M. E.Fay, R.Tandon, T.Latham, et al., “Artificial Intelligence-Based Analysis of Central Nervous System Vasculopathy in Pediatric Sickle Cell Anemia,” American Journal of Hematology (2026): 1–7, https://doi.org/10.1002/ajh.70251.
Explanation
This study aims to use imaging data from brain scans and artificial intelligence to detect differences in the blood vessels in the brain of children with sickle cell anemia (SCA) to determine their risk of stroke (0 – no changes to blood vessels to 6 – severe narrowing/blocking of multiple blood vessels).
Key Points
- Patients in this study consisted of 42 children ages 2 to 17 with SCA who were part of a larger study known as EXpanding Treatment for Existing Neurological Diseases) EXTEND, which tested hydroxyurea as a treatment to prevent stroke in Jamaica
- Artificial intelligence was used to compile visualized data on patients to create a system that can detect characteristics of blood vessels including vessel length, the winding of a vessel, area, width, and location within the brain
- This system was used to categorize children into three risk categories ranging from moderately at risk, highly at risk, and those who have experienced a stroke
- Findings of the study
- Surprisingly, the number of blood vessels detected in the brain decreased over time in children who were moderately at risk and have experienced a stroke (no changes were detected in the highly at risk category)
- The way blood vessels were connected were distinctly different between each category at the start of the study
- Children who were highly at risk were found have more winding blood vessels
- Artificial intelligence was able to correctly identify the children who had previously experienced a stroke (they were found to have more complex blood vessel patterns)
- Vessel winding can be an early sign of SCA severity
- Artificial intelligence can be used to measure stroke risk
- This study provides a connection between blood vessels in the brain and SCA and the use of artificial intelligence in early stroke detection in SCA patients. This method is also effective in that the data can be collected from previous brain scans, not requiring additional procedures.
Limitations
This study only used prior data and does not follow up with the patient’s current conditions. Also, data on a healthy group of children were not included, observations cannot be attributed to solely SCA patients.
Written by: James La
