The importance of a high rate of nodule detection- i.e.
high sensitivity- is underscored by the fact that most failures in lung cancer diagnosis are due to errors of detection rather than interpretation [6,7].
Considering this in isolation,
any CT screening programme would want readers with the highest possible sensitivity.
In this context,
data from our study suggests that radiographers cannot be considered ideal first readers for lung screening,
since their performance in the majority of cases was statistically significantly lower than that of the radiologists.
However,
a radiographer can act as an aid to nodule detection rather than a first reader,
in the same way that computer-aided detection (CAD) does.
The advantage of having a radiographer rather than a computer as a detection aid is that interaction and discussion regarding missed or incorrect findings is possible with the radiographer,
whereas learning and “tuning” of a CAD system depends on artificial neural networks and algorithms that the radiologist is not privy to,
and has no influence over.
Caution should always be exercised when comparing sensitivities between nodule detection studies,
as differences in the method of reading,
derivation and stringency of the reference standard [8],
and in the types of patients undergoing CT scans (e.g.
patients with multiple metastases versus lung screening studies) can profoundly affect sensitivity.
Nevertheless,
it is reassuring that the mean performance of the radiographers and radiologists in our study was comparable to radiologists in the published literature [9-15].
Intuitively,
it might be expected that radiographers,
who have no experience in pulmonary nodule evaluation,
would tend to falsely interpret any abnormal pulmonary opacity as a pulmonary nodule,
and so increase their false positive rates.
As such,
it is not surprising that in the majority of cases,
radiographers had higher average false-positive rates than radiologists.
However,
these rates were comparable to,
and for the majority of radiographers lower than,
current CAD systems.
It is also noteworthy that the two radiographers with the lowest performance showed an improvement in their sensitivity between the first and second 10 weeks of the study,
and perhaps more importantly that this increased sensitivity did not carry the penalty of an increased false-positive rate.
It can be speculated that (a) the nodule detection ability of a radiographer relies on his or her inherent perceptual ability once some level of basic training has been provided,
and (b) it may be possible to improve suboptimal sensitivity with time and reading experience,
but this improvement will probably still be limited by the individual’s aforementioned inherent ability.
In summary,
this study demonstrates that radiographers’ performance in lung nodule detection is comparable to radiologists in the published literature but lower than that of radiologists reading the same scans,
and with higher false-positive rates.
Radiographers are thus not sensitive enough to be used as first readers in CT lung screening.