Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT

• URL: http://arxiv.org/abs/2609.14106v1
• Date: Sat, 12 Sep 2026 19:07:51 GMT
• Title: Parameter-Efficient Fine-Tuning of Foundation Models for Liver Tumor Segmentation in CT
• Abstract summary: We evaluated parameter-efficient fine-tuning (PEFT) of the Segment Anything Model (SAM) for liver tumor segmentation in abdominal CT of colorectal liver metastases.<n>We compared Low-Rank Adaptation (LoRA), 4-bit Quantized LoRA (QLoRA), a convolutional adapter (Conv-Adapter), and our Directional Spectral Top-K adapter (DiSCo)<n>Conv-Adapter and LoRA achieved the highest accuracy (overall Dice 0.793 and 0.792); single-point Dice 0.795 and 0.792; 95th-percentile Hausdorff distance (HD95) 32
• Score: 1.9403818519796712
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We evaluated parameter-efficient fine-tuning (PEFT) of the Segment Anything Model (SAM) for liver tumor segmentation in abdominal CT of colorectal liver metastases. We compared Low-Rank Adaptation (LoRA), 4-bit Quantized LoRA (QLoRA), a convolutional adapter (Conv-Adapter), and our Directional Spectral Top-K adapter (DiSCo), training only adapters while freezing the SAM backbone. DiSCo derives spectral bases from singular value decomposition of row-normalized weights and learns rank-gated spectral coefficients, per-output magnitude offsets, and a spectral gain, with optional Top-K rank selection at inference and 0.14 M trainable parameters. We benchmarked five prompting regimes: no prompt, single-point, multi-point, and bounding boxes at intersection over union 0.50 and 0.75. Conv-Adapter and LoRA achieved the highest accuracy (overall Dice 0.793 and 0.792; single-point Dice 0.795 and 0.792; 95th-percentile Hausdorff distance (HD95) 32 mm). QLoRA was close (overall Dice 0.766; single-point Dice 0.768; HD95 36.41 mm), with 0.91 M trainable parameters, 120 ms latency, and 4.9 GB peak memory. DiSCo achieved the highest Dice per million trainable parameters (4.66), with overall Dice 0.653, single-point Dice 0.698, and HD95 49.53 mm. These results show an accuracy-efficiency trade-off and support PEFT for liver tumor segmentation with reduced adaptation costs when compute and labeled data are limited. Code: https://github.com/Ramtin-Mojtahedi/PEFT-SAM-Liver-CT

Related papers

• Spectral Adapters for Segment Anything Model-based Segmentation of Colorectal Liver Metastases in Computed Tomography [1.6111769386474124]
  We propose two parameter-efficient spectral adapters for the Segment Anything Model (SAM)<n>We evaluate these methods on 446 contrast-enhanced CT volumes (355 training, 91 testing)<n>SiGA achieves the best single-point performance with a Dice score of 0.77, IoU of 0.69, and HD95 of 35.39 mm.
  arXiv  Detail & Related papers  (2026-09-10T15:26:15Z)
• LeCor: Learning to Be Corrected by Meta-Learned Test-Time Training for Interactive 3D Lung-Tumour Segmentation [45.86925813888675]
  Fine-tuning SAM 3 on lung tumours raises the Dice obtained from a single point prompt from 0.298 to 0.757, and seven rounds of corrections raise it further to 0.765.<n>We propose LeCor, which performs test-time training on a small set of case adapters that are reset for every case.<n>LeCor raises the Dice reached after seven correction rounds from 0.787 with the fine-tuned model to 0.827, reduces the number of cases that never reach a Dice of 0.80 from 47 to 27, and reaches in three correction rounds the accuracy that the fine-tuned model attains in seven.
  arXiv  Detail & Related papers  (2026-09-08T21:49:56Z)
• Parameter-Efficient pretrained-CT-to-MRI Transfer for Rectal Cancer Segmentation: Performance-Calibration Trade-offs [3.0748861313822995]
  We introduce SWIFT, a pretrained model wIth parameter-eF and Tumor-aware fine-tuning for rectal cancer segmentation.<n>A Swin V2 encoder pretrained on 10,444 public 3D CT volumes was adapted to T2-weighted MRI through four cumulative configurations.<n> Geometric accuracy, tumor detection, radiomic agreement, and probability calibration were evaluated.
  arXiv  Detail & Related papers  (2026-08-27T14:27:49Z)
• Post-Operative Glioma Segmentation via Loss Stabilization, Normalization and Subspace Attention [46.03321798937855]
  We show that the standard Generalized Dice Loss (GDL) is unstable under domain shift.<n>We propose a Subspace-Aware Class Attention (SACA) module that re-calibrates the bottleneck features and raises Enhancing Tumor (ET) sensitivity by 8%.
  arXiv  Detail & Related papers  (2026-07-23T09:59:30Z)
• Weight-Space Geometry of Offline Reasoning Training [39.20113202952282]
  Training six methods (SFT, RFT, DFT, RIFT, Offline GRPO, DPO) on identical math rollouts from a single base model (Qwen3-4B) with attention-only LoRA, we analyze the resulting deltas via cosine similarity, principal-angle subspace analysis, linear mode connectivity, and CKA.<n>We observe: (i) SFT, RFT, and RIFT have nearly colinear weight deltas (cosine >= 0.97, top-1 principal angle 7 deg median over 144 modules) and comparable GSM8K accuracy (87-88%, n=1
  arXiv  Detail & Related papers  (2026-06-21T15:34:21Z)
• Improving Generalization of Deep Learning for Brain Metastases Segmentation Across Institutions [5.4585656405659195]
  The goal of this work is to create a domain adaptation framework that will allow for BM segmentation to be used across multiple institutions.<n>We propose a VAE-MMD preprocessing pipeline that combines variational autoencoders (VAE) with maximum mean discrepancy (MMD) loss.
  arXiv  Detail & Related papers  (2026-04-01T02:25:25Z)
• Coronary Artery Segmentation and Vessel-Type Classification in X-Ray Angiography [0.0]
  Low contrast, motion, foreshortening, overlap, and catheter confounding degrade vessel segmentation.<n>Vessel-type labeling achieves 98.5% accuracy (Dice 0.844) for RCA, 95.4% (0.786) for LAD, and 96.2% (0.794) for LCX.
  arXiv  Detail & Related papers  (2026-01-24T11:56:49Z)
• Fixed-Budget Parameter-Efficient Training with Frozen Encoders Improves Multimodal Chest X-Ray Classification [0.0]
  Multimodal chest X-Ray analysis often fine-tunes large vision-language models, which is computationally costly.<n>We study parameter-efficient training strategies, including frozen encoders, BitFit, LoRA, and adapters for multi-label classification on the Indiana University Chest X-Ray dataset.
  arXiv  Detail & Related papers  (2025-12-25T05:02:19Z)
• Pillar-0: A New Frontier for Radiology Foundation Models [41.640120966890954]
  We introduce Pillar-0, a radiology foundation model pretrained on 42,990 abdomen-pelvis CTs, 86,411 chest CTs, 14,348 head CTs, and 11,543 breast MRIs.<n>Pillar-0 achieves mean AUROCs of 86.4, 88.0, 90.1, and 82.9, outperforming MedGemma (Google), MedImageInsight (Microsoft), Lingshu (Alibaba), and Merlin (Stanford) by 7.8-15.8 AUROC points and ranking best in 87.2% (319/366) tasks.
  arXiv  Detail & Related papers  (2025-11-21T21:50:34Z)
• PulmoFusion: Advancing Pulmonary Health with Efficient Multi-Modal Fusion [31.28925796862013]
  Traditional remote spirometry lacks the precision required for effective pulmonary monitoring.<n>We present a novel, non-invasive approach using multimodal predictive models that integrate RGB or thermal video data with patient metadata.
  arXiv  Detail & Related papers  (2025-01-29T15:10:09Z)
• MRSegmentator: Multi-Modality Segmentation of 40 Classes in MRI and CT [29.48170108608303]
  The model was trained on 1,200 manually annotated 3D axial MRI scans from the UK Biobank, 221 in-house MRI scans, and 1228 CT scans. It demonstrated high accuracy for well-defined organs (lungs: DSC 0.96, heart: DSC 0.94) and organs with anatomic variability (liver: DSC 0.96, kidneys: DSC 0.95) It generalized well to CT, achieving DSC mean of 0.84 $pm$ 0.11 on AMOS CT data.
  arXiv  Detail & Related papers  (2024-05-10T13:15:42Z)
• Hybrid Window Attention Based Transformer Architecture for Brain Tumor Segmentation [28.650980942429726]
  We propose a volumetric vision transformer that follows two windowing strategies in attention for extracting fine features. We trained and evaluated network architecture on the FeTS Challenge 2022 dataset. Our performance on the online validation dataset is as follows: Dice Similarity Score of 81.71%, 91.38% and 85.40%.
  arXiv  Detail & Related papers  (2022-09-16T03:55:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.