1. Advantages and disadvantages of CTP:
Fig. 1 Fig. 2
2. Acquisition:
Fig. 3 Fig. 4 Fig. 5 Fig. 6 Fig. 7
3. Theoretical background [1]
a. Nondeconvolution:
Fig. 8
b. Deconvolution:
Fig. 9 Fig. 10
Another important mathematical theory is delay and dispersion
Fig. 11 Fig. 12 [2]
These different theories lead to substantial differences between the commercially available software that may drastically affect the results of processing [3].
Radiologists should be familiar with the type of postprocessing used,
along with the limitations to avoid misinterpretation.
4. Postprocessing:
Fig. 13
5. Interpretation:
a. Tissue attenuation curve (TAC):
Fig. 14
b. Parameters:
Fig. 15 Fig. 16
i. Cerebral blood flow (CBF):
Fig. 17 Fig. 18
ii. Cerebral blood volume (CBV):
Fig. 19
iii. Time to peak (TTP):
Fig. 20
iv. Time to maximum (Tmax):[4]
Fig. 21
v. Time to start (TTS):
Fig. 22
vi. Mean transit time (MTT): [5]
Fig. 23
vii. Time to drain (TTD): [6].
Fig. 24
viii. Flow extraction products (FEP):[7]
Fig. 25
c. Analysis:
There are two approaches to interpretation:
i. Qualitative analysis:
Uses subjective judgment based on personal experience and comparison between the right and left hemispheres and different brain regions.
This approach is practical for daily routine,
but requires experience Fig. 26.
ii. Quantitative analysis:
Using parametric analysis to reach a conclusion.
This approach is more objective,
but drawing regions of interest (ROIs) is time consuming,
and there is no constant method used in the literature; therefore,
this approach is not always suitable for daily routine,
although different software packages offer more automated methods of quantification,
such as ROI mirroring,
predefined templates,
and automated estimation of the penumbra and infarction core Fig. 27.
1. Absolute values:
using thresholds to define abnormal regions Fig. 28 [8].
2. Relative values:
comparing the suspected pathological regions with normal contralateral brain tissue using interhemispheric differences and ratios.
6. Clinical situations that may affect brain perfusion:
CTP has many indications and perfusion changes may be observed in different clinical situations.
Therefore,
knowledge of such situations is important for radiologist and clinicians reading CTP.
a. Vascular:
i. Stroke
Fig. 29 Fig. 30 Fig. 31
ii. Large vessel occlusion and stenosis:
Perfusion changes can be observed in chronic carotid stenosis.
When there is a low flow state,
the arterioles will be maximally dilated and autoregulation will be exhausted,
causing CBF to be decreased below normal values.
This state of “misery perfusion” puts the brain at risk of infarction.
The perfusion changes resolved after revascularization therapy [9].
iii. Cerebral hyperperfusion syndrome (reperfusion injury):
Occurs usually in the first days after revascularization therapy of carotid stenosis,
and in minority of cases,
it becomes symptomatic (triad of ipsilateral headache,
focal seizures,
neurological deficit).
The mechanism appears to be the postoperative increase in systemic blood pressure and the impaired cerebral autoregulation in the previously hypoperfused brain parenchyma.
Conventional imaging shows ipsilateral edema and cortical swelling,
and CTP shows increased CBF and CBV and decreased MTT.
iv. Migraine:
Fig. 32
v. Vasospasm /delayed cerebral ischemia:
Occurs 3-4 days after subarachnoid hemorrhage.
TTD is considered one of the most sensitive parameters [10] Fig. 33.
CTP can identify the extent as well as the degree of ischemia[11] Fig. 34.
vi. Posterior Reversible Enecphalopathy Syndrome (PRES): [12]
Fig. 35
vii. Hypoxic ischemic injury after resuscitation:
Fig. 36
viii. Brain death:
The functional information acquired by CTP can augment the anatomical information obtained by the 2-phase CTA,
which will reduce the false positive results and increase the sensitivity.
Reduction of CBV and CBF in the brain stem reflects the absence of brain stem reflexes clinically [13].
N.B.: Pronouncing brain death differs among countries and even institutions,
and there are usually specific criteria according to medicolegal and ethical guidelines and certain procedures to be followed.
ix. Cerebral venous thrombosis:
Fig. 37
x. Sinking skin flap:
Impaired hemodynamic causes prolonged MTT and reduced CBF and CBV in structures adjacent to the decompressive craniectomy,
which improves after cranioplasty [14].
xi. Variants,
anomalies and malformations:
1. Normal variants: [15]
Fig. 38
2. Developmental venous anomaly (DVA):
Fig. 39
3. Arteriovenous malformation (AVM): [16]
Fig. 40 Fig. 41
b. Traumatic: [17]
Fig. 42 Fig. 43 Fig. 44
c. Neoplastic: [18]
Fig. 45 Fig. 46
d. Degenerative:
Normal pressure hydrocephalus (NPH):
Fig. 47
e. Inflammatory and infectious diseases:
Different entities can cause perfusion changes such as encephalitis and abscess.
f. Metabolic:
MELAS:
Fig. 48
g. Others:
i. Epilepsy:
Fig. 49
ii. Crossed cerebellar diaschisis:
Interruption of corticopontocerebellar tracts,
regardless of the cause of this interruption (e.g.,
ischemia,
hemorrhage,
neoplasia,
status epilepticus,
migraine with aura),
causes a reduction in the excitatory input as well as a decrease cerebellar flow[19].
CBF and CBV may be reduced,
and temporal parameters prolonged contralateral to the lesion.
N.B.: Different hemodynamic effects can play a role in the same disease.
For example,
Sturge weber syndrome may show decreased perfusion due to impaired vascular drainage or due to impaired metabolism.
Seizure activity of the involved cortex may contribute to the changes as well [20].
Fig. 50 summarizes the conditions that may affect perfusion.
7. CTP in the era of artificial intelligence (AI):
With the increasing use of AI in radiology,
understanding the possibilities of this powerful tool as well as the factors that may affect its performance is crucial for radiologists.
Regarding CTP,
there are plenty of potential,
some of these are already under investigations;
a. Acquisition:
i. Improvement of image quality by decreasing noise
ii. Reduction of radiation dose while maintaining the same quality
iii. Reduction of contrast dose
b. Post-processing:
Use of AI to substitute for the conventional algorithms (i.e.,
convolution and nondeconvolution methods) to achieve more accurate parameters.
c. Workflow:
Use of AI in triaging acute scans for urgent review
d. Interpretation:
i. Improvement of diagnostic accuracy
ii. Detection of perfusion abnormalities
iii. Characterization of perfusion abnormalities: whether there is hypo- or hyperperfusion,
the extent of the changes as well as the degree of hypoperfusion (i.e.,
critical or hemodynamic irrelevant hypoperfusion).
iv. Lesion segmentation,
which allows subsequent follow-up monitoring and can guide biopsy or intervention.
v. Prediction:
Prediction of outcome,
especially in stroke patients,
according to perfusion and other clinical and imaging data,
is helpful in decision-making.
vi. Reaching an improved understanding of complex diseases.
However,
it is important to acknowledge certain difficulties that may affect the performance of using AI,
such as the following:
a. The complexity of the techniques used.
b. The diversity of results according to the technique,
vendor and different algorithms used for postprocessing.
c. Diversity of results according to inherent patient factors,
such as age,
sex or cardiovascular function.
d. The complexity of the pathomechanism of the disease processes
e. The variety of the diseases that may cause perfusion changes and even similar patterns,
such as stroke and stroke mimics.
f. Medicolegal aspects
These features necessitate an improved understanding of the methods used (both perfusion analysis and machine learning algorithm),
as well as engagement of radiologist with enough expertise to provide a correct input to ensure optimal results that could be used in clinical settings and help in the diagnosis and patient management.