New autoradiography image analysis tool combining
overlain microscopy and histology in Carimas
multimodal DICOM image analysis software
M. Miner1,2, S. Piirola3,2, E. D. Atencio Herre1, H. Liljenbäck1, R. K.
Aarnio1,2, M. Ståhle1, A. Roivainen1,2,4;

Aim/Introduction: Autoradiography is an essential tool for the
many different stages of radiopharmaceutical development. It is
often used in the assessment of thin layer chromatography (TLC) of
synthesis products, metabolite products and can directly measure
ex vivo tissues (whole and sections) with high spatial resolution.
A new autoradiography analysis module was built into Carimas
medical image analysis software to improve workflow and move
closer to an all-in-one radiopharmaceutical image suite for in
vitro, in vivo, and ex vivo studies. This new program was rigorously
tested against existing software to examine its comparability
and validity with existing workflows. Materials and Methods:
Comparative testing of Carimas (Turku PET Centre) versus
commercially available software (AIDA and TINA; Raytest) was
done via practical methods and pixel-by-pixel testing in random
clusters. First, fluorine-18-containing samples were exposed to
phosphor imaging plates for 4 hours and scanned with a Fujifilm
BAS-5000 using different encoding and resolution settings (8- and
16-bit depth as well as 25 μm and 50 μm resolution). Segmentation
was performed on macro regions, pixel-by-pixel in random 6-by-
3 clusters, and small 3×3 pixel areas extracting photo-stimulated
luminescence (PSL) values. Results: All small region analyses were
identical (after adjusting for a 1 pixel shift present in TINA) in PSL
and PSL/mm2 output suggesting that the same quantification
equation is employed in all three software suites. Macro regional
analyses varied only by an average (n = 72 regions from 4 images
and all analysed in each 3 software programs) of 0.084% ± 2.163,
which was attributed to subtle differences in manual region
drawing. A single factor ANOVA test suggested no significant
differences between the three programs (P = 0.999, F = 0.001).
Conclusion: The results showed no significant difference in data
output between programs. Carimas was demonstrated to be an
accurate program for analysing autoradiography image data and
includes the ability to overlay histology images to conveniently
guide segmentation.