Interpretable photoacoustic phenotyping of distal microcirculation for peripheral artery disease diagnosis with exploratory perioperative assessment
Deng, H.; Yuwen, T.; Li, Z.; Xiang, J.; Bai, Y.; Zhang, N.; Fu, W.; Wang, X.; Guo, J.; Wu, W.; Ma, C.; Liu, M.-Y.
Show abstract
Peripheral artery disease (PAD) spans a continuum from large-vessel obstruction to distal microvascular dysfunction, yet routine non-invasive tests, including the ankle-brachial index (ABI), do not provide structurally resolved assessment of the foot microvascular bed and may be unreliable in the setting of medial arterial calcification or perioperative follow-up. Here we developed a clinic-oriented multispectral compound-scanning photoacoustic tomography system (MCPATS) for compression-free distal toe imaging, and an interpretable photoacoustic tomography distal microcirculation score, termed PACT-DMS, for phenotyping PAD-related distal vascular abnormalities. PACT-DMS was derived from anatomically standardized distal toe sections and integrated seven prespecified vascular features spanning trunk-vessel morphology, microvascular distribution and pulsation-related dynamics through a traceable linear support vector machine. In a prospective single-centre cohort of 45 participants, the bilateral fusion PACT-DMS model distinguished patients with PAD from healthy controls with an area under the receiver operating characteristic curve of 0.964 (95% CI, 0.907-1.000) and an accuracy of 91.1% (95% CI, 82.2%-97.8%) under subject-level leave-one-out cross-validation, supported by complementary robustness analyses. Exploratory analyses further showed that PACT-DMS identified abnormal distal vascular phenotypes in 6 of 9 clinically diagnosed PAD limbs with non-abnormal ABI and visualized distal vascular-bed changes before and after revascularization. These findings support MCPATS-enabled interpretable photoacoustic vascular phenotyping as a candidate adjunctive approach for distal microcirculatory assessment in PAD; larger multicentre studies with external validation and prespecified analysis protocols will be required to define its clinical role.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Indexing Cerebrovascular Health Using Near-infrared Spectroscopy 92%
- Opportunistic Assessment of Ischemic Heart Disease Risk Using Abdominopelvic Computed Tomography and Medical Record Data: a Multimodal Explainable Artificial Intelligence Approach 92%
- AngioNet: A Convolutional Neural Network for Vessel Segmentation in X-ray Angiography 92%
Similar papers in this journal
Similar papers in this journal
- A Pumpless Microfluidic Neonatal Lung Assist Device for Support of Preterm Neonates in Respiratory Distress 89%
- Reinnervation of Muscle Targets Enhances the Separability of Motor Unit Signals Following Peripheral Nerve Transfers 89%
- Shear stress targeted delivery of nitroglycerin to brain collaterals improves ischaemic stroke outcome 89%
Similar papers in this journal
- A novel art of continuous non-invasive blood pressure measurement 92%
- Physics-informed deep generative learning for quantitative assessment of the retina 90%
- Clinically Applicable Deep Learning for Intracranial Aneurysm Detection in Computed Tomography Angiography Images: A Comprehensive Multicohort Study 90%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.