The research team, led by pediatric surgeon Mehmet Metin together with Nurcan Çoşkun, Hülya İpek, Gül Doğan, and senior author Çağatay Evrim Afşarlar, analyzed 675 patients younger than 18 years who underwent appendectomy between 2019 and 2024 at a single center. Rather than relying on clinical impressions, the investigators used histopathology — the microscopic examination of the removed appendix — as the reference standard for whether appendicitis was truly present. This design anchors the study’s numbers to biological ground truth, a level of rigor that many imaging studies in this field lack.

The headline finding concerns how often each imaging modality got the diagnosis right. Ultrasonography, the radiation-free first-line test recommended by pediatric guidelines, detected appendicitis with 79.9 percent sensitivity but only 51.1 percent specificity. In practical terms, ultrasound missed roughly one in five confirmed cases of appendicitis, and when it did flag a problem, it was wrong nearly half the time. Computed tomography showed a different profile: 76.8 percent sensitivity paired with a substantially better 71.7 percent specificity, meaning CT was somewhat more likely than ultrasound to correctly rule out appendicitis when the appendix was actually normal.

These figures illuminate a persistent tension in pediatric radiology. Ultrasound is favored first because it involves no ionizing radiation, is widely available, and children tolerate it well. But the appendix is a small, mobile, gas-filled target, and in children with abdominal pain and guarding, ultrasound visualization can be technically difficult. CT, by contrast, images the entire abdomen with high spatial resolution and is less operator-dependent, but it delivers a radiation dose that, in a developing body, carries a small lifetime cancer risk — a concern that has driven decades of effort to reduce pediatric CT use through clinical scoring systems, MRI protocols, and staged imaging pathways.

The study also explored what happens when both modalities are used in the same patient. In a subgroup of 198 children who had undergone both ultrasound and CT, the investigators tested an exploratory

The exploratory combined rule tested in the dual-imaging subgroup deserves careful interpretation. When the investigators treated a positive result on either ultrasound or CT as sufficient to call appendicitis, sensitivity climbed to 89.2 percent and overall accuracy reached 84.8 percent — figures exceeding either modality alone. The cost was specificity, which fell to 67.5 percent, below CT’s standalone performance. This pattern illustrates a fundamental property of diagnostic test combinations: serially interpreting two tests with an either-positive rule amplifies sensitivity because a patient needs only one of two chances to test positive, but it simultaneously compounds false positives. The authors themselves are explicit that this analysis was exploratory and does not establish that combined imaging is superior overall; the subgroup was not randomized, and the decision to obtain both studies may itself have been driven by clinical uncertainty, embedding selection effects into the result.

That caveat connects to one of the study’s most consequential limitations: CT was performed in 333 of the 675 patients, or 49.3 percent of the cohort, yet the timing, sequence, and clinical indications for those scans could not be consistently retrieved from the records. This is more than a bookkeeping gap. Current pediatric guidelines envision CT as a second-line tool reserved for equivocal ultrasound findings, so nearly half of the cohort receiving CT raises the question of how often the scan followed a genuinely indeterminate ultrasound versus how often it was obtained for other reasons — surgeon preference, emergency department throughput, or medicolegal caution. Without patient-level indication data, the study cannot quantify CT overuse or assess whether the scans were appropriate, which is precisely the utilization question its title promises to address. The authors acknowledge this directly, noting that missing indications prevent any assessment of CT appropriateness.

The design also carries a verification bias that the investigators flag candidly. Because every child in the cohort underwent appendectomy, the study includes no children who were imaged, found to be free of appendicitis, and sent home without surgery. Specificity estimates derived only from operated patients are therefore systematically distorted: children with negative imaging who avoided surgery — the very cases in which ultrasound’s rule-out ability matters most — never entered the denominator. This inflates the apparent false-positive rate of ultrasound and complicates direct comparison with studies that include non-operated patients. It is a well-known structural limitation of appendectomy-based cohorts, and it means the reported specificity figures should be read as conditional on the decision to operate rather than as the real-world performance of the tests across all comers.

Beyond the imaging results, the study contributes a detailed picture of the inflammatory markers that accompany appendicitis in children. C-reactive protein, white blood cell count, absolute neutrophil count, and the neutrophil-to-lymphocyte ratio were all significantly associated with both the diagnosis and the severity of appendicitis, with P values below 0.001 across the comparisons. The neutrophil-to-lymphocyte ratio, a cheap and widely available index of systemic inflammation, has attracted growing interest in surgical literature as a potential adjunct to imaging, and its association with severity in this cohort is consistent with that broader literature. These markers cannot replace imaging, but they can help clinicians stratify pretest probability and decide when an equivocal ultrasound warrants further workup.

Another underappreciated finding is the systematic discrepancy in appendiceal diameter. Both ultrasound and CT underestimated the diameter of the appendix relative to the measurements obtained on histopathological examination, quantified through intraclass correlation coefficients. This is plausible on mechanistic grounds: the appendix is a collapsible tubular structure, and imaging captures it under physiological conditions of intraluminal pressure and muscular tone, whereas fixation and processing of the surgical specimen can alter luminal distension and wall thickness. The practical implication is that diameter thresholds used to declare an appendix abnormal on imaging are calibrated against imaging-based norms, not against the pathological specimen, and clinicians should be cautious about dismissing appendicitis solely because a measured diameter falls just below a published cutoff.

The statistical apparatus of the study also merits note. Beyond conventional sensitivity and specificity, the investigators reported Cohen’s kappa and Gwet’s AC1 to quantify categorical agreement between imaging results and histopathology, and used intraclass correlation coefficients for the continuous diameter measurements. Gwet’s AC1 is particularly useful in settings with high prevalence of agreement, where Cohen’s kappa is known to behave paradoxically and can underestimate agreement; including both gives a more robust picture of how well imaging and pathology correspond. Reporting both metrics reflects an increasing trend toward methodological rigor in diagnostic accuracy research.

Placed in context, the findings reinforce the existing guideline architecture rather than overturning it. Ultrasound’s 79.9 percent sensitivity, while imperfect, is consistent with the range reported in pediatric literature, and its role as a first-line test rests on the absence of radiation rather than on diagnostic supremacy. CT’s higher specificity supports its designated niche as the problem-solving modality for equivocal cases. The study’s real contribution is its unvarnished documentation of how imaging is actually used in practice — with CT deployed in nearly half of appendectomy patients and indication records incomplete — and its demonstration that even in a pathology-anchored cohort, the operational details of imaging pathways remain difficult to reconstruct. For institutions seeking to tighten stewardship of pediatric CT, the message is that structured documentation of imaging indications is a prerequisite: without knowing why each scan was ordered, quality improvement efforts cannot distinguish appropriate escalation from reflexive utilization.

The single-center design in Çorum, Türkiye, also shapes generalizability. Ultrasound performance is notoriously operator-dependent, and results from a high-volume pediatric surgery service with experienced sonographers may not transfer to settings with different expertise or equipment availability. Conversely, the cohort’s size — 675 appendectomies over five years, with 571 histopathologically confirmed cases of appendicitis, a positive rate of 84.6 percent — provides statistical weight that smaller single-center series often lack. The high proportion of confirmed appendicitis also reflects the reality that surgical decision-making in children is conservative, with many equivocal cases proceeding to the operating room, which is precisely the clinical behavior the study’s verification-bias caveat illuminates.

For clinicians, the actionable synthesis is straightforward. Ultrasound remains the appropriate starting point, inflammatory markers can sharpen pretest probability, and CT should be deployed selectively when ultrasound leaves the diagnosis genuinely unresolved. The combined either-positive rule, while attractive on paper for its 89.2 percent sensitivity, should not be adopted as a routine strategy without prospective validation, since its specificity penalty would translate into more negative appendectomies — the outcome pediatric surgery works hardest to avoid. What the study ultimately documents is a gap between guideline intent and bedside practice, and it makes the case that closing that gap begins with better data on why, when, and in whom each imaging study is ordered.

Subject of Research: Diagnostic imaging performance and computed tomography utilization in children undergoing appendectomy for suspected appendicitis

Article Title: Computed tomography utilization patterns and diagnostic imaging performance in children undergoing appendectomy for suspected appendicitis: a real-world cohort study

Article References: Metin, M., Çoşkun, N., İpek, H., Doğan, G., & Afşarlar, Ç. E. (2026). Computed tomography utilization patterns and diagnostic imaging performance in children undergoing appendectomy for suspected appendicitis: a real-world cohort study. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02957-1

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02957-1

Keywords: appendicitis, computed tomography, ultrasonography, pediatric surgery, diagnostic imaging, histopathology, inflammatory markers, C-reactive protein, medical imaging, appendectomy, diagnostic accuracy, verification bias

A new study published in Annals of Gastroenterological Surgery asks whether machine learning could tip that balance. Researchers from the University Hospital of Split in Croatia analyzed the records of 1,547 adults operated on for suspected appendicitis between January 2020 and June 2024, ultimately assembling a rigorously curated dataset of 623 patients. Of these, 66 had a histologically normal appendix, 248 had uncomplicated appendicitis, and 309 had complicated appendicitis confirmed by pathologists. Crucially, every case was verified by histopathology, eliminating the diagnostic guesswork that plagues many prior studies in which patients were classified by clinical or radiological impression alone.

The team built and compared four machine learning models: logistic regression, random forest, balanced random forest, and XGBoost, an extreme gradient boosting algorithm. All models relied exclusively on routinely available clinical and laboratory data—age, sex, symptom duration, fever, nausea, vomiting, rebound tenderness, pain migration, white blood cell count, neutrophil and lymphocyte percentages, platelet indices, C-reactive protein, and serum sodium. No imaging data were included. Missing values, which ranged from zero to 5.3 percent across variables, were imputed with a bagged trees algorithm carefully configured to prevent data leakage between training and test sets.

The methodological centerpiece of the study was a nested cross-validation design, with five inner and outer folds repeated ten times, yielding fifty independent test evaluations. Hyperparameters were tuned exclusively on inner validation data, and performance was then measured on untouched outer test folds, keeping performance estimates unbiased. On top of that, the researchers applied bootstrap resampling with 2,000 iterations to confirm that results were stable and not artifacts of a particular split of the data.

The most distinctive innovation, however, was threshold optimization. Rather than accepting a default probability cutoff of 0.5, the team deliberately shifted decision thresholds to meet six predefined minimum sensitivity targets, ranging from 0.95 to 0.995. Within each outer fold, thresholds were derived from pooled inner validation predictions, fixed, and then applied unchanged to the test set. The logic is clinical, not statistical: in appendicitis, missing a true case is far more dangerous than operating on a healthy appendix, so a decision support tool must be calibrated to catch virtually every case of appendicitis while identifying the small minority of patients who might safely avoid immediate surgery.

The results were revealing. For detecting acute appendicitis, logistic regression outperformed the more complex algorithms, achieving an area under the ROC curve of 0.765. At the strictest sensitivity target of 0.995, the model detected 99.8 percent of appendicitis cases, with specificity of just 0.038. Translated to a clinical scale, using the observed case distribution, this operating point would miss approximately two appendicitis cases per 1,000 patients while correctly flagging about four patients without appendicitis who could be spared surgery. At the more lenient sensitivity target of 0.95, roughly 41 appendicitis cases per 1,000 would be missed, but approximately 24 patients per 1,000 without appendicitis would avoid an unnecessary operation. The numbers lay bare the inherent trade-off between diagnostic safety and operative selectivity.

For distinguishing complicated from uncomplicated disease, the random forest model performed best, with an AUC of 0.785 and a maximum achievable sensitivity of 0.997 paired with specificity of 0.057. This capability carries real clinical weight: complicated appendicitis generally demands urgent surgery and carries higher morbidity and mortality, whereas some centers now attempt conservative, non-operative management of uncomplicated cases. A model that reliably identifies complicated disease could support more urgent operative decisions, temper enthusiasm for non-operative strategies, or prompt additional imaging and closer monitoring when the predicted risk is ambiguous.

To address the notorious black-box problem that undermines clinical trust in machine learning, the researchers applied SHAP analysis, a framework from explainable artificial intelligence that quantifies each feature’s contribution to individual predictions. For appendicitis detection, neutrophil percentage, the neutrophil-to-lymphocyte ratio, the platelet-to-lymphocyte ratio, sex, and lymphocyte percentage emerged as the most influential predictors. For complication prediction, C-reactive protein, age, symptom duration, and the neutrophil-to-lymphocyte ratio dominated. These rankings align closely with the pathophysiology of appendicitis, in which bacterial invasion of the appendiceal mucosa triggers a systemic inflammatory cascade, and they mirror the feature importances reported by other research groups using different algorithms and populations.

The authors are careful to frame what these models are and are not. The cohort consisted exclusively of adults already selected for surgery by a surgeon’s judgment, so the models should not be treated as stand-alone diagnostic tools for undifferentiated abdominal pain in the emergency department, nor as replacements for ultrasound or CT where those are routinely available. Instead, the researchers envision them as adjunctive decision-support and reassessment tools: a low-risk prediction should trigger repeat examination, laboratory reassessment, or additional imaging rather than discharge, particularly in settings with limited imaging access or after equivocal imaging findings. Notably, the superiority of the simple linear model over tree-based methods in this adult cohort contrasts with findings in pediatric populations, suggesting that inflammatory markers relate to appendicitis in a more linear fashion in adults and underscoring the need for population-specific models.

Limitations remain. The study was retrospective and single-center, and its performance may reflect local diagnostic pathways, surgeon thresholds, and imaging practices that do not transfer to other institutions. The dataset is also surgically enriched, meaning there are no true negative patients who avoided surgery, so specificity must be interpreted conditionally within the operated population. Still, the combination of histopathological verification, rigorous nested cross-validation, bootstrap stability testing, deliberate sensitivity-first threshold calibration, and transparent SHAP-based interpretability marks this work as a methodologically serious step toward machine learning that could meaningfully reduce negative appendectomies. The authors call for prospective, multicenter external validation before clinical deployment, along with future exploration of multimodal models integrating imaging and of tools to help select patients with uncomplicated appendicitis for non-operative management—a direction that could ultimately reshape how one of surgery’s most routine emergencies is decided.

Subject of Research: Machine learning with sensitivity-focused threshold optimization to reduce negative appendectomies in adults with suspected acute appendicitis

Article Title: Can Machine Learning Reduce Unnecessary Surgeries? A Retrospective Analysis Using Threshold Optimization to Prevent Negative Appendectomies in Adults

Article References: Males, I., Kumric, M., Boban, Z., Vrdoljak, J., Pecenkovic, D., Ivanda, M., Grahovac, M., Pogorelic, Z., & Bozic, J. (2026). Can Machine Learning Reduce Unnecessary Surgeries? A Retrospective Analysis Using Threshold Optimization to Prevent Negative Appendectomies in Adults. Annals of Gastroenterological Surgery, 10(5), 1631-1645. https://doi.org/10.1002/ags3.70225

Image Credits: AI Generated

DOI: 10.1002/ags3.70225

Keywords: machine learning, appendicitis, negative appendectomy, logistic regression, random forest, XGBoost, threshold optimization, SHAP explainability, clinical decision support, C-reactive protein, nested cross-validation, emergency surgery