Tumor-non-tumor discrimination by a β^- detector for Radio Guided Surgery on ex-vivo neuroendocrine tumors samples

This paper provides a first insight of the potential of the β^- Radio Guided Surgery (β^--RGS) in a complex surgical environment like the abdomen, where multiple sources of background concur to the signal at the tumor site. This case is well reproduced by ex-vivo samples of 90^Y-marked Gastro-Entero-Pancreatic Neuroendocrine Tumors (GEP NET) in the bowel. These specimens indeed include at least three wide independent sources of background associated to three anatomical districts (mesentery, intestine, mucose). The study is based on the analysis of 37 lesions found on 5 samples belonging to 5 different patients. We show that the use of electrons, a short range particle, instead of γ particles, allows to limit counts read on a lesion to the sum of the tumor signal plus the background generated by the sole hosting district.The background on adjacent districts in the same specimen/patient is found to differ up to a factor 4, showing how the specificity and sensitivity of the β^--RGS technique can be fully exploited only upon a correct measurement of the contributing background. This locality has been used to set a site-specific cut-off algorithm to discriminate tumor and healthy tissue with a specificity of 100% and a sensitivity, on this test data sample, close to 100%. Factors influencing the sensitivity are also discussed. One of the specimens set allowed us evaluate the volume of the lesions, thus concluding that the probe was able to detect lesions as small as 0.04 mL in that particular case.