Pith. sign in

REVIEW 4 major objections 5 minor 6 cited by

Read Like a Radiologist: Efficient Vision-Language Model for 3D Medical Imaging Interpretation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Reading CT slices in sequence beats fixed 3D patch encoders for medical VLMs.

desk verdict Z-former is a sensible 3D VLM backbone, but the experiments conflate architecture with pretraining and data, so the 'surpasses' claim is not yet supported. read the letter →

arxiv 2412.13558 v1 pith:ZQFA4EX2 submitted 2024-12-18 eess.IV cs.CLcs.CVcs.LG

Changsun Lee , Sangjoon Park , Cheong-Il Shin , Woo Hee Choi , Hyun Jeong Park , Jeong Eun Lee , Jong Chul Ye This is my paper · ORCID
classification eess.IVcs.CLcs.CVcs.LG
keywords 3Dmedicalimagingvision-languagemodelradiologyreportgenerationself-supervisedlearningvisiontransformerZ-formerCT-RATErectalMRI
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 argues that 3D medical vision-language models should read a scan the way radiologists do: slice by slice, then synthesize across slices, planes, and phases, rather than carving the volume into fixed 3D patches. It introduces MS-VLM, built from a self-supervised 2D vision transformer that embeds each slice, a sparse-attention Z-former that captures inter-slice dependencies, and a perceiver resampler that turns the slice sequence into a fixed-size visual prompt for a language model. On the CT-RATE chest CT benchmark and an in-house rectal MRI dataset, MS-VLM reports higher natural-language and clinical-accuracy metrics than fixed-volume baselines such as CT2Rep and 3D-CT-GPT, with the largest gains when multiple planes and phases are available. If the architecture is what drives these gains, it would make 3D medical report generation more accurate, cheaper to train, and adaptable to variable scan lengths.

What carries the argument

The Z-former is the load-bearing object: a transformer encoder with Big Bird sparse attention that operates on the concatenated [CLS] embeddings of per-slice 2D DINO ViT features, synthesizing neighboring-slice context while preserving slice-specific detail. It is trained by masked embedding modeling, which predicts masked slice embeddings rather than pixels, and its output serves as keys and values for a perceiver resampler that compresses any slice length into a fixed 32-query visual prompt for the LLM. Together these components implement the paper's radiologist-reading analogy: independent per-slice encoding, sequential synthesis across slices, and flexible integration of multiple views and phases.

What would settle it

Train CT2Rep (or the MS-VLM 3D-encoder variant) with the same 180,000 synthetic VQA pairs and 121,000-136,000 RadGenome-ChestCT pairs used by MS-VLM, and compare on CT-RATE; if the F1 gap shrinks to near zero, the slice-sequence architecture is not the cause. Also, run 3D-CT-GPT trained only on CT-RATE without private pretraining and check whether it still trails MS-VLM.

Watch

Extended reading notes

Core claim

The central discovery is that representing a 3D volume as a sequence of slice-level [CLS] embeddings and modeling cross-slice relations with a sparse-attention transformer yields better radiology-report generation than embedding the whole volume as sub-volumetric 3D patches. MS-VLM's Z-former is pretrained by masked embedding modeling, regressing masked slice features with L1 loss, and then the frozen encoder plus Z-former feed a perceiver resampler whose learned queries are aligned to a frozen LLM (Vicuna-7B) via LoRA fine-tuning. The authors show that this slice-sequence design handles variable slice lengths without performance loss, outperforms its own 3D-ViT variant on CT-RATE, and integrates multi-plane, multi-phase rectal MRI better than a T2-axial-only model. They claim the approach reduces over-correlated z-axis representations that fixed 3D patchification introduces.

Load-bearing premise

The load-bearing premise is that the reported performance gap comes from the slice-sequence architecture itself, not from unequal training conditions: MS-VLM receives extra VQA data and different pretraining than the baselines it is compared against.

Editorial extensions

If this is right

  • On CT-RATE, MS-VLM reports higher ROUGE-L, METEOR, and clinical F1 than CT2Rep and 3D-CT-GPT, including better recall of subtle findings such as satellite masses and fibrotic change.
  • MS-VLM (proposed) outperforms its own 3D-ViT variant (MS-VLM 3D encoder) in NLG and CA metrics, attributing the gain to the slice-sequence plus Z-former design.
  • Using original slice length instead of fixed 240 slices avoids information loss on volumes with L>240, where MS-VLM bests both CT2Rep and its fixed-length variant.
  • On rectal MRI, integrating multi-view and multi-phase volumes with MS-VLM improves T-stage, CRM, and MLNI prediction over T2-axial-only models.
  • Joint report-generation and VQA fine-tuning improves report quality over report-generation-only training, supporting the use of decomposed finding-level VQA data.

Reading between the lines

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

  • Our inference: if the architecture is the cause, the same slice-sequence recipe should transfer to other anisotropic 3D modalities such as PET-CT, ultrasound sweeps, and abdominal CT; a quick check is whether the Z-former variant keeps its edge when the 3D-ViT baseline is given identical VQA data and training epochs.
  • Our inference: because the 3D-CT-GPT numbers come from a model pretrained on private data, a fair head-to-head requires training 3D-CT-GPT on CT-RATE alone; until then the public comparison conflates architecture with pretraining.
  • Our inference: the variable-length property suggests a clinical deployment pattern not tested in the paper: reading a scan as it is acquired, slice by slice, and producing preliminary reports before the full volume is reconstructed.
  • Our inference: the lower hallucination score versus CT2Rep is attributed to CT2Rep's report-specific tokenizer; that claim is testable by swapping tokenizers while holding the rest of the pipeline fixed.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes MS-VLM, a 3D medical vision-language model that processes volumetric images slice-by-slice: a 2D DINO-pretrained ViT encodes each slice, a novel sparse-attention Z-former aggregates the slice-level [CLS] tokens to capture inter-slice dependencies, and a perceiver resampler with an MLP projects the resulting representation to fixed-size learnable queries that are fed into a LoRA-fine-tuned Vicuna-7B. Training is staged (DINO fine-tuning, Z-former masked embedding modeling, bridger alignment, and joint report generation plus VQA instruction fine-tuning). The authors evaluate on the public CT-RATE chest CT dataset and an in-house rectal MRI dataset, reporting gains over CT2Rep and 3D-CT-GPT, over their own 3D-encoder variant, and over a no-VQA variant, and they demonstrate the model's ability to handle variable slice lengths and multi-view/multi-phase input.

Significance. If the headline comparison were controlled, the slice-sequence design would be a practically valuable alternative to fixed 3D patch encoders for 3D medical VLMs: it is initialization-friendly, handles variable z-lengths, and naturally admits multi-plane and multi-phase input. The paper has several strengths: evaluation on the public CT-RATE dataset with an external RadBERT classifier and LLM judges reduces circularity; the CT2Rep baseline was reproduced with its official implementation; per-abnormality results and training-detail documentation are unusually complete; and the variable-length and multi-view experiments address clinically relevant conditions. However, the central architecture claim is currently confounded by unequal pretraining and unequal training supervision across the compared systems, so the significance of the contribution cannot be assessed until controlled ablations are provided.

major comments (4)
  1. [§4.1; Tables 1 and 4] The central claim that the slice-sequence architecture outperforms fixed 3D patch encoders is not isolated by the reported ablation. The MS-VLM (3D encoder) variant is trained only in Stages 2 and 3, whereas MS-VLM uses Stage 0 DINO fine-tuning on CT-RATE slices (§3.2.1, 50 epochs) and Stage 1 Z-former masked-embedding modeling (§3.2.2, 20 epochs) before Stage 2/3. The 3D encoder therefore starts from a weaker initialization and receives less total training signal, so the gains in Tables 1 and 4 could reflect pretraining rather than the slice-sequence design. Please report an ablation with a 3D encoder given analogous self-supervised pretraining, or an MS-VLM variant trained without Stages 0–1, so that architecture and pretraining are not confounded.
  2. [§3.3.1; Table 1] The comparison with CT2Rep is also not matched in supervision. In addition to report generation, MS-VLM is jointly fine-tuned on about 180,000 synthetic VQA pairs and 381,000 selected RadGenome-ChestCT QA pairs (121k type, 136k presence, 110k location, 14k size), while CT2Rep is trained on report generation only. Since the MS-VLM (w/o VQA) row still includes Stages 0–1, no reported row isolates the architecture from the extra data. Please add a baseline trained with the same VQA data, or an MS-VLM without Stages 0–1 and without VQA, to support the claim that MS-VLM 'surpasses existing methods' in Table 1.
  3. [§4.1; Table 1] The 3D-CT-GPT numbers are not from a controlled comparison. As stated in §4.1, the 3D-CT-GPT (T2) model was pretrained on a private chest CT dataset and validated, not trained, on CT-RATE, and its code is unavailable. Quoting its BLEU/ROUGE/METEOR values in the same table as models trained on CT-RATE overstates the comparison. Either reproduce 3D-CT-GPT under the CT-RATE protocol or present the row as an external reference point with an explicit caveat; it should not be used as evidence that MS-VLM 'surpasses existing methods.'
  4. [§5; §1] The motivating mechanism—that fixed 3D patchification creates harmful over-correlated z-axis representations and loses slice-specific detail—is asserted rather than measured. The paper does not quantify z-axis correlation, slice-level information retention, or how the Z-former's sparse attention affects these quantities. Because the architecture comparison is confounded (first major comment), the mechanistic explanation in the Discussion is not yet supported by the experiments. A direct measurement or a targeted experiment (e.g., pathology appearing or disappearing between adjacent slices) would substantially strengthen the claim.
minor comments (5)
  1. [Abstract and §4.1] The statement that MS-VLM 'surpasses existing methods' is too broad given Table 2's hallucination category, where CT2Rep scores higher (0.186 vs 0.127), and Table 1's precision, where CT2Rep is higher (0.355 vs 0.222). Please qualify the claim to reflect these category-level results.
  2. [§3.2.2, Eq. (1)] The notation \hat{Z}_{masked} is confusing because the quantity being regressed is the reconstruction of the masked embeddings, not the masked input itself. Please define it explicitly as the Z-former output at the masked positions.
  3. [§3.3.1] The description of the synthetic VQA data says 'approximately 180,000 QA pairs' and then lists selected RadGenome-ChestCT counts (121k, 136k, 110k, 14k); please clarify whether these subsets are disjoint and how they are combined with the synthetic pairs during training.
  4. [§3.3.2] For the rectal MRI experiments the paper says 20 slices were sampled from each of six selected phases, producing a 120-slice input, but the number and identity of the six phases is not fixed across patients given the availability table in Appendix B; please specify the selection rule for the six phases.
  5. [Table 3] The claim that MS-VLM (z = L) exhibits 'no significant performance degradation' and outperforms fixed-length variants relies on small differences (F1 0.261 vs 0.245 in the full set; 0.246 vs 0.237 in the L > 240 subset) without confidence intervals or significance tests; please report variability or temper the wording.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: claims are empirical comparisons against external benchmarks; the noted ablation confound is a validity issue, not a definitional reduction.

full rationale

This paper is an empirical systems comparison and does not derive its headline results from its own definitions or fitted parameters. The central claims are evaluated against held-out test sets (the CT-RATE test split and an in-house rectal MRI test set) using external and standard metrics: the CT-CLIP/RadBERT text classifier for clinical accuracy, GPT-4o-mini for semantic categories, o1-mini for rectal MRI findings, and BLEU/ROUGE/METEOR for language quality. None of the equations in the paper (the MEM loss in Eq. 1 or the instruction-tuning NLL in Eq. 2) define these evaluation metrics in terms of the model's own outputs, so there is no self-definitional reduction. The self-citations to Lee et al. 2024 and Park et al. 2024 appear in related-work or data-generation motivation and are not load-bearing for the reported superiority claim. The most substantial methodological concern is that the key ablation, MS-VLM (3D encoder), is trained only in stages 2 and 3, omitting the DINO and Z-former pretraining stages, and that MS-VLM receives substantially more VQA training data than CT2Rep; this makes the architecture comparison confounded. However, a confounded experimental comparison is a correctness or validity risk, not circularity: the paper does not fit a parameter and then present that same fitted quantity as a prediction, and it does not invoke a self-citation chain to forbid alternative explanations. The evaluation is externally anchored, so the appropriate circularity finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard transfer-learning assumptions (DINO to medical slices, RadBERT and LLM-based evaluators), plus one method-specific assumption that masked embedding modeling on [CLS] embeddings is sufficient to learn inter-slice dependencies. The trained model weights themselves are the main fitted quantities; the hand-selected hyperparameters listed above are not ablated. No new physical entities are introduced; the Z-former is a neural network module, not a postulated entity in the sense of a new force or particle.

free parameters (4)
  • Z-former MEM masking probability = 0.3
    Chosen by hand for masked embedding modeling; affects Z-former pretraining but not ablated.
  • Big Bird sparse attention configuration = sliding window 16, 3 random blocks, no global blocks
    Hand-selected; no ablation in the paper.
  • Number of learnable queries in perceiver resampler = 32
    Hand-selected; sets the fixed-size visual prompt for the LLM.
  • Joint fine-tuning sample ratio (report:VQA) = 1:1
    Hand-selected; the paper shows VQA helps but does not ablate the ratio.
assumptions (5)
  • domain assumption ImageNet DINO-pretrained 2D ViT features transfer to medical CT and MRI slices after fine-tuning.
    Stage 0 fine-tunes DINO on slices; the premise is that 2D pretraining provides useful slice representations.
  • domain assumption The RadBERT classifier from CT-CLIP is a valid measure of clinical accuracy for generated reports.
    CA metrics rely on this classifier (Section 3.4).
  • domain assumption GPT-4o-mini and o1-mini LLMs provide reliable evaluation of generated reports.
    Used for semantic evaluation (Sections 3.4-3.5, Appendix C).
  • domain assumption CT-RATE preprocessing protocol (1.5 mm z, 0.75 mm xy, 480x480x240) preserves clinically relevant information.
    Adopted from CT-CLIP (Section 3.3.1).
  • ad hoc to paper Masked embedding modeling on [CLS] tokens is a sufficient pretext task to learn inter-slice dependencies.
    MEM is introduced in Section 3.2.2; its sufficiency for report generation is not independently validated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Read Like a Radiologist: Efficient Vision-Language Model for 3D Medical Imaging Interpretation." pith.science (2026). https://pith.science/paper/ZQFA4EX2

@misc{pith2026241213558,
  author       = {Pith},
  title        = {Pith review of: Read Like a Radiologist: Efficient Vision-Language Model for 3D Medical Imaging Interpretation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZQFA4EX2}},
  note         = {Machine review of arXiv:2412.13558}
}
read the original abstract

Recent medical vision-language models (VLMs) have shown promise in 2D medical image interpretation. However extending them to 3D medical imaging has been challenging due to computational complexities and data scarcity. Although a few recent VLMs specified for 3D medical imaging have emerged, all are limited to learning volumetric representation of a 3D medical image as a set of sub-volumetric features. Such process introduces overly correlated representations along the z-axis that neglect slice-specific clinical details, particularly for 3D medical images where adjacent slices have low redundancy. To address this limitation, we introduce MS-VLM that mimic radiologists' workflow in 3D medical image interpretation. Specifically, radiologists analyze 3D medical images by examining individual slices sequentially and synthesizing information across slices and views. Likewise, MS-VLM leverages self-supervised 2D transformer encoders to learn a volumetric representation that capture inter-slice dependencies from a sequence of slice-specific features. Unbound by sub-volumetric patchification, MS-VLM is capable of obtaining useful volumetric representations from 3D medical images with any slice length and from multiple images acquired from different planes and phases. We evaluate MS-VLM on publicly available chest CT dataset CT-RATE and in-house rectal MRI dataset. In both scenarios, MS-VLM surpasses existing methods in radiology report generation, producing more coherent and clinically relevant reports. These findings highlight the potential of MS-VLM to advance 3D medical image interpretation and improve the robustness of medical VLMs.

Figures

Figures reproduced from arXiv: 2412.13558 by the authors.

Figure 1
Figure 1. Overview of MS-VLM architecture. novel Z-former that integrates slice-by-slice 2D ViT embeddings, while the bridger module features a perceiver resampler and an MLP layer to project volumetric representation to a fixed-size learnable query. The learnable query is then used as a visual prompt when instruction fine-tuning the LLM to perform vision-language tasks (report generation or VQA). Given a 3D medical image, ea… view at source ↗
Figure 2
Figure 2. Overview of MS-VLM training and evaluation. a) MS-VLM is evaluated on public chest CT dataset CT-RATE and in-house Rectal MRI dataset. MS-VLM learns volumetric representation off of a single axial view for chest CT and multi-view for rectal MRI. 3.2.1. Training a Domain-Specific DINO Vision Encoder We utilize a 2D ViT-B/16 model pre-trained on ImageNet-1K using DINO (Self-Distillation with No Labels) (Caron et al., … view at source ↗
Figure 3
Figure 3. Instruction prompts used for instruction fine-tuning the LLM for report genera￾tion [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Generating synthetic VQA pairs from CT-RATE reports using LLaMA-3-8B. 3.2.4. Instruction Fine-tuning We employ Vicuna-7B-v1.5 (Chiang et al., 2023) as our LLM, selected for its exemplary instruction-following capabilities among publicly accessible checkpoints (Liu et a…
Figure 5
Figure 5. Figure 5: An example of a rectal MRI report that describes the seven most common rectal tumor-related abnormalities. Each sentence is color-coded based on the abnormality it describes. For the chest CT VQA task, we utilize both synthetic VQA pairs and pairs from the RadGenome-Ch…
Figure 6
Figure 6. Figure 6: Qualitative evaluation on CT-RATE. Each sentence from the report is color coded to identify True Positive, True Negative, False Positive and False Negatives on the findings mentioned in the ground truth report. MS-VLM achieves greater recall on the true positives and i…
Figure 7
Figure 7. Figure 7: Average F1 score across 18 chest abnormalities over training epochs. MS-VLM achieves superior average F1 score compared to baseline methods. A key limitation lies in the relative under-performance of vision encoders compared to LLMs. While LLMs excel in understanding d…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 6 Pith papers

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

  1. Disorder-induced stress-flow misalignment in soft glassy materials revealed using multi-directional shear

    cond-mat.soft 2025-08 unverdicted novelty 6.0 of 10

    Soft glassy materials show a transient stress response orthogonal to a newly applied shear direction, which a mesoscopic elasto-plastic model attributes to local yield-stress disorder.

  2. Computed Tomography Visual Question Answering with Cross-modal Feature Graphing

    cs.CV 2025-07 conditional novelty 5.0 of 10

    A cross-modal graph connecting CT slices and question tokens, aggregated by an attentive GCN, improves LLM-based CT visual question answering on M3D-VQA.

  3. Aligning Proteins and Language: A Foundation Model for Protein Retrieval

    q-bio.BM 2025-05 conditional novelty 5.0 of 10

    A CLIP-style model aligns protein surface point clouds with GO-derived captions and achieves roughly 60% Top-5 zero-shot retrieval on PDB and 36% on EMDB.

  4. CT-Agent: A Multimodal-LLM Agent for 3D CT Radiology Question Answering

    cs.CV 2025-05 conditional novelty 5.0 of 10

    CT-Agent combines an LLM planner, region-specific LoRA adapters, and global/local token compression to improve 3D chest CT report generation and question answering on CT-RATE and RadGenome-ChestCT.

  5. Region-Aware Multimodal Large Language Model via SlowFast Tokenization and Pseudo-Mask Guidance for 3D CT Report Generation

    eess.IV 2025-06 conditional novelty 4.0 of 10

    MedRegion-CT integrates region-representative tokens, mask-driven segmentation tokens, and patient-specific attribute prompts into a multimodal LLM, reporting state-of-the-art scores on RadGenome-Chest CT report generation.

  6. Taming Vision-Language Models for Medical Image Analysis: A Comprehensive Review

    eess.IV 2025-06 conditional novelty 3.0 of 10

    A survey that classifies vision-language model adaptation for medical imaging into five strategies across eleven tasks, with challenges and future directions.

Reference graph

Works this paper leans on

13 extracted references · 3 canonical work pages · cited by 6 Pith papers

  1. [2]

    arXiv preprint arXiv:2406.06512

    Merlin: A vision language foundation model for 3d computed tomography. arXiv preprint arXiv:2406.06512 . Brady, A.P.,

    arXiv

  2. [3]

    arXiv preprint arXiv:2409.19330

    3d-ct-gpt: Generating 3d radiology reports through integration of large vision-language models. arXiv preprint arXiv:2409.19330 . Chiang, W.L., Li, Z., Lin, Z., Sheng, Y., Wu, Z., Zhang, H., Zheng, L., Zhuang, S., Zhuang, Y., Gonzalez, J.E., et al.,

    arXiv

  3. [6]

    Dosovitskiy, A.,

    doi: 10.5626/ JCSE.2012.6.2.168. Dosovitskiy, A.,

    work page 2012

  4. [8]

    arXiv preprint arXiv:2405.03770

    Foundation models for video understanding: A survey. arXiv preprint arXiv:2405.03770 . McRobbie, D.W., et al.,

    arXiv

  5. [12]

    arXiv preprint arXiv:2308.02463

    Towards generalist foundation model for radiology. arXiv preprint arXiv:2308.02463 . Xie, Z., Zhang, Z., Cao, Y., Lin, Y., Bao, J., Yao, Z., Dai, Q., Hu, H.,

    arXiv

  6. [13]

    Advances in neural information processing systems 33, 17283–17297

    Big bird: Trans- formers for longer sequences. Advances in neural information processing systems 33, 17283–17297. Zhang, K., Zhou, R., Adhikarla, E., Yan, Z., Liu, Y., Yu, J., Liu, Z., Chen, X., Davison, B.D., Ren, H., et al., 2024a. A generalist vision–language foundation model for diverse biomedical tasks. Nature Medicine , 1–13. Zhang, X., Wu, C., Zhao...

    arXiv

  7. [317]

    Kalender, W.A.,

    doi: 10.1038/s41597-019-0322-0 . Kalender, W.A.,

    doi:10.1038/s41597-019-0322-0

  8. [2017]

    Medical image analysis 41, 40–54

    3d deeply supervised network for automated segmentation of volumetric medical images. Medical image analysis 41, 40–54. Hamamci, I.E., Er, S., Almas, F., Simsek, A.G., Esirgun, S.N., Dogan, I., Dasdelen, M.F., Wittmann, B., Simsar, E., Simsar, M., et al., 2024a. A foundation model utilizing chest ct volumes and radiology reports for supervised-level zero-...

    arXiv

Show all 13 references
  1. [2019]

    Volumetric at- tention for 3d medical image segmentation and detection, in: Medical Im- age Computing and Computer Assisted Intervention–MICCAI 2019: 22nd International Conference, Shenzhen, China, October 13–17, 2019, Proceed- ings, Part VI 22, Springer. pp. 175–184. Wang, Z....

    2019

  2. [2020]

    arXiv preprint arXiv:2010.11929

    An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 . Dou, Q., Yu, L., Chen, H., Jin, Y., Yang, X., Qin, J., Heng, P.A.,

    2010 arXiv

  3. [2022]

    arXiv preprint arXiv:2210.10163

    Medclip: Contrastive learning from unpaired medical images and text. arXiv preprint arXiv:2210.10163 . 39 Wu, C., Zhang, X., Zhang, Y., Wang, Y., Xie, W.,

    arXiv

  4. [2023]

    URL https://github

    Stanford alpaca: an instruction-following llama model (2023). URL https://github. com/tatsu-lab/stanford alpaca

    2023

  5. [2024]

    arXiv preprint arXiv:2404.00578

    M3d: Advancing 3d medical image analysis with multi-modal large language models. arXiv preprint arXiv:2404.00578 . Blankemeier, L., Cohen, J.P., Kumar, A., Van Veen, D., Gardezi, S.J.S., Paschali, M., Chen, Z., Delbrouck, J.B., Reis, E., Truyts, C., et al.,

    arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.