Abstract
Poor medical image quality hinders accurate clinical decision making. Image quality in ultrasound is modulated by many variables, including positioning and acquisition parameter choices. Optimising these variables is slow because human judgement of image quality is required. Existing automatic image quality assessment (IQA) methods are overly general, not adaptable, or are designed for static tasks. We argue that the clinical value of an image depends on the quality of specific anatomical regions rather than the image as a whole. Based on this, we propose an IQA method that conditions the quality score on the operator’s anatomical focus at scan time. By exploiting a contrastively pre-trained biomedical vision-language model (BiomedCLIP), TargetIQA requires no dedicated training and outperforms trained organ-specific IQA methods, a pre-trained VLM with a 138× higher parameter count, and general IQA methods across a variety of tasks. Our ablations show that TargetIQA is sensitive to the anatomical context, and this context meaningfully changes its quality scores. These findings show anatomical conditioning is a lightweight but effective basis for clinically meaningful IQA.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/ASMUS_041.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026-sat/supp/ASMUS_041_supp.pdf
Link to Open Review
Open Review Page: https://openreview.net/forum?id=UpUAL6RSDl
BibTex
@InProceedings{LanGeo_TargetIQA_MICCAISAT2026,
author = { Lansdown, George AND Tsaftaris, Sotirios A. AND Smithard, Alison Q. AND Kascenas, Antanas},
title = { { TargetIQA: Anatomy-Aware Ultrasound Image Quality Assessment } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17276},
month = {pending},
page = {pending}
}