Does Vendor Image Processing Matter? Learning Echocardiography Display Appearance from Native Beamspace Data

Elias Stenhede; Arian Ranjbar

eess.IVOct 30, 2026

Does Vendor Image Processing Matter? Learning Echocardiography Display Appearance from Native Beamspace Data

Authors: Elias Stenhede, Arian Ranjbar

Organizations: Medical Technology & E-Health, Akershus University Hospital, Norway

Abstract

Most echocardiography datasets contain vendor-processed images, but processing is often undocumented and may affect model performance and robustness. To address this, we train a small convolutional neural network to reproduce General Electric display appearance from native beamspace brightness-mode ultrasound, using 8 406 recordings from 605 unique patients. The beamspace CNN reduces test mean absolute error by 43% over the optimal pointwise brightness map, outperforms an equivalent Cartesian CNN and runs in real time on a CPU. Across the EchoXFlow and CAMUS datasets, unfiltered training followed by filtered evaluation yields the lowest mean Dice for every class, suggesting filtering may be useful as augmentation when evaluation data are filtered or their processing is unknown. Code and pretrained weights: https://github.com/Ahus-AIM/EchoAppearance

Figures & tables

Figure 1: Left: appearance learning against the vendor display. Right: the same frame without and with filtering.

Method	MAE $\downarrow$				PSNR ( $\mathrm{dB}$ ) $\uparrow$				SSIM $\uparrow$
Scan conversion (input)		8	[33.6,	34.1]		1	[17.1,	17.2]		512	[0.498,	0.525]
Optimal pointwise map		4	[5.2,	5.6]		5	[29.3,	29.8]		892	[0.887,	0.896]
Cartesian CNN		4	[3.2,	3.6]		2	[33.9,	34.5]		939	[0.934,	0.943]
Beamspace CNN	3 .	1	[2.9 ,	3.3]	35 .	2	[34.8 ,	35.4]	0 .	945	[0.939 ,	0.949]

Table 1: Appearance fidelity on the held-out test set: recording means [

95\text{\,}\mathrm{\%}

patient-bootstrap interval].

Figure 2: Five test recordings from different patients, selected at fixed CNN-error quantiles. Rows: input and beamspace CNN.

Figure 3: The optimal pointwise map (median target per input level), fitted on the training recordings.

	EchoXFlow		CAMUS
Filtering applied	LV cavity	Myocardium	LV cavity	Myocardium	Left atrium
None	93.5 $\pm$ 0.3$$	76.0 $\pm$ 0.8$$	93.8 $\pm$ 0.6$$	88.1 $\pm$ 1.2$$	92.2 $\pm$ 0.8$$
Training and evaluation	93.4 $\pm$ 0.3$$	75.7 $\pm$ 0.8$$	93.9 $\pm$ 0.6$$	88.3 $\pm$ 1.1$$	92.4 $\pm$ 0.9$$
Evaluation only	89.2 $\pm$ 0.6$$	64.5 $\pm$ 1.1$$	93.2 $\pm$ 1.0$$	87.3 $\pm$ 1.5$$	91.3 $\pm$ 1.0$$
Training only	92.4 $\pm$ 0.4$$	73.7 $\pm$ 0.8$$	93.6 $\pm$ 0.6$$	87.9 $\pm$ 1.1$$	92.4 $\pm$ 0.9$$

Table 2: Dice (

\mathrm{\%}

)

\uparrow

: mean

\pm

conservative

95\text{\,}\mathrm{\%}

CI.

Figure 4: External application to CAMUS: five patients. Top: input; bottom: beamspace CNN.

Ultrasound Imaging

Ultrasound imaging uses sound waves to visualize anatomy and function, with AI improving image reconstruction, diagnosis, measurement, and scan guidance.

See all 153 papers

EchoXFlow: A Beamspace Echocardiography Dataset for Cardiac Motion, Flow, and Function
May 6, 2026Elias Stenhede, Joanna Sulkowska, Eivind Bjørkan Orstad +2Medical Imaging Echocardiography
Medical Technology & E-Health, Akershus University Hospital, Norway · Department of Technology Systems, University of Oslo, Norway · Department of Cardiology, Akershus University Hospital, Norway
EchoVQA: Enabling Conversational Assistance for Point-of-Care Cardiac Ultrasound
May 22, 2026Filippos Bellos, Yutong Li, Jessie N Dong +6Medical Visual Question Answering Echocardiography
University of Michigan · University of Pennsylvania · Independent Scientist
EAGT: Echocardiography Augmentation for Generalisability and Transferability
May 14, 2026Soroush Elyasi, Sara Adibzadeh, Nasim Dadashi Serej +1Ultrasound Image Segmentation Data Augmentation
THRIVE Centre, University of West London, London, United Kingdom

May 6, 2026J/K move · Enter open · S save

EchoXFlow: A Beamspace Echocardiography Dataset for Cardiac Motion, Flow, and Function

Elias Stenhede, Joanna Sulkowska, Eivind Bjørkan Orstad, Henrik Schirmer+1

Medical Technology & E-Health, Akershus University Hospital, Norway · Department of Technology Systems, University of Oslo, Norway · Department of Cardiology, Akershus University Hospital, Norway

Medical Imaging Echocardiography Ultrasound Imaging Medical Image Benchmarks Multimodal Learning

Open paper AskPDF arXiv

We introduce EchoXFlow, a clinical echocardiography dataset with ultrasound in its native acquisition geometry. Existing public datasets offer limited opportunities to study how cardiac anatomy, myocardial motion, and blood flow relate: Doppler velocities are typically absent or fused as RGB overlays on grayscale anatomical images, and acquisitions are released only after scan conversion and lossy vendor display processing. EchoXFlow comprises 35,377 recordings from 666 routine-care examinations of 622 patients. It preserves the timing, geometry, and modality relationships needed for physically grounded echo learning. Each recording is retained as separable modality-specific streams: temporally resolved 1D, 2D, and 3D data alongside multiple Doppler modalities, paired with a synchronized ECG. Clinical annotations range from guideline-defined measurements to dense 2D myocardial contours and 3D left-ventricular endocardial meshes. With its associated open-source tooling, EchoXFlow enables cross-modal, acquisition-aware learning tasks that cannot be formulated from conventional scan-converted videos alone, and serves as a testbed for 4D vision and physically grounded multi-modal learning more broadly. The dataset is available at https://huggingface.co/datasets/Ahus-AIM/EchoXFlow and the tooling at https://github.com/Ahus-AIM/EchoXFlow

Abstract

Ultrasound Imaging

Explore similar work

EchoXFlow: A Beamspace Echocardiography Dataset for Cardiac Motion, Flow, and Function

EchoVQA: Enabling Conversational Assistance for Point-of-Care Cardiac Ultrasound

EAGT: Echocardiography Augmentation for Generalisability and Transferability