AI Volumetric Response Criteria
AI Volumetric Response Criteria in Mesothelioma: ARTIMES, 10,926 CT Scans, and Automated Tumor Measurement (2026)
Executive Summary
AI volumetric response criteria use artificial intelligence to automatically measure the total tumor volume of pleural mesothelioma on CT scans and track how that volume changes during treatment. Mesothelioma grows as a thin rind wrapping the lung rather than as round masses, so the diameter-based rulers used for most cancers — RECIST and its mesothelioma-specific version, modified RECIST — fit it poorly. In 2026 the ARTIMES study, published in Lancet Oncology, validated an AI-assisted volumetric method across 10,926 CT scans from 2,080 patients in 14 cohorts. It predicted survival more accurately than modified RECIST (concordance index 0.83 versus 0.73) and flagged disease progression roughly five weeks earlier. AI volumetry is a promising research tool that is entering formal staging discussions, but it remains investigational — it is not FDA-cleared and its authors call for prospective validation before routine clinical use.
Why This Matters
For a mesothelioma patient, the question behind response assessment is simple: "is my treatment working?" The answer depends on how accurately doctors can measure the tumor over time. Because mesothelioma's shape defeats ordinary measurement rulers, better tools mean earlier, more reliable answers about whether to continue, change, or stop a therapy. AI volumetry is one of those emerging tools. Understanding what it can and cannot yet do helps patients and families ask informed questions when reviewing scans with their oncology team.
At a Glance
AI-assisted volumetric response assessment in mesothelioma, 2026:
- What it is — software that automatically outlines (segments) the full mesothelioma tumor on a CT scan and calculates its volume, then tracks volume change as a measure of treatment response.
- The problem it addresses — mesothelioma's unique growth pattern makes standard diameter-based RECIST difficult to apply; modified RECIST fits it better, but reliable, reproducible measurement still depends on scrupulous, standardized methodology.[1]
- The anchor evidence (ARTIMES) — validated across 10,926 CT scans from 2,080 patients in 14 cohorts, including data from major treatment trials.[2]
- Prognostic accuracy — ARTIMES volumetric response predicted overall survival with a concordance index of 0.83, compared with 0.73 for modified RECIST.[2]
- Earlier detection — the volumetric method identified disease progression approximately five weeks earlier than modified RECIST.[2]
- Earlier proof of concept — a fully automated deep-learning system demonstrated in 2022, via blinded multicentre validation on 183 CT datasets, that mesothelioma volume can be measured automatically, with moderate agreement against modified RECIST for response classification.[3]
- Still investigational — the ARTIMES authors call for prospective validation before the criteria are ready for routine clinical use, a question a companion editorial in the same issue poses directly in its title, "ready for deployment?"[2][4]
- Entering staging — the 9th edition of the TNM staging system now explicitly incorporates automated tumor-volume assessment into imaging-based staging.[5]
Key Facts
| Essential AI Volumetric Response Information | |
|---|---|
| What Is Measured | Total tumor volume on CT, segmented automatically by artificial intelligence |
| Current Standard It Improves On | Modified RECIST (mRECIST) — tumor thickness measured perpendicular to the chest wall |
| Anchor Study | ARTIMES — retrospective, multicohort, multicentre validation, Lancet Oncology 2026 |
| Dataset Size | 10,926 CT scans from 2,080 patients across 14 cohorts |
| Prognostic Performance | Concordance index 0.83 (volumetric) vs 0.73 (modified RECIST) for overall survival |
| Timing Advantage | Progression detected approximately 5 weeks earlier than modified RECIST |
| Regulatory Status | Investigational — not FDA-cleared; prospective validation pending |
| Complementary Imaging | FDG-PET/CT metabolic tumor volume; TNM 9th-edition volumetric staging |
Why Do Standard Response Criteria Struggle With Mesothelioma?
Most solid cancers are measured with RECIST — Response Evaluation Criteria in Solid Tumors — which tracks the longest diameter of discrete tumors. Pleural mesothelioma does not cooperate. It grows as a thin, irregular rind that wraps around the lung and lines the chest cavity, rather than forming round, easily measured masses. To adapt, radiologists use modified RECIST (mRECIST), which measures tumor thickness perpendicular to the chest wall at several defined levels and sums those measurements.[1] Modified RECIST is the accepted standard, but it remains a simplification of a three-dimensional problem: it samples thickness at a few points, and different readers can measure the same scan differently. Immunotherapy has added a further complication: checkpoint inhibitors can produce unconventional response patterns such as pseudoprogression and hyperprogression that are difficult to classify with thickness-based rules.[6] These limitations are the reason researchers have pursued full three-dimensional volumetric measurement.
What Is Volumetric Response Assessment?
Volumetric response assessment measures the entire three-dimensional volume of the tumor rather than a set of thickness readings. In principle this captures the true burden of disease and small changes in it more faithfully than any diameter- or thickness-based method. The obstacle has always been labor: outlining the full extent of a mesothelioma rind slice by slice across a CT scan by hand is enormously time-consuming and impractical in routine care. This is where artificial intelligence changes the equation. Deep-learning segmentation models can outline the tumor automatically across every CT slice in minutes, making volumetry feasible at the scale of everyday practice and large clinical trials.
What Did the ARTIMES Study Show?
ARTIMES (Artificial intelligence-assisted volumetric response criteria) is the largest validation of AI volumetry in mesothelioma to date. Published in Lancet Oncology in 2026, it was a retrospective, multicohort, multicentre study built on 10,926 CT scans from 2,080 patients across 14 cohorts, including patients from major treatment trials.[2] The system automatically segmented tumor volume and applied response thresholds derived from the biology of how volume change relates to outcome, rather than borrowing cut-offs designed for diameter measurement.
The central finding was that volumetric response classified patients' prognosis more accurately than the current standard. AI volumetric response predicted overall survival with a concordance index of 0.83, compared with 0.73 for modified RECIST — a concordance index of 1.0 would be perfect prediction and 0.5 no better than chance.[2] The volumetric method also identified disease progression approximately five weeks earlier than modified RECIST, which could matter for decisions about whether to continue or switch a treatment.[2] The authors framed ARTIMES as a validation step and explicitly called for prospective testing before the criteria are adopted in routine care. For a patient-focused walkthrough of these findings, including what earlier progression detection could mean for clinical trial design, see this coverage of the ARTIMES results.[7]
How Did Automated Mesothelioma Volumetry Develop?
ARTIMES did not appear in isolation. In 2022, a team reported a fully automated, deep-learning system for measuring mesothelioma volume, validated in a blinded fashion on 183 CT datasets and compared against modified RECIST for both response classification and survival prediction.[3] That study established the essential proof of concept — that a neural network can segment mesothelioma reliably enough to measure it without a human tracing every slice. The broader field has since expanded well beyond measurement: a 2025 scoping review catalogued artificial-intelligence models applied across mesothelioma diagnosis, staging, and outcome prediction, reflecting rapid growth in the area.[8] ARTIMES represents the point at which this line of work reached top-tier, large-scale validation for the specific task of measuring treatment response.[2]
Is AI Volumetry Ready for Clinical Use?
Not yet, and the honest answer matters. An editorial accompanying ARTIMES in Lancet Oncology poses exactly this question in its title, "ready for deployment?"[4] The ARTIMES authors themselves frame the study as a validation step, pending prospective testing, before the criteria are ready for routine clinical use.[2] Open questions for any model at this stage include standardization — different scanners, imaging protocols, and software could produce different volumes — and generalizability, or whether a model validated on research cohorts performs equally well in ordinary hospitals. Prospective studies, in which the tool is tested on new patients in real time rather than on previously collected scans, are the accepted next step. For now, AI volumetric response criteria are best understood as a well-validated research tool with strong momentum, not an approved replacement for modified RECIST. They are investigational and not FDA-cleared.
How Does Volumetry Connect to PET Imaging and Staging?
Volumetric assessment on CT is one part of a wider shift toward quantitative mesothelioma imaging. On the metabolic side, FDG-PET/CT can measure metabolic tumor volume (MTV) and uptake intensity; in the NIPU trial, baseline MTV was independently associated with overall survival in patients treated with ipilimumab and nivolumab.[9] Reviews of PET/CT in mesothelioma now position metabolic imaging alongside anatomical volumetry as complementary ways to read response.[10] Formal staging is moving in the same direction: the 9th edition of the TNM staging system explicitly incorporates automated tumor-volume assessment and artificial intelligence into imaging-based staging, signaling that volumetry is entering official practice, not just research.[5] Together these developments point toward response and stage being defined by measured tumor volume rather than by a handful of thickness readings.
What Does This Mean for Patients?
For patients and families, AI volumetry is a reason for cautious optimism rather than an immediate change in care. If your treatment is being assessed today, your radiologist is most likely still using modified RECIST, which remains the accepted standard. Where AI volumetric tools are available — often at specialized mesothelioma centers or within clinical trials — they may add an earlier or more precise read on whether a therapy is working. Reasonable questions for your oncology team include whether volumetric measurement is available for your scans, whether you are enrolled in or eligible for a trial that uses it, and how any volumetric result compares with the standard assessment. Patients researching active studies can review the mesothelioma clinical trials overview.
Because mesothelioma care is costly, families are encouraged to document treatment and imaging expenses. Compensation for these costs may be available through asbestos trust funds, U.S. Department of Veterans Affairs benefits, and legal claims against the companies responsible for the asbestos exposure that caused the disease. Families weighing their options can consult a mesothelioma law firm such as Danziger & De Llano to understand which compensation channels apply to their circumstances.
Frequently Asked Questions
What are AI volumetric response criteria for mesothelioma?
They are a method that uses artificial intelligence to automatically measure the total volume of a pleural mesothelioma tumor on CT scans and track how that volume changes with treatment. Because mesothelioma grows as a rind around the lung rather than as round masses, volume can capture treatment response more faithfully than the diameter- and thickness-based criteria used today. The approach was validated at large scale by the ARTIMES study in 2026 but remains investigational and is not FDA-cleared.[2]
Is AI volumetry better than modified RECIST for mesothelioma?
In the 2026 ARTIMES validation study, AI volumetric response predicted overall survival more accurately than modified RECIST (concordance index 0.83 versus 0.73) and detected disease progression about five weeks earlier. These results are promising, but they come from a retrospective study; the authors call for prospective validation before AI volumetry replaces modified RECIST, which is still the accepted clinical standard.[2]
Can I get an AI volumetric scan for my mesothelioma today?
AI volumetric assessment is available mainly at specialized mesothelioma centers and within clinical trials rather than as routine care. Most treatment response today is still measured with modified RECIST. If you are interested, ask your oncology team whether volumetric measurement is available for your scans and whether a trial using it is an option.
See Also
- Mesothelioma Diagnosis and Staging
- FDG-PET/CT Scan for Mesothelioma
- Mesothelioma Diagnosis
- Immunotherapy for Mesothelioma
- Pleural Mesothelioma
- Mesothelioma Clinical Trials
References
- ↑ 1.0 1.1 Picone C, Porto A, Fusco R, Granata V, Brunese MC, Montanino A, et al. Malignant Pleural Mesothelioma CT Imaging: How to Measure It Correctly?. Cancer Control. 2025;32:10732748241301901. PMID 40091424.
- ↑ 2.00 2.01 2.02 2.03 2.04 2.05 2.06 2.07 2.08 2.09 2.10 Lipman KBWG, Wittenberg R, de Oliveira Taveira M, Smesseim I, Schmitz AMT, Boellaard TN, et al. Development and validation of artificial intelligence-assisted volumetric response criteria in pleural mesothelioma (ARTIMES): a retrospective, multicohort, multicentre study. Lancet Oncol. 2026;27(7):879-892. PMID 42309108.
- ↑ 3.0 3.1 Kidd AC, Anderson O, Cowell GW, Weir AJ, Voisey JP, Evison M, et al. Fully automated volumetric measurement of malignant pleural mesothelioma by deep learning AI: validation and comparison with modified RECIST response criteria. Thorax. 2022;77(12):1251-1259. PMID 35110367.
- ↑ 4.0 4.1 Blyth KG. Artificial intelligence volumetry in mesothelioma: ready for deployment?. Lancet Oncol. 2026;27(7):769-771. PMID 42309109.
- ↑ 5.0 5.1 Benamore RE, Xie C, Sobirov I. Staging pleural mesothelioma revitalized. Br J Radiol. 2026;99(1178):206-217. PMID 41339274.
- ↑ Martella S, Aiello MM, Bertaglia V, Cau R, Denaro N, et al. Malignant Pleural Mesothelioma: Staging and Radiological Response Criteria in Patients Treated with Immune Checkpoint Inhibitors. Target Oncol. 2024;19(1):13-28. PMID 38063957.
- ↑ Mesothelioma Lawyers Near Me. "AI Volumetric Scans Detect Mesothelioma Progression 5 Weeks Earlier."
- ↑ Ram M, Afrash MR, Moulaei K, Esmaeeli E, Khorashadizadeh MS, Garavand A, et al. Predicting Mesothelioma Using Artificial Intelligence: A Scoping Review of Common Models and Applications. Technol Cancer Res Treat. 2025;24:15330338251341053. PMID 40340549.
- ↑ Thunold S, Hernes E, Farooqi S, Öjlert ÅK, Francis RJ, Nowak AK, et al. Outcome prediction based on 18F-FDG PET/CT in patients with pleural mesothelioma treated with ipilimumab and nivolumab +/- UV1 telomerase vaccine. Eur J Nucl Med Mol Imaging. 2025;52(2):693-707. PMID 39133306.
- ↑ Badarna M, Keidar Z, Arnon-Sheleg E. Current and Future Perspective of PET/CT in Response Assessment of Malignant Pleural Mesothelioma. Semin Nucl Med. 2025;55(2):252-263. PMID 40021361.