Source-linked AI summary
OPTARI: An open-source software framework for clinical translation of optoacoustic imaging
Moritz Schillinger, Maximilian Bader, Adrian Buehler, Felix Wachter, Ferdinand Knieling, Vasilis Ntziachristos, Katharina Breininger
TL;DR
Clinical OPUS studies lack accessible software for reproducible analysis, despite advances in OptA processing. OPTARI combines established tools in an open-source graphical workflow; its automated analysis is comparable to manual annotation for muscle tissue, and batch processing takes under a second per scan.
Problem
Clinical researchers lack straightforward access to recent OptA processing, while subjective analysis choices and complex parameters hinder reproducible workflows.
Source: §1 · p. 2Method
OPTARI extends established open-source tools with a graphical clinical OPUS workflow and batch processing of reconstruction, unmixing, segmentation, and ROI placement.
Source: §1 · p. 2, §2.2.6 · p. 6Results
0.93 ± 0.03 s per scan for the default batch pipeline, while automated segmentation was reported as a viable alternative to manual ROI annotation without clear loss in group discrimination.
Source: §4.3 · p. 11Takeaways & Limitations
OPTARI combines a graphical interface with reconstruction, unmixing, and multi-patient workflow capabilities for large-scale clinical OPUS studies.
Source: §5 · p. 11OPTARI targets clinical two-dimensional handheld OPUS data, lacks volumetric and preclinical support, and automated ROI placement may need finetuning for new sites or scanners.
Source: §5 · p. 12
Abstract
from arXiv · showhide
Combined optoacoustic (OptA) and ultrasound imaging is a promising modality across a wide range of pathologies. Its use in clinical research, however, is limited by a lack of analysis software accessible to users without programming expertise. Existing tools are either script-based engineering frameworks or proprietary vendor applications. We present OPTARI, an open-source graphical software application to facilitate large-scale clinical optoacoustic-ultrasound (OPUS) studies. OPTARI builds on the napari image viewer and the PATATO toolkit for OptA data processing. It supports native vendor and open HDF5 data formats, aligns with the IPASC standard, and is distributed as a standalone executable. Features include region-of-interest annotation, reconstruction, chromophore unmixing, deep-learning-based tissue segmentation, and temporal and spectral analysis modules. OPTARI offers a combination of features for clinical study analysis that none of the compared tools provide. The automated tissue segmentation produced results comparable to expert labeling for muscle tissue, and batch processing reduced analysis time from minutes to under one second per scan. The tool is used in ongoing clinical OPUS studies. OPTARI connects engineering frameworks with medical research and provides an open-source basis for reproducible and standardized analysis in clinical OptA imaging studies.
1 Introduction
Clinical OPUS research lacks accessible software that supports standardized, reproducible analysis across large studies. OPTARI addresses this gap with an open-source application combining established tools and a clinical study workflow.
1 Introduction
- Clinical studies rarely use recent OptA processing algorithms because researchers outside engineering lack a straightforward way to access them.Quantitative analysis requires dedicated reconstruction, spectral unmixing, and analysis software.Source: §1 · p. 2
- Proprietary formats impede dataset migration, inter-center sharing, and custom analysis, while IPASC open formats have not yet been widely adopted in clinical studies.Source: §1 · p. 2
- Subjective choices in reconstruction, spectra, frames, and ROIs introduce operator variability, motivating more objective workflows for reproducible and comparable OPUS studies.Source: §1 · p. 2
- Complex parameters and large datasets make manual configuration impractical, creating a need for intuitive tools with shareable defaults that focus on relevant data and metrics.Source: §1 · p. 2
- Existing vendor tools are hardware- or format-specific, while open-source options lack the graphical interface and structured multi-patient workflow needed for large clinical studies.Source: §1 · p. 2
- OPTARI is an open-source, platform-agnostic application that extends established tools and combines them with a clinical OPUS workflow and shareable configurations.Automated ROI placement and batch processing reduce analysis time for large datasets; OPTARI lets researchers adjust and share configurations rather than defining standards or parameters itself.Source: §1 · pp. 2–3
2 Scope and Features
OPTARI focuses on clinical two-dimensional handheld multispectral optoacoustic-ultrasound imaging.
2 Scope and Features
- OPTARI targets clinical two-dimensional handheld multispectral OPUS imaging, a setup forming the basis for most clinical studies across manufacturers.Reported application areas include inflammatory bowel disease, breast cancer, muscular disorders, and cardiovascular conditions.Source: §2 · p. 3
2.1 System Requirements
OPTARI’s six system requirements address data standardization, operator variability, and technical barriers.
2.1 System Requirements
- The six requirements map data standardization to R6, operator variability to R4, and technical barriers to R1, R2, R3, and R5.Source: §2.1 · p. 3
2.2 Feature Overview
The feature overview presents a GUI for navigating, annotating, processing, and analyzing OPUS scans, with exports and batch workflows for study-scale use.
2.2.1 Viewing and Navigation
- The Scan Browser navigates multi-scan studies, while sliders move through wavelengths, frames, and unmixed components and the viewer overlays ultrasound, optoacoustic, and annotation layers.A motion-scoring algorithm selects the least-motion frame, and metadata are shown with IPASC-compliant tag names and units.Source: §2.2.1 · p. 3
2.2.2 Annotation, ROI Management, and Automated Segmentation
- Custom or predefined ROIs yield configurable measurements, with Live Analysis updating as regions change and Saved Analysis preserving selected measurements.Pretrained tissue segmentation supports automatic ROI placement; ROI presets can be shared and anchored to segmentation masks across frames.Source: §2.2.2 · p. 4, Figure 1 · p. 5
2.2.3 Temporal and Spectral Analysis
- Temporal Analysis plots ROI-group intensities against scan timestamps, while Spectral Analysis displays signal intensity by wavelength.Source: §2.2.3 · p. 5
2.2.4 Data Processing and Conversion
- OPTARI loads HDF5, MSOT Acuity, and MSOT Frontier data and provides GUI access to reconstruction and linear spectral unmixing.Processing includes backprojection and a DeepMB adapter; reference spectra support chromophore concentrations and derived features such as oxygen saturation and total hemoglobin, and parameters can be saved or shared.Source: §2.2.4 · pp. 5–6
2.2.5 Data Export
- Processed scans, layers, ROIs, and segmentation masks can be exported as HDF5 with IPASC-compliant metadata, while measurements and plots support spreadsheet or image export.Saved measurements can be re-imported to resume analysis or combine results across sessions or machines.Source: §2.2.5 · p. 6
2.2.6 Batch Mode
- Batch Mode runs a saved reconstruction, unmixing, segmentation, and ROI-placement pipeline across a dataset without manual interaction.Presets are tuned on a representative scan, and the analysis plan is checked for incompatibilities or missing requirements before execution; failures are logged and skipped.Source: §2.2.6 · p. 6
3 Implementation and Architecture
OPTARI separates its interface from processing through a layered, modular architecture, integrating external tools for data handling and analysis. It is packaged as a standalone application with configurable settings and presets.
3.1 Software Architecture
- OPTARI divides the application into UI, Control, Utilities, and External layers, with the OptariController coordinating global state and specialized task controllers handling specific operations.The UI widgets contain no internal logic; a stateless UIManager connects interface signals to backend routines.Source: §3.1 · p. 6
- Pretrained models segment three common examination-site muscles, enabling automatic ROI placement; ONNX adapters allow additional models, and background tasks keep the interface responsive during processing.The included target muscles are gastrocnemius, tibialis anterior, and vastus lateralis.Source: §3.1 · p. 7, Figure 2 · p. 7
3.2 External Integrations
- OPTARI connects to PATATO for proprietary vendor and HDF5 data, uses a PATATO-inspired HDF5 schema with IPASC metadata, and imports or exports native IPASC files through PACFISH.The chosen schema also accommodates ultrasound and processed or reconstructed layers, unlike the raw-time-series-scoped native IPASC format.Source: §3.2 · p. 8
- A motion score based on structural similarity and zero-mean normalized cross-correlation automatically suggests the least-motion ultrasound frame for analysis.The scoring approach is adapted from Kukačka et al.22.Source: §3.2 · p. 9
3.3 Deployment and Installation
- OPTARI is distributed as a standalone executable with precompiled binaries for Windows, macOS, and Linux, avoiding local compiler requirements and administrative rights.Settings and presets are stored as human-readable JSON files in the user’s home directory.Source: §3.3 · p. 9
4 Evaluation
The evaluation compares OPTARI with existing graphical OptA tools, describes its clinical-study use, and assesses automated muscle segmentation and processing speed. Automated ROI placement was comparable to expert annotation in PAD classification, while the batch pipeline processed a scan in under a second.
4.1 Feature Comparison
- Among the compared graphical tools, OPTARI combines clinical annotation and multi-patient workflow features with reconstruction and unmixing, which the other tools lack in combination.PATATO offers reconstruction and unmixing but is largely script-based and lacks a structured multi-patient workflow; iLabs is closed-source and hardware- and format-locked.Source: §4.1 · p. 9, Table 2 · p. 10
4 Evaluation
- The comparison covers graphical-access tools for OptA processing, while simulation-focused and preclinical tools such as SIMPA, PHANTOM, and MSOT-RAFT are outside its scope.SIMPA lacks GUI, annotation, and clinical workflow features; PHANTOM targets preclinical fluence compensation, and MSOT-RAFT is script-based.Source: §4.1 · p. 10
4.2 Integration into Clinical Studies
- OPTARI was developed with clinical professionals, its workflows were iteratively revised from their feedback, and it is being used in ongoing clinical studies.Documentation includes step-by-step guides and training videos.Source: §4.2 · p. 10
4.3 Performance of the Automated Analysis
- Automated muscle segmentation produced PAD-versus-healthy classification comparable to expert annotation: AUROC was 0.76 for automatic ROIs and 0.77 for manual ROIs.The underlying UNet segmentation achieved a Dice score of 0.86 and a 95th-percentile Hausdorff distance of 1.92 mm across 5-fold cross-validation.Source: §4.3 · p. 11
- 0.93 ± 0.03 s per scan for the default batch pipeline, with segmentation and automatic ROI placement taking 0.31 ± 0.02 s; 1,977 scans from 102 subjects took roughly 31 minutes, excluding manual quality checks.The representative scan had 13 wavelengths reconstructed into 400×400 px images; manual delineation of one complex ROI could take up to one minute.Source: §4.3 · p. 11
5 Discussion
OPTARI combines open-source infrastructure with clinical OPUS workflows, offering a feature combination the authors say no other open-source tool provides. Its scope and automated ROI use have practical limits, and it is research software rather than a medical device.
5 Discussion
- OPTARI uniquely combines a graphical interface, reconstruction, unmixing, and multi-patient workflow capabilities for large-scale clinical OPUS studies among open-source tools, building on napari and PATATO.It adds a clinically oriented workflow and automated parameter selection to existing open-source infrastructure.Source: §5 · p. 11
- 0.93 ± 0.03 s per scan for the default batch pipeline on one representative scan, measured over 100 repetitions on a standard laptop.The configuration used HDF5 loading, BP reconstruction, two-chromophore LU, single-class segmentation, and single-frame ROI extraction.Source: Figure 4 · p. 12
- OPTARI supports IPASC priorities through automated ROI delineation, an accessible GUI and documentation, and reading and writing IPASC files with its metadata schema.These features address the roadmap’s priorities for segmentation, clinical adoption, education, and data management.Source: §5 · p. 12
- OPTARI targets clinical two-dimensional handheld OPUS data, not volumetric scans or preclinical systems; automated ROI placement may need finetuning for new anatomical sites or scanner types and should be reviewed.Reviewing automatically placed ROIs reduces but does not eliminate the time saved.Source: §5 · p. 12
- OPTARI is research software that has not undergone medical-device conformity assessment and is not intended for diagnosis or treatment decisions.The tool remains under active development, and the authors welcome community contributions.Source: §5 · p. 12
Code, Data, and Materials Availability
The source code and executables are available under the BSD-3-Clause license, with documentation provided online. The paper identifies the versions used and points to the clinical dataset and segmentation-model resources.
Code, Data, and Materials Availability
- OPTARI source code and executables are available under the BSD-3-Clause license, with documentation at the project website.The analyses reported in the paper used version 0.8.0.Source: Code, Data, and Materials Availability · p. 13
- The clinical PAD dataset used in Section 4.3 is available on Zenodo, and pretrained segmentation models download automatically on first use.DeepMB model weights are available from the original authors upon reasonable request.Source: Code, Data, and Materials Availability · p. 13