0

I read this phrase in "Curvelet based residual complexity objective function for non-rigid registration of pre-operative MRI with intra-operative ultrasound images." 2016:

In the intensity-based methods, proposing a proper similarity measure is the main challenge, particularly in multimodal image registration. In this area, due to different nature of two imaging modalities, objective functions which are used commonly such as correlation ratio, mutual information, and sum square differences are known to fail.

  • possible duplicate of english.stackexchange.com/questions/277141/… – miltonaut Jan 4 '17 at 8:25
  • I don't think so – Mohammad nagdawi Jan 4 '17 at 8:26
  • "Known to fail" in a scientific context can mean two things - (1) It indicates the "in-appropriateness" of applying the said (objective) functions/ test to the scientific problem being probed, or (2) It can also indicate that previous empirical data has proved the inefficiency of the said tests. – Monzoor Jan 4 '17 at 9:13
0

In this area, due to different nature of two imaging modalities, objective functions which are used commonly such as correlation ratio, mutual information, and sum square differences are known to fail.

In this context "Known to fail" is reinforcing that calculations/correlations made between DICOM images originating from two modalities are not always going to "add up" (1:1). So if you were to attempt sum square differences on images from Modality X and Modality Y, you're not going to get valid results every time. This is often due to noise and variation in luminance between images.

Source: I'm a picture archival and communication systems (PACS) engineer.

Some useful reading on correlations for Digital Imaging and Communications in Medicine (DICOM) https://en.wikipedia.org/wiki/Phase_correlation

2

"Be known to fail" is not a set phrase. The root phrase is "to be known to X" where X is a verb. In this case, that verb is "fail."

It means that something (a function, a machine, whatever) does not always succeed and that users are aware of this. They know it fails. Depending on the situation, it may fail often, frequently, or always.

In your quote, it sounds like objective functions always fail when dealing iwth intensity-based methods.

to be known to X - If something or someone is known to be or do something, people know that it is true or happens, or that someone is or does something

Something can be known to succeed, to taste bad, to be a liar...

2

The phrase "known to fail" just means "unreliable". So, in this context, sum of squares differences are an unreliable measure (of whatever is being measured).

Conversely, the old advertising tag-line, "never known to fail" means "completely reliable" (from the manufacturer's point of view).

0

A phrase such as the one you are asking about simply means that no matter what is practiced, it has bene proven to fail and not succeed under certain circumstances. Just like you would say "It was known that no one was able to climb that mount in 5 minutes", implies that climbing the mountain in 5 minutes was never possible and no one has done it. Hope this simplicity answer's your question!

Not the answer you're looking for? Browse other questions tagged or ask your own question.