From nuclear track characterization to machine learning based image classification in neutron autoradiography for boron neutron capture therapy.

Knowledge of the 10B microdistribution is of great relevance in BNCT studies. Since 10B concentration assesment through neutron autoradiography depends on the correct quantification of tracks in a nuclear track detector, image acquisition and processing conditions should be controlled and verified, in order to obtain accurate results to be applied in the frame of BNCT. With this aim, an image verification process was proposed, based on parameters extracted from the quantified nuclear tracks. Track characterization was performed by selecting a set of morphological and pixel-intensity uniformity parameters from the quantified objects (area, diameter, roundness, aspect ratio, heterogeneity and clumpiness). Their distributions were studied, leading to the observation of varying behaviours in images generated by different samples and acquisition conditions. The distributions corresponding to samples coming from the BNC reaction showed similar attributes in each analyzed parameter, proving to be robust to the experimental process, but sensitive to light and focus conditions. Considering those observations, a manual feature extraction was performed as a pre-processing step. A Support Vector Machine (SVM) and a fully dense Neural Network (NN) were optimized, trained, and tested. The final performance metrics were similar for both models: 93%-93% for the SVM, vs 94%-95% for the NN in accuracy and precision respectively. Based on the distribution of the predicted class probabilities, the latter had a better capacity to reject inadequate images, so the NN was selected to perform the image verification step prior to quantification. The trained NN was able to correctly classify the images regardless of their track density. The exhaustive characterization of the nuclear tracks provided new knowledge related to the autoradiographic images generation. The inclusion of machine learning in the analysis workflow proves to optimize the boron determination process and paves the way for further applications in the field of boron imaging.

Charged particle spectrometry to measure 10B concentration in bone.

Osteosarcoma is the most common primary malignant tumour of bone in young patients. The survival of these patients has largely been improved due to adjuvant and neo-adjuvant chemotherapy in addition to surgery. Boron neutron capture therapy (BNCT) is proposed as a complementary therapy, due to its ability to inactivate tumour cells that may survive the standard treatment and that may be responsible for recurrences and/or metastases. BNCT is based on neutron irradiation of a tumour enriched in 10B with a boron-loaded drug. Low-energy neutron capture in 10B creates charged particles that impart a high dose to tumour cells, which can be calculated only knowing the boron concentration. Charged particle spectrometry is a method that can be used to quantify boron concentration. This method requires acquisition of the energy spectra of charged particles such as alpha particles produced by neutron capture reactions in thin tissue sections irradiated with low-energy neutrons. Boron concentration is then determined knowing the stopping power of the alpha particles in the sample material. This paper describes the adaptation of this method for bone, with emphasis on sample preparation, experimental set-up and stopping power assessment of the involved alpha particles. The knowledge of boron concentration in healthy bones is important, because it allows for any dose limitation that might be necessary to avoid adverse effects such as bone fragility. The measurement process was studied through Monte Carlo simulations and analytical calculations. Finally, the boron content of bone samples was measured by alpha spectrometry at the TRIGA reactor in Pavia, Italy, and compared to that obtained by neutron autoradiography. The agreement between the results obtained with these techniques confirms the suitability of alpha spectrometry to measure boron in bone.

Extending neutron autoradiography technique for boron concentration measurements in hard tissues.

The neutron autoradiography technique using polycarbonate nuclear track detectors (NTD) has been extended to quantify the boron concentration in hard tissues, an application of special interest in Boron Neutron Capture Therapy (BNCT). Chemical and mechanical processing methods to prepare thin tissue sections as required by this technique have been explored. Four different decalcification methods governed by slow and fast kinetics were tested in boron-loaded bones. Due to the significant loss of the boron content, this technique was discarded. On the contrary, mechanical manipulation to obtain bone powder and tissue sections of tens of microns thick proved reproducible and suitable, ensuring a proper conservation of the boron content in the samples. A calibration curve that relates the 10B concentration of a bone sample and the track density in a Lexan NTD is presented. Bone powder embedded in boric acid solution with known boron concentrations between 0 and 100 ppm was used as a standard material. The samples, contained in slim Lexan cases, were exposed to a neutron fluence of 1012 cm-2 at the thermal column central facility of the RA-3 reactor (Argentina). The revealed tracks in the NTD were counted with an image processing software. The effect of track overlapping was studied and corresponding corrections were implemented in the presented calibration curve. Stochastic simulations of the track densities produced by the products of the 10B thermal neutron capture reaction for different boron concentrations in bone were performed and compared with the experimental results. The remarkable agreement between the two curves suggested the suitability of the obtained experimental calibration curve. This neutron autoradiography technique was finally applied to determine the boron concentration in pulverized and compact bone samples coming from a sheep experimental model. The obtained results for both type of samples agreed with boron measurements carried out by ICP-OES within experimental uncertainties. The fact that the histological structure of bone sections remains preserved allows for future boron microdistribution analysis.