MedElevator Войти в исследования
← Поиск публикаций

DOI 10.1016/j.media.2026.104333

Медицинская библиотека · Medical image analysis · 2026

Micro-ultrasound super-resolution with geometry-driven consistency models.

Pang Kaifeng, Zhao Kai, Miao Qi, Hung Alex Ling Yu, Oh Changsuk, Ramirez Raymi, He Qiudi, Klein Jordan, Shao Wei, Brisbane Wayne, Sung Kyunghyun

СтатьяПеревод в очереди

Аннотация

Prostate cancer diagnosis typically relies on image-guided biopsy. Micro-ultrasound (MicroUS) has recently emerged as a promising imaging modality, offering high spatial resolution with comparable diagnostic performance to multi-parametric MRI (mpMRI) at a lower cost. However, unlike MRI, MicroUS volumes are acquired in a fan-shaped, angular geometry, making direct alignment and comparison across imaging modalities challenging. High-quality multi-planar reformation (MPR) is therefore expected in treatment planning, yet conventional MPR produces blurred images with slice discontinuities. In this paper, we present UltraCCM, a fully self-supervised, geometry-driven conditional consistency model for super-resolution of reformatted MicroUS images. UltraCCM explicitly incorporates the geometric characteristics of MicroUS acquisition and formulates MPR as an angular super-resolution problem, enabling fast, single-step recovery of fine anatomical details without requiring high-resolution target-plane supervision. Extensive experiments on in vivo and ex vivo datasets, including quantitative evaluation, expert reader studies, and a downstream prostate segmentation task, demonstrate that UltraCCM improves perceptual quality and fine-detail visualization while preserving anatomy-relevant information for downstream image analysis. The code of UltraCCM is available at: https://github.com/Calvin-Pang/UltraCCM.

Для цитирования: Pang Kaifeng, Zhao Kai, Miao Qi, Hung Alex Ling Yu, Oh Changsuk, Ramirez Raymi, He Qiudi, Klein Jordan, Shao Wei, Brisbane Wayne, Sung Kyunghyun. Micro-ultrasound super-resolution with geometry-driven consistency models.. Medical image analysis. 2026. DOI: 10.1016/j.media.2026.104333.
Подготовка и перевод публикации
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…

История · было → стало

Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.

Тип публикации указан по данным источника. Уровень доказательности автоматически не присваивается.