a follow-up chest radiograph at least six months following the initial examination, or a follow-up CT scan of the chest within three months was performed to establish diagnostic accuracy.
Of 324 DR exams, the researchers performed follow-up imaging for 214 patients (67 percent) according to the parameters available.
Notably, the investigators found lung nodules in the initial group and subsequently in 35 patients (10 percent) without CAD. Using CAD, nodules were found and subsequently confirmed in 51 patients (15 percent), improving sensitivity from 63.8 percent to 92.7 percent.
Van Beek and colleagues said that the nodules were subsequently proved to be malignant in five of the 16 additional cases (31 percent). False-positive readings increased from three to six cases; specificity decreased from 98.1 percent to 96.2 percent, which is not statistically significant. There were 153 true negative cases (71.4 percent), according to the researchers.
The authors also noted that the improvement in reader performance comes with a minimal number of false-positive interpretations.
“The present study demonstrates that the use of a CAD program to assist in the interpretation of chest radiographs can enhance the diagnostic performance of radiologists,” according to van Beek et al.