Machine learning methodology for high throughput personalized neutron dose reconstruction in mixed neutron + photon exposures

Abstract

We implemented machine learning in the radiation biodosimetry field to quantitatively reconstruct neutron doses in mixed neutron + photon exposures, which are expected in improvised nuclear device detonations. Such individualized reconstructions are crucial for triage and treatment because neutrons are more biologically damaging than photons. We used a high-throughput micronucleus assay with automated scanning/imaging on lymphocytes from human blood ex-vivo irradiated with 44 different combinations of 0–4 Gy neutrons and 0–15 Gy photons (542 blood samples), which include reanalysis of past experiments. We developed several metrics that describe micronuclei/cell probability distributions in binucleated cells, and used them as predictors in random forest (RF) and XGboost machine learning analyses to reconstruct the neutron dose in each sample. The probability of “overfitting” was minimized by training both algorithms with repeated cross-validation on a randomly-selected subset of the data, and measuring performance on the rest. RF achieved the best performance. Mean R2 for actual vs. reconstructed neutron doses over 300 random training/testing splits was 0.869 (range 0.761 to 0.919) and root mean squared error was 0.239 (0.195 to 0.351) Gy. These results demonstrate the promising potential of machine learning to reconstruct the neutron dose component in clinically-relevant complex radiation exposure scenarios.

A MATLAB-based program for three-dimensional quantitative analysis of micronuclei reveals that neuroinflammation induces micronuclei formation in the brain

“…Since the micronucleus is defined by several parameters, such as nuclear components and size, the precise quantification of micronuclei is challenging. Several methods to quantify micronuclei have been reported to date 9 , 23 – 26 . The most frequently used method involves counting the small nuclei by analyzing the nuclear size as the index.…”

“…CAMDi can elicit each nucleus using ellipse fitting functions and analyze the suitable size of nuclei from Z-stack images, showing an accurate number of both nuclei and micronuclei. Previously, several groups have reported a high-throughput screening system for quantifying micronuclei in culture cells 9 , 23 – 26 . Although the precise quantification methods in vivo have not been developed, investigating micronuclei is crucial to understanding the pathological condition in each tissue.…”

“…Several studies have reported approaches to quantify micronuclei in vitro and in vivo. The most common approach involves detecting the small nuclei by nuclear staining and imaging manually or automatically 9 , 23 – 26 . These standard approaches are employed to count the number of small nuclei by setting the threshold, such as the nuclear size.…”

“…This approach is used to obtain data effectively and for general use. Quantification in conjunction with high-throughput screening and machine learning is powerful for analyzing the genomic condition in vitro 25 , 26 . It is technically difficult, however, to obtain precise data on the micronucleus in tissues due to stereoscopic challenges (see the Results sections).…”

Cytogenetic Damage of Human Lymphocytes in Humanized Mice Exposed to Neutrons and X Rays 24 h After Exposure

“…Novel insights observed here using the humanized mouse model could be used in the development of future studies to investigate mixed exposures with different percentages of neutrons and photons at longer times after exposure. More recently, we have used a highthroughput automated CBMN assay in human lymphocytes and developed a machine learning approach to reconstruct neutron and photon doses in complex exposure scenarios [Shuryak et al, 2020[Shuryak et al, , 2021.…”