Biodose Tools: Dose Assessment Technologies

• Biodose Tools are advanced systems that integrate cytogenetic software, Monte Carlo simulation, real-time imaging, and embedded sensors to quantify ionizing radiation doses.
• They optimize dose estimation and reporting in clinical therapies, emergency triage, and environmental monitoring through standardized, automated processes.
• Recent innovations include deep learning for rapid dose prediction and GPU-accelerated imaging, which enhance treatment planning and quality assurance.

Biodose tools comprise a diverse suite of hardware systems and software platforms designed to measure, estimate, and interpret biologically relevant ionizing radiation doses in biomedical, clinical, and environmental contexts. These technologies span cytogenetic software for mass-casualty triage, Monte Carlo simulation pipelines for internal and external dosimetry, real-time imaging and monitoring devices for advanced treatment modalities (e.g., BNCT, HDR brachytherapy), and compact systems for environmental radioactivity assessment. Their unifying purpose is to quantify absorbed doses, optimize radiation delivery, validate biological effects, and harmonize dose reporting across different laboratories, protocols, and exposure scenarios.

1. Taxonomy and Principle Classes of Biodose Tools

Biodose tools encompass a range of instrumentation and algorithms, which can be categorized as follows:

• Cytogenetic/Calibration Software: Analytical platforms for estimating dose from biological markers such as dicentric chromosome and translocation assays (e.g., Biodose Tools (Frances-Abellan et al., 10 Nov 2025)).
• Monte Carlo Dose Engines: General-purpose engines utilizing patient imaging data (CT, DICOM, PET/SPECT) to simulate patient- or sample-specific, voxel-level dose (e.g., GEANT4+gMocren (Lee et al., 2015), TOPAS-MC (Rodrigues et al., 2023)).
• Treatment Planning and Optimization Systems: Integration of image-based data, source/applicator libraries, and optimization routines for personalized therapy planning (e.g., RapidBrachyMCTPS 2.0 (Glickman et al., 2020)).
• Real-time/Embedded Systems: In-situ measurement of dose delivery via detector arrays, miniature implantable sensors, or multi-sensor tracking (e.g., i-TED for BNCT (Lerendegui-Marco et al., 9 Sep 2024), DOSION (Boissonnat et al., 2016), IViST (Rosales et al., 2020), injectable TLD (Garty et al., 2019)).
• Environmental and Low-dose Assessment: Compact analytical instruments using spectrometric techniques to measure radioactive element activities and compute environmental dose rates (e.g., μDose (Tudyka et al., 2018)).
• Data-driven and ML-based Systems: Emerging deep-learning pipelines for rapid patient-specific, image-domain dose estimation (e.g., DeepBEVdose (Fan et al., 2023)).

Each of these tool classes addresses different requirements for spatial and temporal resolution, sample throughput, error quantification, and integration with workflow or reporting standards.

2. Cytogenetic Software: Biodose Tools Platform

The Biodose Tools platform (v3.7.1) (Frances-Abellan et al., 10 Nov 2025) implements a modular, Shiny-based R application for cytogenetic dosimetry, focused on large-scale emergency response and laboratory harmonization. Its key modules:

• Characteristic Limits: Automated computation of decision thresholds (L_c) and detection limits (L_D) under ISO 19238:2023 using Poisson statistics. Given a calibration curve $Y(D) = C + \alpha D + \beta D^2$, thresholds are mapped to minimum resolvable dose/inverse quadratic solutions.
• Enhanced Dose-Estimation: Supports batch-mode dose estimation for dicentric/translocation assays in acute/protracted/partial exposure. Dose inversion employs

$$D_{\text{est}} = \frac{-\alpha + \sqrt{\alpha^2 + 4\beta(Y_{\text{obs}} - C)}}{2\beta}$$

with profile-likelihood CIs.

• Criticality Accidents: Mixed neutron-gamma field estimation utilizes the additive model

$$A_{\text{tot}}(D_n, D_g) = C + a D_n + b D_g + c D_g^2$$

with closed-form quadratic solutions when $p = D_n/D_g$ is known.

• Interlaboratory Comparison (ILC): Upload of .rds result files, robust Z-score calculation against reference means (Huber M-estimator, ISO 13528 algorithms), and report/visualization export.
• Reporting: Full audit trails via LaTeX, PDF, DOCX, or XLSX; harmonization of statistical practices across disparate laboratories.

This advances triage and harmonization in mass-casualty situations by providing standardized, auditable, and robust dose estimates.

3. Monte Carlo Image-driven Dose Calculation Engines

Monte Carlo tools underpin many biodose platforms due to their high physical fidelity and generalizability:

• GEANT4+gMocren (Lee et al., 2015): Implements patient-specific absorbed dose using DICOM imaging, custom handlers for CT-HU → material mapping, and G4VParameterised voxel geometry. Outputs are visualized in gMocren, which supports 3D/2D dose overlays, volume rendering, and qualitative isodose analysis. Validation versus TLD/film yields agreement within 3%–5%.
• TOPAS-MC Extension (Rodrigues et al., 2023): Introduces a voxelized radionuclide source for SPECT/PET-driven internal dosimetry. The C++ "VoxelParticleSource" class employs inverse-transform sampling on the normalized intensity vector of the activity map. The system reconstructs per-voxel dose as

$$D_v = \frac{1}{m_v} \sum_{k\in\text{int. in }v} E^{\text{dep}}_k$$

achieving ~2–3% precision with $N \geq 10^8$ histories. The system bridges clinical imaging and detailed MC calculation, though it remains computationally intensive.

Such toolchains are validated against physical dosimeters (TLDs, radiochromic film) and other MC engines, and are now being extended to hybrid, functional imaging pipelines (e.g., PET/CT-driven S values).

4. Real-Time and Embedded Biodose Measurement Devices

Several platforms provide rapid, in situ dose measurement with high temporal and spatial resolution:

• i-TED Compton Camera in BNCT (Lerendegui-Marco et al., 9 Sep 2024): Utilizes a double-plane LaCl₃(Ce) array with SiPM readout to map prompt 478 keV γ-rays from $^{10}$B(n,α)$^7$Li reactions in Boron Neutron Capture Therapy. Sensitivity reaches sub-μg $^{10}$B with cm-scale dose resolution. GPU-accelerated analytical reconstruction algorithms enable near real-time 2D mapping, paving the way for electronic online biodose monitoring.
• DOSION Air-Ionization Chamber (Boissonnat et al., 2016): A 64-channel strip-segmented ionization chamber providing per-sample, 2D fluence and dose maps in heavy-ion radiobiology settings. Calibrated via LET measurement and cross-referenced with plastic scintillator/PMT, it delivers ~4% RMS homogeneity over 24-sample irradiations.
• IViST Multi-sensor HDR Brachytherapy Platform (Rosales et al., 2020): Implements a 3-point plastic scintillator array with dichroic/PMT spectral separation, achieving ≤1 mm source localization and dose precision within 5% (up to 6 cm from $^{192}$Ir source). Real-time QA flags inconsistencies in dwell time and source placement during therapy.
• Injectable TLD Rods (Garty et al., 2019): Glass-encapsulated LiF:Mg,Ti rods allow in-vivo, passive dose recording in small animals, with reliable reconstruction within ±10% of nominal dose, enabling individualized validation in mobile, long-term irradiation studies.

These devices close the feedback loop between planned, delivered, and biologically effective dose on timescales relevant for both clinical corrections and experimental quality control.

5. Image-Domain and Data-driven Dose Engines

Recent developments harness deep learning for rapid, scalable dose prediction directly from imaging data:

• DeepBEVdose (Fan et al., 2023): Replaces explicit ray tracing and kernel convolution with a "beam’s-eye-view" CNN (UNet-Res architecture), taking as input (BEV-sliced) CT, fluence, and SSD maps. This achieves voxel-wise dose differences ≤2–3% versus a commercial TPS (Pinnacle CCC), with isodose Dice-coefficients 0.85–1 and computational time per beam reduced to ~3–5 s (NVIDIA 2080Ti). Training is currently limited to single-energy, static IMRT, but the framework supports extension to more complex beam geometries.

Such methods offer state-of-the-art speed for scenarios (e.g., online adaptive) where batch MC or TPS engines are too slow, provided rigorous clinical QA is maintained.

6. Environmental and Low-dose Biodose Instrumentation

Biodose assessment for environmental and low-level exposures leverages spectrometric/decay-sequence analysis:

• μDose Compact System (Tudyka et al., 2018): Integrates dual α/β scintillation, full waveform/time stamping, and a custom software suite to derive activities of U/Th decay series and $^{40}$K by multivariate solution of decay-pair and single-rate equations. Dose-rate conversion employs the infinite matrix assumption with state-of-the-art conversion tables, and quantifies U/Th/K activities to ∼ 2.5% (1 g, 44 h). Results agree within 2–3σ of HPGe gamma spectrometry.

Such systems facilitate precise, operator-agnostic assessment of environmental and archaeological samples with minimal infrastructure needs.

7. Integration, Limitations, and Future Directions

Biodose tools increasingly support automation, scalability, and interoperability:

• Data Integration: Standardized output (CSV/XLSX/PDF, DICOM RT-Dose) and APIs for clinical/trial datasets are widespread.
• Limitations: High-fidelity MC approaches remain computationally expensive for large cohorts or 4D planning; simplified phantoms or incomplete cross-validation may introduce systematic bias.
• Trends: Incorporation of GPU-accelerated reconstruction, ML-guided error quantification, batch QA tools, digital workflow automation, and harmonization with international reporting standards (ISO, RENEB, IAEA).

Significant ongoing work includes extension to tomographic (3D) real-time reconstruction (i.e., i-TED array (Lerendegui-Marco et al., 9 Sep 2024)), patient-specific internal dosimetry (TOPAS-MC (Rodrigues et al., 2023)), and direct hybridization of cytogenetic, physical, and computational biodose evidence streams under large-scale emergency conditions.

In summary, contemporary biodose tools constitute an ecosystem of complementary hardware and computational platforms, targeting reproducible, validated, and standardized dose assessment across all modalities and exposure contexts. This ecosystem underpins both advanced therapy individualization and emergency response triage in modern radiological practice.