Towards artificial intelligence based preoperative dosimetry for liver 90 Y selective internal radiation therapy
Abstract
Objective. Selective internal radiation therapy delivers Y‐90 microspheres through the hepatic arterial system to achieve tumoricidal absorbed dose while limiting normal‐tissue irradiation. Current pre‐treatment dosimetry requires angiography and 99 mTc‐MAA imaging, resulting in an invasive and resource‐intensive workflow. This study evaluates the feasibility of estimating post‐therapy Y‐90 dose distribution directly from diagnostic multiphasic CT using an automated deep‐learning pipeline, without requiring angiography or surrogate‐particle imaging. Approach. In 152 treatments from the multicenter PROACTIF registry, arterial‐ and portal‐phase CT, together with clinically documented perfusion territories and injected activities, were processed through automated segmentation, rigid registration, differential CT generation, and a generative model synthesizing a PET‐like Y‐90 activity map. Absorbed‐dose maps were computed using the Local Dose Deposition method, and predictive accuracy was assessed using tumor dose–volume metrics and voxelwise spatial agreement. Main results. For the best‐performing configuration, the median tumor D70 error was 23 Gy, with substantial inter‐patient variability reflecting the challenges of CT‐only dose prediction. High‐dose coverage (V200) showed a similar pattern. In contrast, spatial agreement was consistently strong, with perfused‐volume gamma passing rates around 93% and tumor‐level rates around 85%, indicating that the macro‐scale distribution of microspheres can be approximated from diagnostic CT. The full pipeline was fully automated and produced predictive dosimetry in under 30 S. Significance. Pre‐angiography CT contains exploitable information about arterial enhancement and perfusion patterns that enables feasible CT‐only prediction of Y‐90 spatial distribution. While tumor‐level absorbed‐dose metrics remain variable, the strong spatial agreement suggests potential utility as an early planning tool to support patient selection or guide catheterization strategy prior to angiography.
Related
- Selective Recognition, Covalent Linkage and Chemical Exchange of Artificial Cells through Engineered α-Hemolysin
- Toward Enactive Artificial Intelligence
- Growing up with artificial intelligence: An amplifier of inequalities and a public health issue
- Notification: Artificial Intelligence in Autism Spectrum Disorder Diagnosis: A Scoping Review of Face, Voice, and Text Analysis Methods.
- Early Detection of Red Palm Weevil Infestation: A Systematic Review of Sensing Technologies, Artificial Intelligence, and Field Readiness
BibTeX
@article{morelcorlu2026towards,
title = {{Towards artificial intelligence based preoperative dosimetry for liver 90 Y selective internal radiation therapy}},
author = {Ewan Morel--Corlu and Florent Lalys and Antoine Petit and Pascal Haigron and Simon Esneault and Y Rolland and Mireille Garreau},
journal = {Physics in Medicine and Biology},
year = {2026},
month = oct,
doi = {10.1088/1361-6560/aeabea},
url = {https://hal.science/hal-05785902v1},
note = {CC BY},
}
RIS
TY - JOUR
TI - Towards artificial intelligence based preoperative dosimetry for liver 90 Y selective internal radiation therapy
AU - Ewan Morel--Corlu
AU - Florent Lalys
AU - Antoine Petit
AU - Pascal Haigron
AU - Simon Esneault
AU - Y Rolland
AU - Mireille Garreau
PY - 2026
DA - 2026/10/08
JO - Physics in Medicine and Biology
DO - 10.1088/1361-6560/aeabea
UR - https://hal.science/hal-05785902v1
AB - Objective. Selective internal radiation therapy delivers Y‐90 microspheres through the hepatic arterial system to achieve tumoricidal absorbed dose while limiting normal‐tissue irradiation. Current pre‐treatment dosimetry requires angiography and 99 mTc‐MAA imaging, resulting in an invasive and resource‐intensive workflow. This study evaluates the feasibility of estimating post‐therapy Y‐90 dose distribution directly from diagnostic multiphasic CT using an automated deep‐learning pipeline, without requiring angiography or surrogate‐particle imaging. Approach. In 152 treatments from the multicenter PROACTIF registry, arterial‐ and portal‐phase CT, together with clinically documented perfusion territories and injected activities, were processed through automated segmentation, rigid registration, differential CT generation, and a generative model synthesizing a PET‐like Y‐90 activity map. Absorbed‐dose maps were computed using the Local Dose Deposition method, and predictive accuracy was assessed using tumor dose–volume metrics and voxelwise spatial agreement. Main results. For the best‐performing configuration, the median tumor D70 error was 23 Gy, with substantial inter‐patient variability reflecting the challenges of CT‐only dose prediction. High‐dose coverage (V200) showed a similar pattern. In contrast, spatial agreement was consistently strong, with perfused‐volume gamma passing rates around 93% and tumor‐level rates around 85%, indicating that the macro‐scale distribution of microspheres can be approximated from diagnostic CT. The full pipeline was fully automated and produced predictive dosimetry in under 30 S. Significance. Pre‐angiography CT contains exploitable information about arterial enhancement and perfusion patterns that enables feasible CT‐only prediction of Y‐90 spatial distribution. While tumor‐level absorbed‐dose metrics remain variable, the strong spatial agreement suggests potential utility as an early planning tool to support patient selection or guide catheterization strategy prior to angiography.
SN - 0031-9155
SN - 1361-6560
ER -