Medical Image Segmentation Techniques

RnR-ExM: Robust Non-rigid Registration Challenge for Expansion Microscopy Volumes

The paper presents a challenge for 3D registration of expansion microscopy volumes, evaluates top methods, and highlights future directions for ExM registration.

A critical assessment of LGN segmentation quality based on volume, contour, and location

This study benchmarks seven automated methods and manual segmentation for LGN delineation, finding HIPS-THOMAS and DeepThalamus perform best, while manual segmentation remains precise but variable.

The SWAN atlas: an open multimodal resource combining neuroimaging and high-resolution histology of the human brain in stereotactic space

The paper presents the SWAN atlas, a multimodal resource combining neuroimaging and high-resolution histology of the human brain in stereotactic space to improve localization accuracy for clinical applications.

A public-data reproducibility benchmark of a 2.5-D U-Net with a slice-level topology term for type-B aortic dissection segmentation

The paper benchmarks a 2.5-D U-Net with a slice-level topology term for type-B aortic dissection segmentation using public CTA data, finding no consistent benefit from the topology term.

LuTuFlow: a deep learning-based pipeline for quantitative characterization of mouse lung tumors in CT imaging

LuTuFlow automates mouse lung tumor analysis in CT scans by integrating deep learning for segmentation and tracking, providing quantitative results and visualizations.

A robust and scalable image analysis framework for phenotyping human thoracic aorta

This paper introduces a framework for automated thoracic aorta phenotyping from CT scans, generating anatomical landmarks, meshes, and biomechanical phenotypes.

A large, comprehensive labeled MRI dataset of the human lateral ventricle choroid plexus

The paper introduces DataPlex, a large labeled MRI dataset of the human lateral ventricle choroid plexus to improve segmentation accuracy for clinical and research applications.

UniMedSeg: Unified Multi-Modal Medical Image Segmentation via Uncertainty Guided Encoding and Frequency-Spatial Dual-Prompt Decoding

UniMedSeg improves multi-modal medical image segmentation by integrating uncertainty-guided encoding and dual-prompt decoding for better domain generalization and spatial risk estimation.

Physics and Morphology Constrained Quantitative Susceptibility Based Segmentation of Cerebral Veins

Dipole-field physics and Frangi vesselness losses lift QSM vein segmentation to centreline Dice 0.70 (U-Net 0.57); derived venous OEF tracks calibrated fMRI better than vessel filtering.

Comparison of Automated White Matter Lesion Segmentation Approaches for Use in Large, Multi-Site Data Analyses in Parkinson's Disease

This paper evaluates automated white matter lesion segmentation methods in Parkinson's disease and healthy controls, finding U-Net-pgs to be the most accurate and reliable approach across multiple metrics and datasets.

RecallRefine: Coverage-Constrained Coarse-to-Fine Segmentationfor Small 3D Lesions

RECALLREFINE improves 3D lesion segmentation by addressing coarse-stage misses through coverage-constrained candidate selection, increasing sensitivity and Dice score for small lesions.

High-resolution postmortem MRI of the human infant brain: cortical reconstruction and anatomical parcellation

This paper introduces a computational framework for automated anatomical segmentation and cortical reconstruction in high-resolution postmortem infant MRI, showing improved performance across diverse datasets.

ErrorRoute: Budgeted Utility-Aware Resolution Routingfor 3D Medical Image Segmentation

ERRORROUTE improves 3D medical image segmentation by selectively allocating computation to regions where additional processing provides the most utility, achieving higher accuracy with lower computational cost.

A Hybrid Residual-Swin Transformer Design with Attention for Prostate Cancer Segmentation

The paper introduces RSAUNet, a deep learning model for prostate cancer segmentation in MRI scans, achieving high Dice and Jaccard scores.

Automatic Quality Control and Error Correction in MRI linear registration via a Residual Parameter Prediction Network for T1w MRI

A 3D ResNet18 trained on synthetically misaligned T1w scans predicts residual affine parameters, correcting synthetic misalignments to 0.778 mm residual RMSE and flagging failed registrations better than RegQCNET or DARQ.

A precise atlas of the human subcortex

FOCUS, a 100 µm atlas of 180 manually segmented subcortical grey and white matter regions and 77 hand-built tracts, registered to MNI 2009b space for warping to individual brains.

Anatomy-aware, label-informed approach improves image registration for challenging datasets

This paper introduces a label-informed image registration method that improves registration accuracy for challenging datasets, particularly in developmental studies.

DRUMS - A Flexible New Deep Learning Tool for Precise Cortical Surface Alignment across Individuals

DRUMS improves cortical surface alignment by using deep learning with multimodal supervision, outperforming existing methods in accuracy and reproducibility.

Charting neuroimaging-based head circumference and cranial shape across development

CranioTrace automates head circumference and cranial shape measurement from neuroimaging, enabling population growth modeling and identifying disease-related growth trajectory shifts.

ACPC-MAP: A protocol for manually aligning structural T1-weighted Magnetic Resonance Images to the Anterior Commissure - Posterior Commissure plane according to clinical standards.

The paper introduces a standardized protocol for manually aligning structural MR images to the clinically used Schaltenbrand AC-PC plane to improve anatomical consistency and validation of automated methods.

A Generalizable Deep Learning Model for Automated 3D Segmentation of Orthopteran Head Anatomy in Micro-CT

This paper develops a deep learning model for automated 3D segmentation of orthopteran head anatomy from micro-CT images, showing comparable performance between sequential fine-tuning and from-scratch training strategies.

MAE-UNETR++: Masked Autoencoder Pretraining for 3-D Lung Nodule Segmentation

This paper proposes MAE-UNETR++ for 3-D lung nodule segmentation, showing that MAE pretraining outperforms random initialization and Decathlon transfer learning in low-data scenarios.

BRAIN CAST: An MRIQC-guided pipeline for age- and sex-specific pediatric brain MRI template construction, validated by downstream structural fidelity

Morphological bias of the MNI152 brain

The paper evaluates the MNI152 brain template's fidelity to real human morphology and finds it systematically mismatches population anatomy, suggesting the need for more accurate templates.

Leveraging Segmentation Variability to Improve Brain Age Prediction

This study improves brain age prediction by integrating features from multiple segmentation pipelines, showing that variability across pipelines reflects structured demographic differences.

Deep-learning based 3D segmentation of heterogeneous lizard claw tissue from CT data

This paper develops a 2.5D deep learning framework for segmenting heterogeneous lizard claw tissues from CT data, achieving high accuracy with an Inception-ResNet-v2 model.

A surface-intrinsic framework for topology-preserving hippocampal alignment and precision morphometry

This paper introduces a surface-intrinsic framework for hippocampal unfolding that improves alignment accuracy and morphology preservation compared to volumetric methods.

AI segmentation requires accounting for brain size to maintain performance on developmental MRI cohorts

The paper evaluates SynthSeg's performance on developmental MRI scans and finds that rescaling to adult brain sizes improves segmentation accuracy in infants and children.

A hierarchical framework for cortical and subcortical gray-matter parcellation across rodents, primates, and humans

The paper develops a hierarchical atlas for gray-matter parcellation across multiple species, validated through cross-species comparisons and connectivity analysis.

Deep Learning Based Cross-Modality Histological Brain Section Registration in Multiple Species Using Synthetic Images

The paper develops a deep learning framework for cross-modality histological brain section registration across species using synthetic images, achieving accuracy comparable to alternative methods.