Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Doctor of Philosophy


Medical Biophysics


Lee, Ting-Yim


CT perfusion (CTP) imaging is a validated treatment decision support tool in acute ischemic stroke. Automated analysis of CTP cerebral blood flow (CBF) and Tmax maps produces estimates of ischemic core and penumbra volumes used to determine target mismatch profiles for treatment. However, availability and utilization of CTP is low due to diagnostic variability between CTP software and technical, logistical, and radiation dose considerations that may limit its routine adoption. The objective of this doctoral research was to improve the reliability and accessibility of CTP by (1) improving diagnostic agreement between CTP software, (2) enabling perfusion imaging with standard acute stroke CT studies, and (3) reducing CTP radiation dose.

Mismatch profiles may disagree between software due to different optimal stroke lesion thresholds. In Chapter 2, CTP thresholds were calibrated between software by quantifying the accuracy of each CTP software using simulated CTP data. With threshold calibration, mismatch profiles determined between three CTP software had 95% agreement. The proposed method may help systematically improve diagnostic agreement between CTP software.

Resource-limited hospitals may not have the capacity for routine CTP due to technical and logistical challenges. Non-contrast CT and multiphase CT angiography (mCTA) are widely available standard CT studies that together form a low temporal resolution CTP study. In Chapter 3, we demonstrated that mismatch profiles determined from mCTA perfusion maps had 82% concordance with that of standard CTP. Diagnostic agreement may improve with better optimized non-contrast CT and mCTA scan protocols. Our proposed technique may be a practical alternative to CTP when it is unavailable.

Sparse-view CT can reduce radiation dose by acquiring fewer x-ray projections but at the expense of streaking artifacts in the reconstructed image. In Chapter 4, we demonstrated that CTP is relatively insensitive to streaking artifacts and that radiation dose could be reduced by 50 to 66% compared to routine levels with sparse-view CTP. Low-dose CTP may improve the routine adoption of this useful diagnostic tool.

CTP plays an important role in the selecting patients with acute ischemic stroke for reperfusion treatment. The proposed improvements to CTP may allow wider adoption of this validated decision support tool.

Summary for Lay Audience

Stroke is mainly caused by a blood clot that limits blood flow to the brain. Permanent brain injury and paralysis can occur without early treatment to restore blood flow. Treatments are effective at reducing disability if a large volume of brain can be saved relative to that which is permanently injured, but may be harmful for patients who have little brain tissue to save. Treatment decisions must be carefully guided by evidence of brain injury. X-ray CT imaging of the head can non-invasively reveal brain tissue status. Patients normally receive a plain CT scan and a CT with an injection of x-ray contrast for visual assessment of tissue status. A special CT scan called CT perfusion may also be obtained, which uses x-ray contrast and rapid CT scanning to monitor the passing of contrast material through the brain. By analyzing the wash-in and wash-out of contrast using automated software, regional brain blood flow can be estimated. CT perfusion brain blood flow can distinguish permanent from reversible brain injury. Clinical trials have shown that CT perfusion diagnosis of brain injury can help reliably select patients for stroke treatment. However, because of technical challenges, CT perfusion is infrequently used, especially at rural hospitals where automated diagnostic tools may be most useful. Challenges include disagreement in diagnosis between CT perfusion software, expertise required for CT perfusion scanning, and concerns with CT radiation dose. My doctoral research aimed to address each of these challenges preventing the routine use of CT perfusion in stroke patients. In Chapter 2, quantitative differences between CT perfusion software were established then calibrated to improve diagnostic agreement. Chapter 3 showed that brain blood flow could be estimated from plain CT and CT with contrast scans instead of a specialized CT perfusion scan. Diagnosis of a treatable stroke was possible with these simple, widely available CT scans. Lastly, Chapter 4 showed that CT perfusion scan radiation dose could be reduced by 50 to 66% compared to routine levels. My doctoral research may help lift the barriers preventing CT perfusion from routine use, improving access to this important automated stroke diagnosis tool.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.