CMOS Camera Calibration for Quantification of Cerenkov Emission during Radiotracer Synthesis
Abstract
Purpose
To develop robust CMOS camera calibration methodology for non-destructive quantification of radionuclides during high-throughput radiotracer synthesis. Estimation of radioactivity is achieved through the detection of Cerenkov light emission from positrons. The available signal is weak enough that minor fluctuations in camera bias and, in a high radiation environment, direct gamma strikes on the CMOS chip affect quantification.
Methods
Conventional methodologies for master bias and dark frame subtraction were compared with the newly developed adaptive image correction algorithm. This method, developed in MATLAB, removes weak but patterned bias fluctuations and an adaptive median filter removes direct CMOS sensor strikes. 100 dark images (0C, 60 second exposures) were taken using a cooled monochrome camera (ASI-533MM Pro) and corrected with each method. Resulting images were binned into large ROIs, and the average value of each ROI, representing residual noise per pixel, was calculated in electrons. To compare correction methods, the Root Mean Squared Error (RMSE) and a Welch’s t-test of the squared errors were performed. A weak standard light source (3H in scintillator) was independently imaged for 60 second exposures and corrected with the adaptive correction algorithm to ensure consistent quantification.
Results
The developed corrections (RMSE = 0.017 e-/pixel) significantly reduced the RMSE of the residual noise, (p < 0.001), as compared to the conventional correction method (RMSE = 0.055 e-/pixel). After adaptive correction, the mean residual noise was 0.0013 e-/pixel, resolving very weak Cerenkov signals at or below 0.1 e-/pixel. Despite image-to-image and spatially varying noise, the 3H light source had less than 1% variation across images.
Conclusion
This work demonstrates that the developed methodology effectively removes patterned fluctuations and direct strikes, lowering detection limits as compared to conventional corrections. The new low detection limit translates into higher precision in measuring radiotracer yield after synthesis with a relatively low-cost implementation.