Measurements Discrepancy from x-ray images

Hello Community,
I have three x-ray images (exported as JPEG) taken from the same patient. those three images are taken at different positions:1. patient standing erect 2. patient performing left lateral bending 3. patient performing right lateral bending.
I am trying to measure the length of the spine using ImageJ, and I am converting the measurement from pixels to millimeters using the annotated ruler (real-life scale).
ideally, I am supposed to get very similar spine length, however, I am getting about 60-millimeter difference between the erect and the lateral bending and a difference of about 17 mm between the right and the left lateral bending.
please guide me what might be the reason for the discrepancy? what should be fixed?

Thank you,

1 Like

Hey @Ausi,

welcome to the forum and thanks for posting these interesting images. As my radiology lessons are some time ago, I just tried to measure the length of the thoracic spine. I did so by finding the last vertebra where a rib starts in caudal direction. Then, I marked 12 vertebrae in cranial direction:

Then I measured the length of these three lines and the length of the scale resulting in six length measurments given in pixels:

When I calculate the length in mm, I get 270mm, 307mm and 307mm for the three images. Thus, between the last two images, the length measurement is reproducible - compared to the first image not. Furthermore, the first image also looks a bit different in contrast, zoom and the labels on the image are different.

Thus, some counter questions:

  • Was the patient standing at the same position in all three images?
  • Have all three images been imaged with the same X-ray device?
  • What’s the time between the measurements? I assume it’s a child - if there is a year between these images, it might have grown?
  • Have similar images been taken from a different perspective? The spine may bend unnaturally in view direction of these images as we are obviously looking at scoliosis…

Just guessing… :wink:

I hope that helps.


1 Like

Hi Robert,

Thank you very much for your detailed response.

The answers to the questions:

  • No. the patient in the erect image is standing and at a distance of 72 inches from the detector while in both bending images (right and Left) the patient is lying on the table and the detector is at a 52 inches from the patient.

  • The erect image was imaged using different x-ray device than the lateral bending images.

  • All images were taken on the same day

I wonder shouldn’t the scale take into account the distance of the object from the detector?

Thank you :slight_smile:

1 Like

Hey @Ausi,

can you check with the operator of the machine if the scale was used correctly? You could also take some images of a piece of wood of 200mm length to check if the scale is correct…


Hi Ausi,

there are two reasons why you can get different results for the length of the spine:

  • X-ray images are taken with the rays coming from one point (the x-ray source), not with parallel rays, so the scale depends on the distance from the x-ray source.
  • If the spine (or part of it) is not perpendicular to the image plane, it will appear shorter by a factor cos(α), with α being the angle between the spine and the image plane (the plane perpendicular to the rays from the source to the image detector).

– Michael