The research, led by Mengyuan Jing, Haoxiang Lu, and Qing Liu with colleagues at Lanzhou University Second Hospital, the Guangdong Cardiovascular Institute, and partner institutions, set out to build an explainable machine learning model that could predict functional improvement after ablation in patients with atrial fibrillation combined with heart failure. Rather than relying on a single measurement, the team fused three complementary streams of information: the geometry of the left atrium and pulmonary veins as seen on cardiac CT angiography, quantitative radiomics features extracted from the left atrial wall, and a handful of routine clinical variables. The result was a combined model, dubbed COMB, that achieved an area under the receiver operating characteristic curve of 0.866 in the training set, 0.803 in the validation set, and 0.845 in an independent testing set drawn from other hospitals.

Those numbers deserve unpacking, because the area under the curve, or AUC, is the workhorse metric for judging how well a model separates patients who will improve from those who will not. A value of 0.5 would mean the model performs no better than a coin flip, while 1.0 would indicate perfect discrimination. Scores consistently above 0.80 across three separate patient cohorts, including an external test set the model had never encountered during development, represent solid, clinically meaningful performance. Equally important is how the team arrived at that performance. Patients from the primary institution were randomly divided into training and validation sets in a seven-to-three ratio, while two other organizations contributed entirely separate testing cohorts, a design that guards against the model simply memorizing the quirks of one hospital’s scanner or patient population.

The technical heart of the study lies in what the researchers chose to measure. Cardiac CT angiography, already a routine part of pre-ablation planning because it maps the pulmonary veins before catheter insertion, provides exquisitely detailed three-dimensional images of the left atrium, the chamber where fibrillation typically originates. From these images the team extracted morphological features describing the shape of the left atrium and the pulmonary veins, capturing subtle geometric signatures such as chamber distortion and remodeling that a human reader might overlook. In parallel, they applied radiomics, a computational approach that converts medical images into hundreds of quantitative descriptors of texture, intensity, and spatial pattern, to the left atrial wall itself, the thin muscular sleeve where ablation lesions are created and where fibrotic change often determines whether the procedure succeeds.

From this high-dimensional feature space, the researchers distilled two focused models. Two shape features and three left atrial wall radiomics features survived rigorous screening and were used to construct what they called the Shape model and the Wall model, respectively. Each model produced a continuous score, the Shape_score and the Wall_score, which quantified how strongly a given patient’s cardiac anatomy resembled that of patients who went on to improve. These scores were then integrated with the clinical variables that proved most informative, namely gender, hyperlipidemia, blood urea level, and the type of atrial fibrillation, to yield the final combined model. The parsimony is striking: out of the vast number of features that could have been included, the final predictor rests on just a handful of imaging and clinical inputs, all obtainable from examinations and blood tests that are already standard practice.

What elevates the study beyond a typical prediction exercise is its commitment to explainability. Black-box algorithms have long been a stumbling block for clinical adoption, because physicians are understandably reluctant to act on a probability they cannot interrogate. The researchers addressed this by applying SHAP, or Shapley additive explanations, a technique borrowed from cooperative game theory that assigns each input feature a precise contribution to every individual prediction. In effect, SHAP reveals which factors pushed a particular patient’s predicted probability up or down, and by aggregating these contributions across the cohort, the team identified the features with the greatest impact on outcomes in both the Shape and Wall models. This transparency allows a cardiologist to see, for example, whether an unfavorable prediction stems from pronounced atrial remodeling, abnormal wall texture suggesting fibrosis, or a clinical factor such as persistent rather than paroxysmal fibrillation.

The clinical stakes of this kind of tool are considerable. Patients with atrial fibrillation and concomitant heart failure represent a particularly vulnerable group, and ablation in these individuals carries procedural risk, substantial cost, and a demanding recovery. If a model can reliably flag patients unlikely to experience functional improvement, clinicians could counsel them more honestly, weigh alternative management strategies such as optimized medical therapy, or reserve ablation for those with the greatest expected benefit. Conversely, identifying patients with a high predicted probability of improvement could support earlier referral and shared decision-making grounded in quantitative evidence rather than clinical intuition alone. Because the model’s inputs come from CT angiography and routine laboratory work, it could in principle be deployed without any additional testing beyond what pre-ablation workups already include.

The multicenter architecture of the study strengthens its claims in another important way. Models trained and tested within a single institution often flatter themselves, absorbing center-specific artifacts in scanner protocol, image reconstruction, and patient mix. By including 240 patients in the training set, 101 in the validation set, and 75 in external testing sets from two other organizations, the team demonstrated that the model’s discrimination held up across different hospitals and imaging environments. The ethics committees of all three participating institutions approved the retrospective study, and the requirement for individual informed consent was waived because of its retrospective design, with the work conducted in accordance with the Declaration of Helsinki.

The authors themselves are careful about the limits of what they have shown. The study was retrospective, meaning it looked backward at patients who had already undergone ablation rather than prospectively assigning the model to guide care in real time. The team explicitly states that further prospective evaluation is required before the model can be implemented clinically, a caveat that applies to nearly every prediction algorithm now emerging in cardiovascular medicine. Prospective validation would involve using the COMB model to generate predictions before ablation and then tracking whether those predictions match observed outcomes, ideally across diverse populations and healthcare systems. Questions about calibration, the agreement between predicted probabilities and actual event rates, and about how the model behaves in subgroups underrepresented in the data, would also need answers.

Even with those caveats, the study offers a compelling glimpse of where cardiac imaging and artificial intelligence are converging. The same CT scan obtained to map a patient’s pulmonary veins becomes, through shape analysis and radiomics, a quantitative portrait of atrial health, and machine learning converts that portrait, together with a few clinical facts, into an individualized forecast of recovery. The SHAP framework keeps the forecast auditable, showing clinicians exactly which anatomical and biological signals drove the conclusion. For the millions of patients whose fibrillation and heart failure travel together, and for the physicians deciding whether to offer them ablation, a tool that turns pre-procedural images into evidence-based expectations could reshape the conversation, replacing uncertainty with a number that both doctor and patient can understand, question, and trust.

Subject of Research: Explainable machine learning on cardiac CT angiography to predict functional improvement after atrial fibrillation ablation in patients with heart failure

Article Title: Explainable machine learning model based on cardiac CT angiography for predicting functional improvement after atrial fibrillation ablation: a multicenter study

Article References: Jing, M., Lu, H., Liu, Q., Jing, Y., Lei, F., Yang, X., Chen, G., Xi, H., Xin, W., Zhu, H., Sun, Q., Zhang, Y., Ren, J., Ren, W., Liu, Z., Wang, G., & Zhou, J. (2026). Explainable machine learning model based on cardiac CT angiography for predicting functional improvement after atrial fibrillation ablation: a multicenter study. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02822-1

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02822-1

Keywords: atrial fibrillation, heart failure, catheter ablation, cardiac CT angiography, machine learning, radiomics, left atrium, SHAP explainability, predictive modeling, multicenter study, BMC Medical Imaging, cardiac imaging

The study, published in the European Journal of Nuclear Medicine and Molecular Imaging, was led by Rong Wang of Gansu Provincial Hospital in Lanzhou and Wei Dong of Beijing Anzhen Hospital, Capital Medical University. The researchers enrolled 174 patients who had undergone percutaneous coronary intervention, or PCI, the umbrella term for stenting and related catheter-based procedures. Crucially, every participant had angiographically confirmed complete revascularization, meaning their large epicardial arteries showed no significant residual blockages. In other words, by conventional measures, these hearts had been fixed. The question was whether a deeper look would reveal lingering trouble.

The deeper look came from a technology called dynamic SPECT, abbreviated D-SPECT, which uses fast cadmium-zinc-telluride detectors to track a radioactive tracer moving through the heart muscle in real time. Traditional perfusion scans capture a single static snapshot of blood distribution, revealing only whether some regions of the heart look dimmer than others. Dynamic imaging goes further: by measuring how quickly the tracer arrives and washes out, clinicians can calculate absolute blood flow in milliliters per gram per minute, both at rest and during pharmacological stress, when a drug dilates the vessels to simulate exercise. The ratio of stress flow to rest flow yields the myocardial flow reserve, or MFR — a numerical gauge of how much extra blood delivery the heart can summon when demand rises.

A healthy heart can multiply its resting blood flow several-fold under stress. When that reserve falls below 2.0, it signals that the coronary microvasculature — the dense network of arterioles and capillaries too small to appear on any angiogram — cannot dilate properly. The researchers defined coronary microvascular disease, or CMD, as a global or territorial MFR below this threshold in the absence of any epicardial stenosis greater than 50 percent. This definition matters because microvascular dysfunction is increasingly recognized as a disease entity in its own right, one that causes ischemia, drives symptoms, and worsens outcomes even when the conduit arteries are wide open. Prior work has linked it to adverse remodeling of the heart muscle, diastolic dysfunction, and elevated risk in patients with diabetes and prior infarction.

The results were striking. Over a median follow-up of 12 months, during which some patients were followed for as long as 21 months, 42 of the 174 patients — 24.1 percent — suffered a major adverse cardiac event. The composite endpoint was deliberately broad, encompassing all-cause death, myocardial infarction, unscheduled coronary revascularization, stroke, hospitalization for heart failure, and rehospitalization for angina. When the investigators split the cohort by microvascular status, the difference was unmistakable: patients with CMD experienced significantly higher event rates, a separation visible on survival curves with a log-rank statistic of 11.375 and a p-value of 0.001.

What makes the finding clinically persuasive is that it survived rigorous statistical scrutiny. The team applied multivariable Cox regression, adjusting for traditional cardiovascular risk factors and for abnormal findings on conventional myocardial perfusion imaging. Even after accounting for these established predictors, CMD remained an independent predictor of adverse outcomes, carrying a hazard ratio of 2.46 — meaning patients with impaired flow reserve faced roughly two and a half times the risk of a major event compared with those whose microvessels functioned normally. The 95 percent confidence interval ranged from 1.27 to 4.76, and the p-value of 0.007 indicated the association was unlikely to be chance. Notably, the association held consistently across all prespecified patient subgroups.

The study also delivered a technical refinement with practical implications. Rather than relying solely on a single global MFR value averaged over the whole left ventricle, the researchers analyzed regional MFR — flow reserve measured separately in the territories supplied by each of the major coronary arteries. Statistical reclassification analyses quantified how much this regional approach added. The net reclassification improvement, or NRI, was 0.524, and the integrated discrimination improvement, or IDI, was 0.036, both statistically significant. In plain terms, looking at the heart territory by territory reclassified a meaningful fraction of patients into more accurate risk categories and improved the model’s ability to distinguish those who would have events from those who would not. A globally normal flow reserve can mask a severely diseased single territory, and the regional analysis catches what the average conceals.

The implications extend beyond the cardiology ward. Coronary microvascular dysfunction has long been difficult to assess outside specialized centers, historically requiring invasive catheter-based measurements such as the index of microcirculatory resistance, which involves threading a pressure-temperature sensor wire into the coronary artery. Noninvasive quantification with dynamic CZT-based SPECT offers a way to screen patients after PCI without additional catheterization. If confirmed, the approach could change post-procedural care: patients flagged with low flow reserve might be candidates for intensified medical therapy targeting the microcirculation — including aggressive risk-factor control, antiplatelet and lipid-lowering regimens, and emerging microvascular-directed treatments — while those with robust reserves might be spared unnecessary interventions and anxiety.

The authors are careful to frame their conclusions appropriately. This was a pilot study, exploratory in design, with a relatively modest sample size and a median follow-up of only one year — short by the standards of cardiovascular outcomes research. The researchers themselves call for validation in larger multi-center studies with longer follow-up and direct comparisons against invasive reference measurements. The cohort was drawn from two Chinese hospitals, Gansu Provincial Hospital and Beijing Anzhen Hospital, and whether the findings generalize to other populations remains to be demonstrated. Ethical approval was obtained from both institutions, and the work was funded by the National Natural Science Foundation of China along with provincial and municipal research programs.

Even with those caveats, the study adds an important piece to one of cardiology’s most persistent puzzles: why symptoms and events persist after anatomically successful revascularization. It reinforces a conceptual shift that has been gathering momentum in the field — that the coronary circulation must be evaluated as a whole, from the epicardial conduits down to the microscopic vessels where oxygen actually exchanges with heart muscle. For the roughly one in four patients in this cohort destined for a major cardiac event despite complete stenting, a 20-minute dynamic nuclear scan may soon offer something that angiography never could: a genuine look at the plumbing that matters most, and a head start on preventing the next crisis before it strikes.

Subject of Research: D-SPECT-derived myocardial flow reserve for detecting coronary microvascular disease and predicting outcomes after complete revascularization

Article Title: Prognostic value of D-SPECT-derived myocardial flow reserve in detecting coronary microvascular disease and predicting outcomes after complete revascularization: a two-center pilot study

Article References: Wang, R., Bai, J., Xie, P., Wang, H., Wang, X., Gao, Z., Chen, Y., Mi, H., & Dong, W. (2026). Prognostic value of D-SPECT-derived myocardial flow reserve in detecting coronary microvascular disease and predicting outcomes after complete revascularization: a two-center pilot study. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08162-0

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08162-0

Keywords: D-SPECT, myocardial flow reserve, coronary microvascular disease, percutaneous coronary intervention, major adverse cardiac events, nuclear cardiology, SPECT imaging, coronary microcirculation, prognosis, revascularization, cardiac imaging, Prognostic

The central question the Iranian team asked was deceptively simple: can a quantitative calcium score measured on a gated CT scan predict whether a PTMC procedure will succeed on the table, in real time, before the patient leaves the catheterization laboratory? For decades, cardiologists have relied on the Wilkins score, an echocardiographic grading system that scores four features of the mitral valve, leaflet mobility, leaflet thickening, calcification, and the state of the subvalvular apparatus, each on a scale from one to four. A combined score above eight generally warns of a difficult procedure and poorer outcomes. The Wilkins score has endured because it is cheap and uses equipment every cardiology unit already owns, but it is subjective, semi-quantitative, and depends heavily on the experience of the reader. CT, by contrast, offers an objective, reproducible number derived directly from the density of calcium deposits in the valve tissue.

The study enrolled thirty patients with severe rheumatic mitral stenosis who were scheduled for PTMC. Each participant underwent both a standard transthoracic echocardiographic assessment, from which the Wilkins score was calculated, and an electrocardiogram-gated CT scan of the heart, from which the researchers derived a mitral valve calcium score, abbreviated MVCS. The investigators then defined procedural success in clinically meaningful terms: either a post-procedure mitral valve area of at least 1.5 square centimeters, or at least a fifty percent increase in valve area from the baseline measurement, without the development of significant mitral regurgitation, meaning moderate-severe or severe leakage through the valve. This dual criterion matters because a balloon that splits the valve too aggressively can convert a stenotic valve into a leaking one, an outcome that is often worse than the disease it was meant to treat.

The headline result was that procedural success was achieved in eighteen of the thirty patients, a rate of sixty percent. When the researchers compared the patients in whom the procedure failed with those in whom it succeeded, several patterns emerged. Failed procedures were associated with numerically higher Wilkins scores, averaging 9.55 compared with 8.58 in the success group, a difference that approached but did not clearly cross the threshold of conventional statistical significance, with a p value of 0.05. More strikingly, patients with failed procedures had dramatically higher CT-derived calcium scores on average, 16.08 versus 0.76, and a far greater burden of significant post-procedural mitral regurgitation, affecting 41.7 percent of the failure group compared with none of the success group.

Yet the researchers were careful, and admirably so, about what these numbers do and do not prove. Because the CT-derived calcium measurements followed a zero-inflated distribution, meaning that many patients had no detectable valve calcium at all while a few had very high scores, the team performed an exploratory binary analysis comparing patients with detectable calcium against those without. In that analysis, the association between calcium burden and procedural failure was numerically apparent but statistically inconclusive, with a p value of 0.929 for the comparison of mean calcium scores. In other words, the raw averages looked compelling, but with only thirty patients, the study lacked the statistical power to confirm that the CT calcium score independently predicts outcome. This is a crucial nuance that separates responsible science from headline-grabbing overclaiming, and it is a nuance the authors themselves emphasize in their conclusions.

What the study did establish more firmly was the continued relevance of the Wilkins score. Using receiver operating characteristic analysis, the investigators found that the Wilkins score showed moderate discrimination for the immediate procedural outcome, with an area under the curve of 0.72 and a ninety-five percent confidence interval spanning 0.51 to 0.93. An AUC of 0.72 means that if you picked one patient who succeeded and one who failed at random, the Wilkins score would correctly rank the higher-risk patient about seventy-two percent of the time. That is useful but far from definitive, which is precisely why the search for complementary imaging markers continues. The confidence interval, stretching from barely better than chance to quite good, reflects the small sample size and underscores the need for larger validation cohorts.

The biological logic behind the calcium hypothesis is sound and worth understanding. Rheumatic mitral stenosis is fundamentally a disease of fusion, the commissures, or junctions, between the valve leaflets are glued together by inflammation and scarring following rheumatic fever. The balloon works by forcibly splitting these fused commissures. When the leaflets are pliable and the fusion is the dominant problem, the balloon splits cleanly and the valve opens. But when the leaflets are heavily calcified, the calcium deposits act like concrete within the valve structure, resisting the balloon and, worse, fracturing unpredictably. A crack through a calcium plate rather than along a commissural line can tear the leaflet itself, producing the abrupt, severe regurgitation that the study observed in the failed procedures. Calcium, in this sense, is not merely a marker of chronicity but a direct mechanical obstacle to the procedure’s mechanism of action.

This is where CT scoring offers a genuine conceptual advance over echocardiography. The Wilkins calcification grade asks a sonographer to eyeball the valve and assign a subjective number. A CT calcium score, by contrast, applies an objective attenuation threshold, typically around 130 Hounsfield units, the same standard used for coronary artery calcium scoring, and quantifies the actual volume or mass of calcified tissue. The measurement is reproducible across readers and centers, can be acquired during a single breath-hold, and increasingly comes free of charge when patients already undergo cardiac CT for other reasons, such as pre-procedural planning for left atrial appendage occlusion or AFib ablation. For patients in whom a CT scan is already planned, extracting a mitral valve calcium score adds no radiation, no cost, and no additional visit.

The clinical implications, if the findings are confirmed, could be significant for the global burden of rheumatic heart disease. Although rheumatic fever has become rare in high-income countries, it remains endemic across South Asia, sub-Saharan Africa, the Middle East, and parts of Latin America, where PTMC is often the only realistic treatment option given the scarcity of cardiac surgery capacity. A reliable pre-procedural predictor of failure would allow clinicians to triage patients more intelligently, reserving balloon commissurotomy for valves likely to respond and directing high-calcium valves toward surgical repair or replacement where feasible. It could also inform procedural technique itself, since operators might approach heavily calcified valves with smaller balloon sizes or staged dilations to reduce the risk of catastrophic leaflet tearing.

The authors of the study are appropriately measured in their conclusions, stating that CT calcium scoring may serve as a complementary anatomical marker in selected patients but that larger studies are needed to define its incremental value over the Wilkins score. That framing is exactly right for a thirty-patient exploratory investigation. What the study provides is a proof of concept and a methodological template: a rigorous definition of success, a quantitative imaging biomarker, and an honest statistical treatment of a skewed data distribution. If subsequent multicenter trials with hundreds of patients confirm that a threshold CT calcium score reliably identifies valves destined to fail balloon dilation, the mitral valve calcium score could join the Wilkins score as a standard element of the pre-PTMC workup, giving cardiologists in resource-limited settings a sharper tool for deciding which valves can be saved by a balloon and which require something more. For now, the message for patients and clinicians alike is one of cautious optimism: the technology to see inside a stiffened valve in exquisite detail already exists, and the evidence that it can predict procedural fate is accumulating, one carefully designed study at a time.

Subject of Research: CT-based mitral valve calcium scoring as a predictor of immediate outcomes after percutaneous transvenous mitral commissurotomy in rheumatic mitral stenosis

Article Title: CT mitral valve calcium score and immediate outcomes after percutaneous transvenous mitral commissurotomy

Article References: Eini, P., Jolfayi, A. G., Firouzi, A., Rafiee, F., Ranjbar, B., Kaviani, R., Parsaee, M., Alemzadeh-Ansari, M. J., Bakhshandeh, H., Maleki, M., & Houshmand, G. (2026). CT mitral valve calcium score and immediate outcomes after percutaneous transvenous mitral commissurotomy. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02843-w

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02843-w

Keywords: rheumatic mitral stenosis, PTMC, mitral valve calcium score, computed tomography, Wilkins score, echocardiography, mitral regurgitation, balloon commissurotomy, cardiac imaging, procedural outcomes, valvular heart disease, BMC Medical Imaging