OpenPros : A Large-Scale Dataset for Limited View Prostate Ultrasound Computed Tomography

Hanchen Wang, Yixuan Wu, Yinan Feng, Peng Jin, Luoyuan Zhang, Shihang Feng, James Wiskin, Baris Turkbey, Peter A. Pinto, Bradford J. Wood, Songting Luo, Yinpeng Chen, Emad Boctor, Youzuo Lin · 2025

Read OpenPros : A Large-Scale Dataset for Limited View Prostate Ultrasound Computed Tomography by Hanchen Wang, Yixuan Wu, Yinan Feng, Peng Jin, Luoyuan Zhang, Shihang Feng, James Wiskin, Baris Turkbey, Peter A. Pinto, Bradford J. Wood, Songting Luo, Yinpeng Chen, Emad Boctor, Youzuo Lin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-12261.txt.

Introduction

Hanchen Wang and a large team present OpenPros in two thousand twenty-five. OpenPros is a data set for prostate ultrasound computed tomography. The method rebuilds tissue traits like speed of sound from waves. Few angle views and bone warps make prostate scans hard.

Models come from clinical scans and dozens of real specimens. Hundreds of thousands of paired maps and wave records fill the set. Wave runs use set clinical layouts and open solvers. Learning methods beat physics routes on speed and accuracy in tests. Broad skill and fine detail still lag, the team notes. The set is public.

Catalog record

CREATORHanchen Wang, Yixuan Wu, Yinan Feng, Peng Jin, Luoyuan Zhang, Shihang Feng, James Wiskin, Baris Turkbey, Peter A. Pinto, Bradford J. Wood, Songting Luo, Yinpeng Chen, Emad Boctor, Youzuo Lin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-12261

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

OpenPros : A Large-Scale Dataset for Limited View Prostate Ultrasound Computed Tomography Hanchen Wang, Yixuan Wu, Yinan Feng, Peng Jin, Luoyuan Zhang, Shihang Feng, James Wiskin, Baris Turkbey, Peter A. Pinto, Bradford J. Wood, Songting Luo, Yinpeng Chen, Emad Boctor, Youzuo Lin Abstract Prostate cancer is one of the most common and lethal cancers among men, making its early detection critically important. Ultrasound computed tomography (USCT) has emerged as an accessible and cost-effective method that reconstructs quantitative tissue parameters, which can serve as potential biomarkers for malignancy. However, current prostate USCT faces considerable barriers: limited-angle acquisitions due to anatomical constraints, tissue heterogeneity, proximity to organs and bony pelvic structures, and lengthy processing times. The lack of large-scale, anatomically precise datasets significantly hampers the development of high-quality, efficient, and generalizable methods. To address this gap, we introduce OpenPros , the first large-scale benchmark dataset for limited-angle prostate USCT, designed to evaluate machine learning algorithms for inverse problems systematically. Our dataset includes over 280,000 paired samples of realistic 2D speed-of-sound (SOS) phantoms and corresponding ultrasound full-waveform data, generated from anatomically accurate 3D digital prostate models derived from 4 real clinical MRI/CT scans and 62 ex vivo prostate specimens with experimental ultrasound measurements, annotated by medical experts. Simulations are conducted under clinically realistic configurations using advanced finite-difference time-domain (FDTD) and Runge-Kutta acoustic wave solvers, both provided as open-source components. Through comprehensive benchmarking, we find that deep learning methods significantly outperform traditional physics-based algorithms in inference efficiency and reconstruction accuracy. However, our results also reveal that current machine learning methods fail to deliver clinically acceptable, high-resolution reconstructions, underscoring critical gaps in generalization, robustness, and uncertainty quantification. By publicly releasing OpenPros , we provide the community with a rigorous benchmark that not only enables fair method comparison but also motivates new advances in physics-informed learning, foundation models for scientific imaging, and uncertainty-aware reconstruction—bridging the gap between academic ML research and real-world clinical deployment. The dataset is publicly accessible at https://open-pros.github.io/ . Figure 1: OpenPros dataset creation and benchmarking pipeline. Top panel: Starting from clinical MRI and CT scans, we employ expert annotations to generate detailed 3D anatomical segmentations. We then incorporate real ultrasound speed-of-sound (SOS) measurements from ex vivo prostate samples acquired using the QTscan platform. These are integrated into comprehensive 3D abdominal SOS models. Clinically relevant 2D slices are extracted from these models to simulate limited-angle ultrasound tomography scenarios. Bottom panel: The extracted 2D SOS maps form the ground truth for ultrasound simulations governed by the acoustic wave equation. The resulting simulated ultrasound data are organized into the OpenPros dataset. We utilize these data to train and benchmark physics-based and deep-learning inversion methods, facilitating the evaluation and development of rapid, clinically relevant SOS reconstruction methods under challenging limited-angle conditions. 1 Introduction Prostate cancer is the second most common malignancy in men and is one of the leading causes of cancer-related deaths worldwide. One in eight men suffers from it ( Radtke and Hadaschik, 2020 ; Tosoian et al., 2024 ) . Since the 5-year survival rate for prostate cancer patients significantly drops from nearly 100% to approximately 34% once the disease progresses from localized or regional stages to distant metastases ( Institute, 2024 ) , early detection of aggressive prostate cancer is of vital importance. Medical imaging plays an essential role in this early detection. Among the available imaging modalities, multiparametric MRI (mpMRI) is currently recognized as the most advanced and accurate imaging tool for detecting and localizing clinically significant prostate cancer. However, the high cost and limited accessibility of mpMRI restrict its widespread adoption, particularly in rural or low-resource settings ( De Rooij et al., 2014 ; Kasivisvanathan et al., 2018 ) . In contrast, ultrasound imaging is widely accessible, cost-effective, and capable of real-time imaging. Prostate ultrasound is typically performed transrectally, producing B-mode (brightness-mode) images. Although transrectal ultrasound (TRUS) is the clinical standard for routine prostate evaluations and biopsy guidance, it has a sensitivity of only 30%–50% for detecting clinically significant tumors and a specificity of 70%–80% ( Beemsterboer et al., 1999 ; Chen et al., 2016 ) . Studies have further shown that tumors located in the anterior or apical prostate regions are often undetectable with TRUS due to poor soft-tissue contrast and restricted acoustic windows, and TRUS cannot reliably distinguish malignant lesions from benign conditions such as chronic prostatitis ( Maričić et al., 2010 ) . Ultrasound computed tomography (USCT) has emerged as a promising alternative, reconstructing quantitative tissue parameters like speed-of-sound (SOS) and acoustic attenuation that serve as potential biomarkers for malignancy ( Wu, 2024 ; Williams et al., 2021 ) . However, the anatomical constraints of prostate imaging inherently limit the acquisition aperture, creating a challenging limited-angle condition . Unlike idealized setups where transducers surround the entire imaging domain, prostate imaging is anatomically restricted to transrectal and transabdominal placements, resulting in sparse and angularly limited data. Traditional physics-based methods typically struggle under these conditions, with slow convergence, severe ill-posedness, and significant reconstruction artifacts ( Wang et al., 2025 ; Gilboy et al., 2020 ) . Developing robust USCT algorithms capable of accurately handling limited-angle data is thus critically needed for clinical prostate imaging. Furthermore, clinical translation of prostate USCT faces considerable barriers due to the complexity and specialization of current imaging systems. To date, only two USCT systems (SoftVue and QTscan) have received U.S. FDA approval, and both systems focus exclusively on full-angle breast imaging with custom hardware setups unsuited for prostate applications ( Sandhu et al., 2015 ; Malik et al., 2018 ) . These existing systems operate at relatively low frequencies, rely on patient positioning incompatible with prostate imaging, and require hours for reconstruction. Thus, there is an urgent need for efficient, generalizable, and clinically adaptable prostate-specific USCT platforms. Crucially, this advancement depends on the availability of realistic, anatomically precise digital phantoms and datasets, which are currently lacking in the field ( Gilboy et al., 2020 ; Aalamifar et al., 2017 ) . Additionally, prostate imaging complexity is increased by high tissue heterogeneity and proximity to multiple adjacent organs and bony pelvic structures, invalidating simplified fluid medium assumptions typically used in breast imaging. These factors severely compromise USCT image reconstruction quality, further emphasizing the necessity of specialized prostate-specific datasets. Recent advances in deep learning, particularly convolutional neural networks (CNNs), have shown potential for overcoming these limitations by learning complex mappings directly from ultrasound data to high-resolution SOS maps ( Chugh et al., 2021 ; Havaei et al., 2017 ) . Data-driven approaches bypass computational bottlenecks encountered by iterative solvers and demonstrate the ability to reconstruct detailed tissue properties even under sparse and noisy acquisition conditions. Instead of hours to days of image reconstruction using physics-based methods and the requirement of expert reading as a follow-up, the relatively short inference time and the automatic analysis enable faster and easier diagnosis for better patient experience. Furthermore, transformer-based architectures have recently demonstrated remarkable performance in medical imaging by effectively modeling long-range spatial dependencies, a feature particularly beneficial for ultrasound tomography due to the extensive spatial interaction of acoustic waves. Despite these advancements, progress has been significantly hampered by the lack of large-scale, high-fidelity datasets supporting the development, evaluation, and reproducibility of innovative reconstruction algorithms. Existing USCT datasets are typically

Read the complete text