The study evaluates whether a fine‑tuned open‑weight model (Gemma‑3‑12B) can match the performance of GPT‑4o in extracting multi‑label intracranial hemorrhage acuity from non‑contrast head‑CT reports. Using a 2×2 design that varied adaptation strategy (classification head vs. instruction fine‑tuning) and training‑data source (distilled real GPT‑4o labels vs. synthetic GPT‑4o‑generated reports), the distilled instruction‑tuned model achieved macro‑F1 scores comparable to GPT‑4o and surpassed the untuned base model. The key finding is that the source of training data—distilled real reports—was more important than the fine‑tuning method, and that the entire fine‑tuning and inference process fits on a single 24 GB consumer GPU.
By Aawez Mansuri, Kush Mehta, Mohammadreza Chavoshi, Jahanzaib Malik, Theodorus Dapamede, Frank Li, Rohan Isaac, Beatrice Brown-Mulry, Chiratidzo Rudado Sanyika, YoungSeok Jeon, Judy W. Gichoya, Ali Emami, Hari Trivedi
Converting free-text radiology reports into structured labels supports cohort building, quality assurance, and monitoring of clinical imaging models, but the strongest label extractors are hosted prop...
A specialized large multimodal model, LLaVA‑NeXT, was fine‑tuned on a curated two‑level curriculum of PET/CT image‑conversation pairs to interpret head and neck cancer scans. In external validation across four institutions, the model achieved high ROUGE and similarity scores and outperformed generalist models such as ChatGPT, with primary tumor classification accuracy of 83.14% internally and 69.03% externally. The study demonstrates that domain‑specific LMMs can provide fast, accurate diagnostic support for PET/CT imaging.
By Haengbok Chung, SunGyu Kim, Joo hyun Lee, Sangjin Bae, Min Jeong Cho, Minseok Suh, Jae Sung Lee
arXiv:2606. 07721v1 Announce Type: new Abstract: Objectives: Automatic data extraction from free-text radiology reports enables large-scale research, but few studies assessed the performance of large language models (LLMs) on Dutch neuroradiology reports.
By Kaouther Mouheb, Amos Pomp, Antoine Manenti, Romy de Haan, Farog Faghir, Joy Martens, Harro Seelaar, Francesco Mattace-Raso, Meike W. Vernooij, Frank J. Wolters, Stefan Klein, Esther E. Bron
The paper introduces NeuroFusion, an assistive brain‑MRI report generator that surfaces latent tumor signals from a frozen Mistral‑7B backbone. By adding discriminative field‑classifier heads over per‑lesion features, NeuroFusion restores accurate diagnoses (meningioma 0.92, metastasis 0.75) and improves prose quality while reducing latency 5–6×. A controlled negative result shows that overriding the decoder with a learned diagnosis pin harms performance, and grammar‑constrained decoding yields high schema‑validity (92.3%).
By Khawaja Murad ul Hassan, Ruqiyya Adil, Adil Qayyum, Rida Hassan, Asad Mansoor Khan, Muhammad Usman Akram, Mehran Ebrahimi
The paper introduces DAMM‑Net++, a 2.5D neural network for thoracic organ‑at‑risk and target volume segmentation that tackles inter‑slice surface incoherence, small low‑contrast target failure, and lack of per‑case reliability signals. Its core is an anatomy‑change‑aware bidirectional selective state‑space memory that propagates context across axial slices, complemented by a boundary‑aware decoder and an uncertainty head for calibrated per‑voxel confidence. Evaluations on 2,146 patients, an external cohort, and a reader study show high Dice scores (0.955), low HD95 (3.78 mm), significant time savings (75‑80 %) for clinicians, and improved junior‑reader performance, with the system fully integrated into a clinical workflow.
By Galib Ahmed, Istiak Ahmed, Aritra Islam Saswato, Asib Mostakim Fony, Kazi Shahriar Sanjid, Md. Tanzim Hossain, Md. Anwarul Islam, Md. Nishan Khan, Md. Misbah Khan, Labiba Faiza Karim, Jobaer Rahman, S M Hasibul Hoque, Rahnuma Shahrin Rista, Kamruzzaman Rumman, Md Arifur Rahman, Syed Md. Akram Hussain, Mohammad Ashrafuzzaman Khan, M. Monir Uddin