Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Transforming Multimodal Models into Action Models for Radiotherapy

T0 review · 4 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Reward-guided prompting of a multimodal model beats a DQN at simulated radiotherapy planning.

desk verdict A novel prompt-based action model for radiotherapy that has a solid pilot design but an unresolved action-space mismatch makes the headline comparison uncontrolled. read the letter →

arxiv 2502.04408 v1 pith:J6IS2CX5 submitted 2025-02-06 cs.LG cs.AI

Matteo Ferrante , Alessandra Carosi , Rolando Maria D Angelillo , Nicola Toschi This is my paper · ORCID
classification cs.LGcs.AI
keywords radiotherapytreatmentplanningmultimodalfoundationmodelsfew-shotreinforcementlearningMonteCarlodosesimulationgantryangleoptimizationprostatecancertext-to-planpromptingaction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a large pretrained multimodal model can be turned into an action model for radiotherapy treatment planning by giving it a Monte Carlo simulator, a scalar reward, and a few iterative prompts, without updating its weights. The authors report that this 'text-to-plan' prompting method achieved a mean reward of -211.88 in a simulated prostate-cancer environment, significantly better than a deep Q-network baseline (-259.26) and random gantry selection (-294.24), with ANOVA F(2,87)=67.66, p<0.001 and pairwise t-tests all p<0.001. If true, the result matters because it suggests that the anatomical and dosimetric knowledge already stored in large models can be harnessed to speed up and standardize a traditionally manual, time-consuming step in radiotherapy, without the long training runs that standard reinforcement learning requires.

What carries the argument

The mechanism is a reward-guided few-shot prompting loop that turns the model's hidden knowledge into actions. At each iteration, the model receives the patient CT, a textual goal, and the numeric reward from the previous plan, then outputs a JSON list of gantry angles; the Monte Carlo simulator (the MatRAD environment) computes a dose distribution and a reward that penalizes dose to organs at risk and rewards target homogeneity. The paper calls this 'optimizing the internal state of the pretrained model rather than its weights': the prompts and the model's internal activations are refined, not its parameters.

What would settle it

Run the same experiment on the MatRAD prostate data with a properly tuned deep Q-network (including hyperparameter search and a training budget comparable to the reported seven days) or with the published 3D-Unet DRL beam-angle method from the paper's reference [1], using the same reward function; if the text-to-plan mean reward no longer exceeds the RL baseline at the reported significance level, the central claim would be refuted. A cheaper additional falsifier is to check whether the advantage persists when the environment's maximum of five gantry angles is strictly enforced, since the paper's own appendix shows output with eight angles.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that a pretrained vision-language model, prompted with a CT image and a text instruction to choose gantry angles, can iteratively improve its plans when given scalar reward feedback from a Monte Carlo dose simulator, and that in the MatRAD prostate benchmark this procedure outperforms a conventional deep reinforcement-learning baseline and random selection in plan quality. The reported statistics, mean rewards of -211.88, -259.26, and -294.24 for text-to-plan, RL, and random, with pairwise t-tests all below p<0.001, are offered as evidence that the gap is not noise. The authors interpret the result as showing that large multimodal models carry enough implicit knowledge of anatomy and dosimetry to bootstrap plan optimization that otherwise would have to be learned from scratch.

Load-bearing premise

The load-bearing premise is that the deep Q-network baseline is a fair, representative instance of conventional reinforcement learning for this problem; if the baseline is weak or undertrained, the reported advantage of text-to-plan over 'RL' would not generalize to stronger RL methods.

Editorial extensions

If this is right

  • If the text-to-plan method is genuinely better, treatment planning can be improved in far fewer trials than RL needs, because the model starts with prior knowledge of anatomy and beam geometries.
  • The approach requires no gradient-based training of the policy, so it could be applied to a new patient or treatment site by simply changing the prompt and running the simulator, lowering the engineering barrier for automated planning.
  • The reward design used here, target homogeneity minus organ-at-risk penalties, can be reused as the objective for any future model or optimizer in the same environment.
  • The authors' observed 'dose bath' behavior, spreading beams to avoid critical structures, suggests the model can describe and follow a clinically sensible strategy, which is a step toward more explainable automated plans.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: The headline comparison rests on a single, apparently lightly tuned DQN baseline; a fair test of the paper's claim would pit text-to-plan against stronger published beam-angle reinforcement-learning methods, such as the 3D-Unet DRL approach the authors cite, on the same data and with comparable compute.
  • Inference: If the method scales beyond prostate and beyond this simulator, a natural extension is to let the model propose dose constraints or beam energies, not just angles, and to use open-weights multimodal models so the whole loop is reproducible.
  • Paper-stated limitation: The Discussion notes that the vision-language backbone sees only 2D slices and that using GPT-4 beyond its intended scope limits generalizability; these are acknowledged by the authors as barriers to clinical adoption.
  • Inference: The appendix sample outputs list eight gantry angles although the paper's environment description allows up to five; enforcing that constraint is a natural robustness check that could change the measured rewards.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The manuscript introduces 'text-to-plan', a framework that turns GPT-4V into an action model for radiotherapy beam-angle planning. The model is prompted with CT images and a planning task, receives a scalar reward from the MatRAD Monte Carlo simulator, and iteratively proposes new gantry angles in JSON. The authors compare this method with a random baseline and a DQN baseline on prostate cases, reporting mean rewards of -211.88, -259.26, and -294.24 respectively, with ANOVA and pairwise t-tests. They conclude that the approach outperforms conventional RL in quality and efficiency.

Significance. The proof-of-concept is timely and could be of interest to the ML-for-health community if the comparison is valid. Strengths include the use of an open-source simulator, a transparent interactive prompt log in the Appendix, explicit statistical testing, and the fact that no fitted parameters are introduced in the reward. However, the manuscript currently does not establish that the text-to-plan and DQN methods solve the same planning problem, the reported sample size is inconsistent with the reported degrees of freedom, and the efficiency claim is not backed by timing data. These issues are load-bearing for the headline claims.

major comments (4)
  1. [Sections 2.1-2.2 and Appendix] The environment and DQN are described as taking 'up to 5' gantry angles, but every text-to-plan interaction in the Appendix outputs eight angles (e.g., [10, 50, 90, 130, 170, 210, 250, 290]). The paper never states how these outputs are mapped to the five-beam environment. If the environment ignores or truncates angles, the scored plan is not the plan the model reasoned about; if all eight are accepted, the DQN was restricted to a strictly smaller action space. This makes the reward gap in Section 3 an uncontrolled comparison. The authors must specify the exact parsing and, ideally, rerun the comparison with matched action spaces.
  2. [Section 2.4 and Section 3] The text says the models produced 100 treatment plans, but the reported ANOVA F(2, 87) implies 90 total rewards (30 per group) and the pairwise t-tests with df = 58 imply 30 per group. This discrepancy changes the sample size and the statistical support for the headline result. Please report the exact per-group sample sizes and rerun the tests accordingly.
  3. [Abstract, Introduction, and Discussion] The claim that the method outperforms RL 'in both quality and efficiency' (Abstract) is not supported by any timing measurement in the paper. Section 2.2 reports that DQN training took about 7 days, but no wall-clock time, API call counts, or reward-versus-iteration curves are given for the text-to-plan method. Without such data, the efficiency claim should be removed or substantiated.
  4. [Section 2.2] The DQN baseline is under-specified. The paper gives no hyperparameters, no architecture definition beyond '3D convolutions and batch normalization,' no reward curve, and no convergence check, and it does not compare against published beam-angle RL methods such as the 3D-Unet DRL in reference [1]. The conclusion that the method 'outperforms conventional RL-based approaches' therefore rests on a single, unvalidated baseline. Please provide full training details, learning curves, and ideally a comparison to a published RL baseline.
minor comments (4)
  1. [Appendix] The text 'There is now training' appears to be a typo for 'There is no training'; similarly, 'trought' should be 'through'.
  2. [Section 2.1, Eq. (1)] The symbol P is used both as the OAR penalty and as an exponent in the same equation; please clarify whether P is a multiplicative penalty scalar or an exponent, and use distinct notation if needed.
  3. [Figure 2] The DVH panel lacks axis labels and a legend identifying which curve corresponds to which structure, and the box plot panel does not state the sample size or the treatment of outliers.
  4. [Section 2.3] The description 'optimizing the internal state of the large pretrained model rather than its weights' is vague; clarify what is actually updated between iterations, such as the conversational context or prompt history.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found: the text-to-plan result is an empirical comparison against baselines, and the reward-guided refinement is standard RL feedback rather than a definitional shortcut.

full rationale

The paper's central claim that the text-to-plan method outperforms conventional RL-based approaches rests on measured reward distributions from 100 plans per method (mean rewards -211.88 vs -259.26 vs -294.24; ANOVA F(2,87)=67.66, p<0.001; pairwise t-tests all p<0.001) reported in Section 3. The reward function in Section 2.1 is defined independently of the model outputs, and the same scalar reward is used both as the optimization signal during iterative prompting and as the evaluation metric; this is standard practice in reinforcement learning and does not make the reported outcome true by construction. No fitted parameter is renamed as a prediction, and no load-bearing result is justified only by a self-citation chain: the environment is built on the external MatRAD toolkit [7] and the action model is an externally pretrained GPT-4V [19]. The potential internal inconsistency between the eight-angle JSON outputs shown in the Appendix and the 'up to 5' beam environment/DQN action space described in Sections 2.1-2.2 is a correctness and controllability concern, not a circularity, because it does not reduce the claimed performance advantage to an identity or to the paper's own assumptions. Accordingly, no circular step can be exhibited, and the appropriate score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The empirical claim rests on the faithfulness of the simulator and reward, the transfer of pretrained knowledge to 2D slices, and the fairness of the DQN baseline. The free parameters are hand-chosen reward weights and design limits, and the 8-to-5 angle mapping is an unstated assumption.

free parameters (4)
  • Rmax = 1
    Set by hand in Section 2.1; it scales the PTV homogeneity term and, jointly with P, determines the trade-off baked into the reward used for both optimization and evaluation.
  • P = 1
    Set by hand in Section 2.1; it sets the penalty for exceeding OAR dose limits. No sensitivity analysis is reported.
  • Prescription dose Ti = 100 Gy
    Chosen in Section 2.1; unusually high for prostate and shifts the reward landscape, but is treated as a fixed clinical input.
  • Maximum beam count = 5
    Environment action limit in Section 2.1; the model outputs 8 angles in the appendix, so the effective number of beams used in the simulations is unclear.
assumptions (4)
  • domain assumption MatRAD's dose calculation is an accurate surrogate for clinical dose distributions in 3D conformal radiotherapy.
    The entire reward and evaluation pipeline treats MatRAD output as ground truth (Section 2.1); no comparison against a clinical treatment planning system is provided.
  • domain assumption The scalar reward function is a valid proxy for treatment plan quality.
    The reward is used both to guide the model's refinements and to measure success (Sections 2.1 and 2.4), but it is a simplified formula with no clinical validation.
  • domain assumption GPT-4V's pretrained knowledge transfers from 2D CT slices to useful anatomical reasoning about beam angles.
    The method relies on this in Section 2.3 and the appendix; the authors themselves note in Section 4 that vision is limited to 2D slices.
  • ad hoc to paper The unstated mapping from the model's 8 output angles to the environment's 5-beam cap preserves the plan that was scored.
    The appendix shows 8 angles per iteration while Section 2.1 limits input to 5; without a described truncation or selection rule, the scored action is not well defined.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transforming Multimodal Models into Action Models for Radiotherapy." pith.science (2026). https://pith.science/paper/J6IS2CX5

@misc{pith2026250204408,
  author       = {Pith},
  title        = {Pith review of: Transforming Multimodal Models into Action Models for Radiotherapy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J6IS2CX5}},
  note         = {Machine review of arXiv:2502.04408}
}
read the original abstract

Radiotherapy is a crucial cancer treatment that demands precise planning to balance tumor eradication and preservation of healthy tissue. Traditional treatment planning (TP) is iterative, time-consuming, and reliant on human expertise, which can potentially introduce variability and inefficiency. We propose a novel framework to transform a large multimodal foundation model (MLM) into an action model for TP using a few-shot reinforcement learning (RL) approach. Our method leverages the MLM's extensive pre-existing knowledge of physics, radiation, and anatomy, enhancing it through a few-shot learning process. This allows the model to iteratively improve treatment plans using a Monte Carlo simulator. Our results demonstrate that this method outperforms conventional RL-based approaches in both quality and efficiency, achieving higher reward scores and more optimal dose distributions in simulations on prostate cancer data. This proof-of-concept suggests a promising direction for integrating advanced AI models into clinical workflows, potentially enhancing the speed, quality, and standardization of radiotherapy treatment planning.

Figures

Figures reproduced from arXiv: 2502.04408 by the authors.

Figure 1
Figure 1. Workflow of an action model for treatment planning. The model processes a patient’s CT scan to determine optimal gantry angles, generating a dose distribution evaluated to produce a reward score. An example shows an initial plan with a reward of -300, improved to -180 after refinement by the multimodal pretrained vision-language model, highlighting iterative enhancement with Monte Carlo simulation. Developing automa… view at source ↗
Figure 2
Figure 2. A) Dose Volume Histograms (DVHs) for baseline, RL, and Text-to-Plan mod￾els, showing dose distribution to the target (PTV) and organs at risk (OARs). B) Dose distribution maps for the baseline (left), RL (middle), and Text-to-Plan (right) models. C) Box plot of the reward values for Text-to-Plan, RL, and random methods, demonstrating the superior performance of the Text-to-Plan approach in optimizing treatment plans… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI

    cs.LG 2025-08 reject novelty 3.0 of 10

    A survey of reinforcement learning in healthcare that frames RL as a paradigm shift from prediction to agentive clinical intelligence.

Reference graph

Works this paper leans on

28 extracted references · 25 canonical work pages · cited by 1 Pith paper

  1. [1]

    Bao, P., Wang, G., Yang, R., Dong, B.: Deep reinforcement learning for beam angle optimization of intensity-modulated radiation therapy (2023)

    work page 2023

  2. [2]

    Prac- tical Radiation Oncology 6(6), 442–449 (Nov-Dec 2016)

    Berry, S.L., Boczkowski, A., Ma, R., Mechalakos, J., Hunt, M.: Interobserver vari- ability in radiation therapy plan output: Results of a single-institution study. Prac- tical Radiation Oncology 6(6), 442–449 (Nov-Dec 2016). https://doi.org/10. 1016/j.prro.2016.04.005, https://doi.org/10.1016/j.prro.2016.04.005

    work page doi:10.1016/j.prro.2016.04.005 2016

  3. [3]

    Medical Dosimetry 43(2), 168–176 (Summer 2018).https://doi.org/10.1016/j.meddos

    Bodensteiner, D.: Raystation: External beam treatment planning system. Medical Dosimetry 43(2), 168–176 (Summer 2018).https://doi.org/10.1016/j.meddos. 2018.02.013, https://doi.org/10.1016/j.meddos.2018.02.013

    work page doi:10.1016/j.meddos 2018

  4. [4]

    In: Schlegel, W., Bortfeld, T., Grosu, A.L

    Bortfeld, T., Thieke, C.: Optimization of treatment plans, inverse planning. In: Schlegel, W., Bortfeld, T., Grosu, A.L. (eds.) New Technologies in Radiation Oncology, pp. 257–276. Medical Radiology, Springer, Berlin, Heidelberg (2006). https://doi.org/10.1007/3-540-29999-8_17

    work page doi:10.1007/3-540-29999-8_17 2006

  5. [5]

    Brockman, G., Cheung, V., Pettersson, L., Schneider, J., Schulman, J., Tang, J., Zaremba, W.: Openai gym (2016)

    work page 2016

  6. [6]

    Chaput, G., Regnier, L.: Radiotherapy: Clinical pearls for primary care. Can. Fam. Physician 67(10), 753–757 (Oct 2021)

    work page 2021

  7. [7]

    In: Jaffray, D.A

    Cisternas, E., Mairani, A., Ziegenhein, P., Jäkel, O., Bangert, M.: matrad - a multi-modality open source 3d treatment planning toolkit. In: Jaffray, D.A. (ed.) World Congress on Medical Physics and Biomedical Engineering, June 7-12, 2015, Toronto, Canada. pp. 1608–1611. IFMBE Proceedings, Springer International Publishing, Cham (2015). https://doi.org/10...

    work page doi:10.1007/978-3-319-19387-8_391 2015

  8. [8]

    Communications Medicine 3, 141 (2023)

    Clusmann, J., Kolbinger, F.R., Muti, H.S., et al.: The future landscape of large language models in medicine. Communications Medicine 3, 141 (2023). https://doi.org/10.1038/s43856-023-00370-1, https://doi.org/10. 1038/s43856-023-00370-1

    doi:10.1038/s43856-023-00370-1 2023

Show all 28 references
  1. [9]

    Artificial Intelligence in Medicine 121, 102193 (Nov 2021)

    Ebrahimi, S., Lim, G.J.: A reinforcement learning approach for finding op- timal policy of adaptive radiation therapy considering uncertain tumor bio- logical response. Artificial Intelligence in Medicine 121, 102193 (Nov 2021). https://doi.org/10.1016/j.artmed.2021.102193, ht...

    2021

  2. [10]

    Frimodig, S., Enqvist,P., Carlsson, M.,Mercier, C.:Comparing optimization meth- ods for radiation therapy patient scheduling using different objectives (2023)

    2023

  3. [11]

    Kalakonda, S.S., Maheshwari, S., Sarvadevabhatla, R.K.: Action-gpt: Leveraging large-scale language models for improved and generalized action generation (2023) 10 Ferrante et al

    2023

  4. [12]

    JMA J6(3), 342–345 (Jul 2023)

    Katano, A., Minamitani, M., Yamashita, H., Nakagawa, K.: National Survey of Ra- diotherapy Utilization Trends from 2015 to 2019, Based on the National Database of Health Insurance Claims and Specific Health Checkups of Japan. JMA J6(3), 342–345 (Jul 2023)

    2015

  5. [13]

    Nature Com- munications 15, 654 (2024)

    Ma, J., He, Y., Li, F., et al.: Segment anything in medical images. Nature Com- munications 15, 654 (2024). https://doi.org/10.1038/s41467-024-44824-z, https://doi.org/10.1038/s41467-024-44824-z

    2024 doi

  6. [14]

    Mayadev, J.S., Ke, G., Mahantshetty, U., Pereira, M.D., Tarnawski, R., Toita, T.: Global challenges of radiotherapy for the treatment of locally advanced cervical cancer. Int. J. Gynecol. Cancer32(3), 436–445 (Mar 2022)

    2022

  7. [15]

    Technology in Can- cer Research & Treatment17, 1533033818780064 (Jan 1 2018).https://doi.org/ 10.1177/1533033818780064, https://doi.org/10.1177/1533033818780064

    McConnell, K.A., Marston, T., Zehren, B.E., Lirani, A., Stanley, D.N., Bishop, A., Crownover, R., Eng, T., Shi, Z., Li, Y., Baacke, D., Kirby, N., Rasmussen, K., Papanikolaou, N., Gutierrez, A.N.: Dosimetric evaluation of pinnacle’s automated treatment planning software to man...

    2018 doi

  8. [16]

    McDuff,D.,Schaekermann,M.,Tu,T.,Palepu,A.,Wang,A.,Garrison,J.,Singhal, K., Sharma, Y., Azizi, S., Kulkarni, K., Hou, L., Cheng, Y., Liu, Y., Mahdavi, S.S., Prakash, S., Pathak, A., Semturs, C., Patel, S., Webster, D.R., Dominowska, E., Gottweis, J., Barral, J., Chou, K., Corra...

    2023

  9. [17]

    iScience 27(5), 109713 (2024)

    Meng, X., Yan, X., Zhang, K., Liu, D., Cui, X., Yang, Y., Zhang, M., Cao, C., Wang, J., Wang, X., Gao, J., Wang, Y.G.S., ming Ji, J., Qiu, Z., Li, M., Qian, C., Guo, T., Ma, S., Wang, Z., Guo, Z., Lei, Y., Shao, C., Wang, W., Fan, H., Tang, Y.D.: The application of large langu...

    2024

  10. [18]

    Journal of Applied Clinical Medical Physics 22(8), 16–44 (2021)

    Momin, S., Fu, Y., Lei, Y., et al.: Knowledge-based radiation treatment plan- ning: A data-driven method survey. Journal of Applied Clinical Medical Physics 22(8), 16–44 (2021). https://doi.org/10.1002/acm2.13337, https://doi.org/ 10.1002/acm2.13337

    2021 doi

  11. [19]

    OpenAI, Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., Avila, R., Babuschkin, I., Balaji, S., Balcom, V., Baltescu, P., Bao, H., Bavarian, M., Belgum, J., Bello, I., Berdine, J., Bernadett-Shapir...

    2024

  12. [20]

    Medical Physics48(4), 1909–1920 (Apr 2021)

    Shen, C., Chen, L., Gonzalez, Y., Jia, X.: Improving efficiency of training a virtual treatment planner network via knowledge-guided deep reinforcement learning for intelligent automatic treatment planning of radiotherapy. Medical Physics48(4), 1909–1920 (Apr 2021). https://do...

    1909 doi

  13. [21]

    A Bradford Book, Cambridge, MA, USA (2018)

    Sutton, R.S., Barto, A.G.: Reinforcement Learning: An Introduction. A Bradford Book, Cambridge, MA, USA (2018)

    2018

  14. [22]

    Tortora,M.,Cordelli,E.,Sicilia,R.,Miele,M.,Matteucci,P.,Iannello,G.,Ramella, S.,Soda,P.:Deepreinforcementlearningforfractionatedradiotherapyinnon-small cell lung carcinoma. Artif. Intell. Med.119(102137), 102137 (Sep 2021)

    2021

  15. [23]

    Frontiers in Oncology13, 1124458 (Feb 3 2023).https://doi.org/ 10.3389/fonc.2023.1124458, https://doi.org/10.3389/fonc.2023.1124458

    Wang, H., Bai, X., Wang, Y., Lu, Y., Wang, B.: An integrated solution of deep reinforcement learning for automatic imrt treatment planning in non-small-cell lung cancer. Frontiers in Oncology13, 1124458 (Feb 3 2023).https://doi.org/ 10.3389/fonc.2023.1124458, https://doi.org/1...

    2023

  16. [24]

    Xie, Q., Chen, Q., Chen, A., Peng, C., Hu, Y., Lin, F., Peng, X., Huang, J., Zhang, J., Keloth, V., Zhou, X., He, H., Ohno-Machado, L., Wu, Y., Xu, H., Bian, J.: Me llama: Foundation large language models for medical applications (2024)

    2024

  17. [25]

    Physics in Medicine and Biology 67(22), – (Nov 11 2022).https://doi.org/10.1088/1361-6560/ac9cb3, https: //doi.org/10.1088/1361-6560/ac9cb3

    Xu, L., Zhu, S., Wen, N.: Deep reinforcement learning and its applications in medical imaging and radiation therapy: a survey. Physics in Medicine and Biology 67(22), – (Nov 11 2022).https://doi.org/10.1088/1361-6560/ac9cb3, https: //doi.org/10.1088/1361-6560/ac9cb3

    2022 doi

  18. [26]

    Yagnik, N., Jhaveri, J., Sharma, V., Pila, G.: Medlm: Exploring language models for medical question answering systems (2024)

    2024

  19. [27]

    Y., Deasy, J.O.: Automated and clinically optimal treatment planning for cancer radiotherapy

    Zarepisheh, M., Hong, L., Zhou, Y., Huang, Q., Yang, J., Jhanwar, G., Pham, H.D., Dursun, P., Zhang, P., Hunt, M.A., Mageras, G.S., Yang, J.T., Yamada, 12 Ferrante et al. Y., Deasy, J.O.: Automated and clinically optimal treatment planning for cancer radiotherapy. INFORMS J. A...

    2022

  20. [28]

    gantry_angles

    Zhou, S.K., Le, H.N., Luu, K., Nguyen, H.V., Ayache, N.: Deep reinforcement learning in medical imaging: A literature review (2021) Transforming Multimodal Models into Action Models for Radiotherapy 13 Appendix Here we report a complete example of a conversion of the action mo...

    2021

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.