This has come up in Question 20 from the first paper of 2005,  again in  Question 13 from the second paper of 2010, and again in Question 24 from the first paper of 2015.The canonical source for this information would have to be the most recent iteration of the ACCF/AHA Guidelines for Diagnosis and Management of Patients With Thoracic Aortic Disease.

Imaging Modalities for the Evaluation of Aortic Dissection
Imaging modality Advantages Disadvantages
Aortogram
  • High sensitivity (86-88%) and specificity (75-94%)
  • Can detect blocked coronaries in Type A dissection
  • Can assess valves
  • May allow endoluminal repair during the same procedure, expertise permitting
  • Not easily available
  • Large contrast load
  • Time consuming
  • Ineffective in detecting intramural haematoma
    (the contrast cannot get in there!)
  • Potential for false negative results when a thrombosed false lumen prevents contrast entry
  • Slightly lower sensitivity and specificity than TOE, CT or MRI; has been largely replaced by them.
CT
  • Easily available
  • High sensitivity (83-94%) and specificity (87-100%)
  • Information about end-organ ischaemia
  • Imaging of the vascular tree allows planning of surgical or endovascular approach
  • ECG-gated CT = cardiac motion artifact is abolished
  • Able to exclude conditions which mimic aortic dissection
  • Contrast exposure
  • No information about the valves
  • Risk of transfer to CT
  • Motion artifact could be an issue in ungated studies
MRI
  • High sensitivity and specificity (95-100% for both)
  • Contrast is less nephrotoxic
  • Information about end-organ ischaemia
  • Imaging of the vascular tree allows planning of surgical or endovascular approach
  • Occasionally allows assessment of aortic valve pathology, coronary arteries and the LV
  • Significant risk of transfer
  • Not easily available
  • Certain patient groups excluded (eg. recent trauma with surgical staples)
  • Often fails to characterize the relationship of an intimal flap and aortic root structures, specifically the coronary arteries
TOE
  • Can assess valves
  • Decent sensitivity (35-80%) and specificity(39-96%)
  • Performed at the bedside- no risk of transfer
  • Contrast not required
  • Allows detection of tamponade
  • Allows assesment of proximal coronary arteries
  • Able to detect intramural haematoma
  • Invasive
  • Accuracy is operator dependent
  • Requires sedation
  • May cause hypertension
  • Limited by a blind spot caused by interposition of the trachea and left main bronchus between the oesophagus and aorta
  • Unable to visualise the abdominal aorta
CXR
  • Rapidly available
  • Immediate evidence of widened mediastinum
  • A completely normal CXR in low risk patients may be meaningful as a means of excluding dissection
  • Inadequately sensitive (~71%)
  • Rarely able to exclude dissection in most patients

Additional information:

  • In high risk patients, all the modalities are more or less equal in accuracy
  • In moderate risk patients, positive predicitive values are >90% for CT, MRI and TOE but only 65% for aortography
  • MRI is the most sensitive of the lot - in low risk patients, it picks up close to 100% of the dissections
  • All four modalities have a 85% negative predictive value

The ACCF/AHA guidelines make the following recommendations:

  • CT as opposed to echocardiography is the imaging modiality of choice, as it can identify aortic disease but it can also identify disease which mimics aortic disease.
  • If repeated imaging is to be performed as surveillance in a stable patient, MRI is recommended (to reduce radiation and contrast exposure)
  • CXR is inadequately sensitive and won't reliably exclude dissection in any but the lowest risk patients.

References

A good article on this topic is available:

Khan, Ijaz A., and Chandra K. Nair. "Clinical, diagnostic, and management perspectives of aortic dissection." Chest Journal 122.1 (2002): 311-328.

Nagy, Kimberly, et al. "Guidelines for the diagnosis and management of blunt aortic injury: an EAST Practice Management Guidelines Work Group." Journal of Trauma-Injury, Infection, and Critical Care 48.6 (2000): 1128-1143.