Markerless dynamic tumor tracking (MDTT) radiotherapy using diaphragm as a surrogate for liver targets.

PURPOSE: To assess the feasibility of using the diaphragm as a surrogate for liver targets during MDTT. METHODS: Diaphragm as surrogate for markers: a dome-shaped phantom with implanted markers was fabricated and underwent dual-orthogonal fluoroscopy sequences on the Vero4DRT linac. Ten patients participated in an IRB-approved, feasibility study to assess the MDTT workflow. All images were analyzed using an in-house program to back-project the diaphragm/markers position to the isocenter plane. ExacTrac imager log files were analyzed. Diaphragm as tracking structure for MDTT: The phantom "diaphragm" was contoured as a markerless tracking structure (MTS) and exported to Vero4DRT/ExacTrac. A single field plan was delivered to the phantom film plane under static and MDTT conditions. In the patient study, the diaphragm tracking structure was contoured on CT breath-hold-exhale datasets. The MDTT workflow was applied until just prior to MV beam-on. RESULTS: Diaphragm as surrogate for markers: phantom data confirmed the in-house 3D back-projection program was functioning as intended. In patients, the diaphragm/marker relative positions had a mean ± RMS difference of 0.70 ± 0.89, 1.08 ± 1.26, and 0.96 ± 1.06 mm in ML, SI, and AP directions. Diaphragm as tracking structure for MDTT: Building a respiratory-correlation model using the diaphragm as surrogate for the implanted markers was successful in phantom/patients. During the tracking verification imaging step, the phantom mean ± SD difference between the image-detected and predicted "diaphragm" position was 0.52 ± 0.18 mm. The 2D film gamma (2%/2 mm) comparison (static to MDTT deliveries) was 98.2%. In patients, the mean difference between the image-detected and predicted diaphragm position was 2.02 ± 0.92 mm. The planning target margin contribution from MDTT diaphragm tracking is 2.2, 5.0, and 4.7 mm in the ML, SI, and AP directions. CONCLUSION: In phantom/patients, the diaphragm motion correlated well with markers' motion and could be used as a surrogate. MDTT workflows using the diaphragm as the MTS is feasible using the Vero4DRT linac and could replace the need for implanted markers for liver radiotherapy.

Exploring Offline Large Language Models for Clinical Information Extraction: A Study of Renal Histopathological Reports of Lupus Nephritis Patients.

Open source, lightweight and offline generative large language models (LLMs) hold promise for clinical information extraction due to their suitability to operate in secured environments using commodity hardware without token cost. By creating a simple lupus nephritis (LN) renal histopathology annotation schema and generating gold standard data, this study investigates prompt-based strategies using three state-of-the-art lightweight LLMs, namely BioMistral-DARE-7B (BioMistral), Llama-2-13B (Llama 2), and Mistral-7B-instruct-v0.2 (Mistral). We examine the performance of these LLMs within a zero-shot learning environment for renal histopathology report information extraction. Incorporating four prompting strategies, including combinations of batch prompt (BP), single task prompt (SP), chain of thought (CoT) and standard simple prompt (SSP), our findings indicate that both Mistral and BioMistral consistently demonstrated higher performance compared to Llama 2. Mistral recorded the highest performance, achieving an F1-score of 0.996 [95% CI: 0.993, 0.999] for extracting the numbers of various subtypes of glomeruli across all BP settings and 0.898 [95% CI: 0.871, 0.921] in extracting relational values of immune markers under the BP+SSP setting. This study underscores the capability of offline LLMs to provide accurate and secure clinical information extraction, which can serve as a promising alternative to their heavy-weight online counterparts.