Fine-Tuning Open-Source Large Language Models to Improve Their Performance on Radiation Oncology Tasks: A Feasibility Study to Investigate Their Potential Clinical Applications in Radiation Oncology

Background: The radiation oncology clinical practice involves many steps relying on the dynamic interplay of abundant text data. Large language models have displayed remarkable capabilities in processing complex text information. But their direct applications in specific fields like radiation oncology remain underexplored. Purpose: This study aims to investigate whether fine-tuning LLMs with domain knowledge can improve the performance on Task (1) treatment regimen generation, Task (2) treatment modality selection (photon, proton, electron, or brachytherapy), and Task (3) ICD-10 code prediction in radiation oncology. Methods: Data for 15,724 patient cases were extracted. Cases where patients had a single diagnostic record, and a clearly identifiable primary treatment plan were selected for preprocessing and manual annotation to have 7,903 cases of the patient diagnosis, treatment plan, treatment modality, and ICD-10 code. Each case was used to construct a pair consisting of patient diagnostics details and an answer (treatment regimen, treatment modality, or ICD-10 code respectively) for the supervised fine-tuning of these three tasks. Open source LLaMA2-7B and Mistral-7B models were utilized for the fine-tuning with the Low-Rank Approximations method. Accuracy and ROUGE-1 score were reported for the fine-tuned models and original models. Clinical evaluation was performed on Task (1) by radiation oncologists, while precision, recall, and F-1 score were evaluated for Task (2) and (3). One-sided Wilcoxon signed-rank tests were used to statistically analyze the results. Results: Fine-tuned LLMs outperformed original LLMs across all tasks with p-value <= 0.001. Clinical evaluation demonstrated that over 60% of the fine-tuned LLMs-generated treatment regimens were clinically acceptable. Precision, recall, and F1-score showed improved performance of fine-tuned LLMs.

Noisy probing dose facilitated dose prediction for pencil beam scanning proton therapy: physics enhances generalizability

Purpose: Prior AI-based dose prediction studies in photon and proton therapy often neglect underlying physics, limiting their generalizability to handle outlier clinical cases, especially for pencil beam scanning proton therapy (PBSPT). Our aim is to design a physics-aware and generalizable AI-based PBSPT dose prediction method that has the underlying physics considered to achieve high generalizability to properly handle the outlier clinical cases. Methods and Materials: This study analyzed PBSPT plans of 103 prostate and 78 lung cancer patients from our institution,with each case comprising CT images, structure sets, and plan doses from our Monte-Carlo dose engine (serving as the ground truth). Three methods were evaluated in the ablation study: the ROI-based method, the beam mask and sliding window method, and the noisy probing dose method. Twelve cases with uncommon beam angles or prescription doses tested the methods' generalizability to rare treatment planning scenarios. Performance evaluation used DVH indices, 3D Gamma passing rates (3%/2mm/10%), and dice coefficients for dose agreement. Results: The noisy probing dose method showed improved agreement of DVH indices, 3D Gamma passing rates, and dice coefficients compared to the conventional methods for the testing cases. The noisy probing dose method showed better generalizability in the 6 outlier cases than the ROI-based and beam mask-based methods with 3D Gamma passing rates (for prostate cancer, targets: 89.32%$\pm$1.45% vs. 93.48%$\pm$1.51% vs. 96.79%$\pm$0.83%, OARs: 85.87%$\pm$1.73% vs. 91.15%$\pm$1.13% vs. 94.29%$\pm$1.01%). The dose predictions were completed within 0.3 seconds. Conclusions: We've devised a novel noisy probing dose method for PBSPT dose prediction in prostate and lung cancer patients. With more physics included, it enhances the generalizability of dose prediction in handling outlier clinical cases.

RadOnc-GPT: A Large Language Model for Radiation Oncology

This paper presents RadOnc-GPT, a large language model specialized for radiation oncology through advanced tuning methods. RadOnc-GPT was finetuned on a large dataset of radiation oncology patient records and clinical notes from the Mayo Clinic in Arizona. The model employs instruction tuning on three key tasks - generating radiotherapy treatment regimens, determining optimal radiation modalities, and providing diagnostic descriptions/ICD codes based on patient diagnostic details. Evaluations conducted by comparing RadOnc-GPT outputs to general large language model outputs showed that RadOnc-GPT generated outputs with significantly improved clarity, specificity, and clinical relevance. The study demonstrated the potential of using large language models fine-tuned using domain-specific knowledge like RadOnc-GPT to achieve transformational capabilities in highly specialized healthcare fields such as radiation oncology.

Deep-Learning-based Fast and Accurate 3D CT Deformable Image Registration in Lung Cancer

Purpose: In some proton therapy facilities, patient alignment relies on two 2D orthogonal kV images, taken at fixed, oblique angles, as no 3D on-the-bed imaging is available. The visibility of the tumor in kV images is limited since the patient's 3D anatomy is projected onto a 2D plane, especially when the tumor is behind high-density structures such as bones. This can lead to large patient setup errors. A solution is to reconstruct the 3D CT image from the kV images obtained at the treatment isocenter in the treatment position. Methods: An asymmetric autoencoder-like network built with vision-transformer blocks was developed. The data was collected from 1 head and neck patient: 2 orthogonal kV images (1024x1024 voxels), 1 3D CT with padding (512x512x512) acquired from the in-room CT-on-rails before kVs were taken and 2 digitally-reconstructed-radiograph (DRR) images (512x512) based on the CT. We resampled kV images every 8 voxels and DRR and CT every 4 voxels, thus formed a dataset consisting of 262,144 samples, in which the images have a dimension of 128 for each direction. In training, both kV and DRR images were utilized, and the encoder was encouraged to learn the jointed feature map from both kV and DRR images. In testing, only independent kV images were used. The full-size synthetic CT (sCT) was achieved by concatenating the sCTs generated by the model according to their spatial information. The image quality of the synthetic CT (sCT) was evaluated using mean absolute error (MAE) and per-voxel-absolute-CT-number-difference volume histogram (CDVH). Results: The model achieved a speed of 2.1s and a MAE of <40HU. The CDVH showed that <5% of the voxels had a per-voxel-absolute-CT-number-difference larger than 185 HU. Conclusion: A patient-specific vision-transformer-based network was developed and shown to be accurate and efficient to reconstruct 3D CT images from kV images.

Evaluating Large Language Models on a Highly-specialized Topic, Radiation Oncology Physics

We present the first study to investigate Large Language Models (LLMs) in answering radiation oncology physics questions. Because popular exams like AP Physics, LSAT, and GRE have large test-taker populations and ample test preparation resources in circulation, they may not allow for accurately assessing the true potential of LLMs. This paper proposes evaluating LLMs on a highly-specialized topic, radiation oncology physics, which may be more pertinent to scientific and medical communities in addition to being a valuable benchmark of LLMs. We developed an exam consisting of 100 radiation oncology physics questions based on our expertise at Mayo Clinic. Four LLMs, ChatGPT (GPT-3.5), ChatGPT (GPT-4), Bard (LaMDA), and BLOOMZ, were evaluated against medical physicists and non-experts. ChatGPT (GPT-4) outperformed all other LLMs as well as medical physicists, on average. The performance of ChatGPT (GPT-4) was further improved when prompted to explain first, then answer. ChatGPT (GPT-3.5 and GPT-4) showed a high level of consistency in its answer choices across a number of trials, whether correct or incorrect, a characteristic that was not observed in the human test groups. In evaluating ChatGPTs (GPT-4) deductive reasoning ability using a novel approach (substituting the correct answer with "None of the above choices is the correct answer."), ChatGPT (GPT-4) demonstrated surprising accuracy, suggesting the potential presence of an emergent ability. Finally, although ChatGPT (GPT-4) performed well overall, its intrinsic properties did not allow for further improvement when scoring based on a majority vote across trials. In contrast, a team of medical physicists were able to greatly outperform ChatGPT (GPT-4) using a majority vote. This study suggests a great potential for LLMs to work alongside radiation oncology experts as highly knowledgeable assistants.