{
  "schema_version": 1,
  "record_url": "https://ramtin-mojtahedi.github.io/publications/foundation-models-liver-tumor-segmentation-2026/",
  "last_verified": "2026-08-02",
  "profile": {
    "name": "Ramtin Mojtahedi",
    "orcid": "https://orcid.org/0000-0002-3953-3256",
    "google_scholar": "https://scholar.google.com/citations?user=KjUrlGUAAAAJ&hl=en",
    "website": "https://ramtin-mojtahedi.github.io/"
  },
  "publication": {
    "id": "foundation-models-liver-tumor-segmentation-2026",
    "title": "Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT",
    "year": 2026,
    "publication_date": "2026-04-08",
    "group": "2026",
    "status": "Published",
    "authors": [
      "Ramtin Mojtahedi",
      "Mohammad Hamghalam",
      "Jacob J. Peoples",
      "Richard K. G. Do",
      "Amber L. Simpson"
    ],
    "authors_html": "<strong>R. Mojtahedi</strong>, M. Hamghalam, J. J. Peoples, R. K. G. Do, and A. L. Simpson",
    "citation": "Medical Imaging 2026: Computer-Aided Diagnosis, Proc. SPIE 13926, 1392612.",
    "citation_plain": "Medical Imaging 2026: Computer-Aided Diagnosis, Proceedings of SPIE 13926, 1392612.",
    "doi": "10.1117/12.3087835",
    "url": "https://doi.org/10.1117/12.3087835",
    "link_label": "DOI ↗",
    "manual": true,
    "source": "Publisher and DOI metadata; OpenAlex/ORCID; Crossref",
    "last_verified": "2026-08-02",
    "title_aliases": [],
    "venue": "Medical Imaging 2026: Computer-Aided Diagnosis",
    "volume": "13926",
    "issue": "",
    "pages": "260-268",
    "article_number": "1392612",
    "type": "conference-paper",
    "peer_reviewed": true,
    "publisher": "SPIE",
    "keywords": [
      "foundation models",
      "parameter-efficient fine-tuning",
      "liver tumor segmentation",
      "computed tomography"
    ],
    "note": "",
    "detail_url": "/publications/foundation-models-liver-tumor-segmentation-2026/",
    "open_access_url": "",
    "open_access_label": "",
    "open_access_license": "",
    "preprint_url": "",
    "code_url": "https://github.com/Ramtin-Mojtahedi/peft-sam-liver-ct",
    "code_label": "Code",
    "pubmed_url": "",
    "pmc_url": "",
    "citation_pdf_url": "",
    "abstract": "State-of-the-art interactive segmentation with foundation models remains constrained by compute and the need for annotations. We evaluated parameter-efficient fine-tuning (PEFT) of the Segment Anything Model (SAM) by inserting baseline adapters, Low-Rank Adaptation (LoRA), 4-bit Quantized Low-Rank Adaptation (QLoRA), and a convolutional adapter (Conv-Adapter), as well as our proposed Directional Spectral Top-K adapter (DiSCo), training only the adapters while keeping the SAM backbone frozen. DiSCo performs singular value decomposition of row-normalized weights to obtain spectral bases, then learns rank-gated spectral coefficients, per-output magnitude offsets, and a spectral gain; it supports optional Top-K rank selection at inference while keeping only 0.14 M parameters trainable. We benchmarked five prompting regimes, none, single-point, multi-point, and bounding boxes at an intersection over union (IoU) of 0.50 and 0.75, applied to abdominal CT scans of colorectal liver metastases, and reported segmentation metrics (Dice and the 95th-percentile Hausdorff distance (HD95)) and compute metrics (trainable parameters, latency, memory, throughput). Conv-Adapter and LoRA achieved the highest accuracy (overall Dice 0.793 and 0.792; single-point 0.795 and 0.792; HD95 32 mm). QLoRA was close (overall 0.766; single-point 0.768; HD95 36.41 mm) while offering the most favorable compute profile (0.91 M trainable parameters, 120 ms latency, 4.9 GB peak memory). DiSCo maximized parameter efficiency, achieving the highest Dice per million trainable parameters (4.66), showing an accuracy-efficiency trade-off (overall Dice 0.653; single-point 0.698; HD95 49.53 mm). These results indicate that PEFT on foundation models enables more accurate liver tumor segmentation with reduced adaptation costs, supporting rapid scanner-specific tuning, point- or box-prompt workflows, and broader deployability for preoperative volumetrics when compute and labeled data are limited.",
    "bibtex_type": "inproceedings",
    "bibtex_key": "Mojtahedi2026PEFTFoundation"
  },
  "citation": {
    "plain_text": "Ramtin Mojtahedi, Mohammad Hamghalam, Jacob J. Peoples, Richard K. G. Do, and Amber L. Simpson. “Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT.” Medical Imaging 2026: Computer-Aided Diagnosis, Proceedings of SPIE 13926, 1392612. https://doi.org/10.1117/12.3087835",
    "bibtex": "@inproceedings{Mojtahedi2026PEFTFoundation,\n  author = {Ramtin Mojtahedi and Mohammad Hamghalam and Jacob J. Peoples and Richard K. G. Do and Amber L. Simpson},\n  title = {{Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT}},\n  year = {2026},\n  booktitle = {Medical Imaging 2026: Computer-Aided Diagnosis},\n  volume = {13926},\n  pages = {260-268},\n  publisher = {SPIE},\n  doi = {10.1117/12.3087835},\n  url = {https://doi.org/10.1117/12.3087835}\n}",
    "ris": "TY  - CPAPER\nTI  - Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT\nAU  - Ramtin Mojtahedi\nAU  - Mohammad Hamghalam\nAU  - Jacob J. Peoples\nAU  - Richard K. G. Do\nAU  - Amber L. Simpson\nPY  - 2026\nDA  - 2026-04-08\nT2  - Medical Imaging 2026: Computer-Aided Diagnosis\nVL  - 13926\nSP  - 260\nEP  - 268\nDO  - 10.1117/12.3087835\nUR  - https://doi.org/10.1117/12.3087835\nKW  - foundation models\nKW  - parameter-efficient fine-tuning\nKW  - liver tumor segmentation\nKW  - computed tomography\nER  -\n"
  },
  "related_records": [
    {
      "title": "Spectral Adapters for Segment Anything Model-based Segmentation of Colorectal Liver Metastases in Computed Tomography",
      "year": 2026,
      "url": "https://ramtin-mojtahedi.github.io/publications/spectral-adapters-sam-crlm-2026/"
    },
    {
      "title": "Parameter-Efficient Fine-Tuning and Few-Shot Learning of Multiscale Vision Transformers for Liver Tumour Segmentation in CT",
      "year": 2025,
      "url": "https://ramtin-mojtahedi.github.io/publications/peft-few-shot-multiscale-vit-2025/"
    },
    {
      "title": "Radiomic Signatures from Baseline CT Predict Chemotherapy Response in Unresectable Colorectal Liver Metastases",
      "year": 2026,
      "url": "https://ramtin-mojtahedi.github.io/publications/radiomic-signatures-chemotherapy-response-2026/"
    }
  ]
}