I could come across a few studies that perform AP/PA Chest X-ray view classification using machine learning. However I'm puzzled if this is a truly noteworthy/real problem in the clinical setting. I would like to know whether the radiologists find it hard to distinguish between AP/PA in the clinical setting and an automated system is really required in this regard.
As a general rule in the adult population, PA films tend to be performed on patients that are clinically well who are able to stand up independently. AP films usually are reserved for bed bound or more unwell patients.
The use of a PA chest X-ray in the general population where ever possible is mainly to allow greater reproducibility in the measurement of the mediastinal and cardiac silhouettes. The use of PA chest x-rays has the benefit of reducing lordosis of the ribs etc which allows for a more diagnostic image.
When a PA chest x-ray is not possible a AP film will be taken however this will magnify the mediastinal and cardiac silhouettes as they are more anterior in the thorax, there is also increased risk of lordosis (a non-lordotic radiograph should have the clavicular heads at the level of the third rib) both of these factors can reduce the image quality or diagnostic value of the image which can lead to limited, or incorrect information being gleaned.
The principal reason is that the difference in position causes different amounts of magnification of cardiac size. With a PA chest x-ray, the heart is close to the image receptor, and magnification is minimal. With an AP chest x-ray, the heart is further from the image receptor, and appears larger on the image. When evaluating the possible presence of cardiomegaly, only a PA view should be used.
Donald L Miller Many thanks for the responses. I would like to know whether the radiologists find it hard to distinguish between AP/PA in the clinical setting and an automated system is really required in this regard.
There is also the bonus that in the PA view that the breast radiation dose is less.
- Badges
- Science topic