Machine Learning Applied to Pre-Operative Computed-Tomography-Based Radiomic Features Can Accurately Differentiate Uterine Leiomyoma from Leiomyosarcoma: A Pilot Study.

BACKGROUND: The accurate discrimination of uterine leiomyosarcomas and leiomyomas in a pre-operative setting remains a current challenge. To date, the diagnosis is made by a pathologist on the excised tumor. The aim of this study was to develop a machine learning algorithm using radiomic data extracted from contrast-enhanced computed tomography (CECT) images that could accurately distinguish leiomyosarcomas from leiomyomas. METHODS: Pre-operative CECT images from patients submitted to surgery with a histological diagnosis of leiomyoma or leiomyosarcoma were used for the region of interest identification and radiomic feature extraction. Feature extraction was conducted using the PyRadiomics library, and three feature selection methods combined with the general linear model (GLM), random forest (RF), and support vector machine (SVM) classifiers were built, trained, and tested for the binary classification task (malignant vs. benign). In parallel, radiologists assessed the diagnosis with or without clinical data. RESULTS: A total of 30 patients with leiomyosarcoma (mean age 59 years) and 35 patients with leiomyoma (mean age 48 years) were included in the study, comprising 30 and 51 lesions, respectively. Out of nine machine learning models, the three feature selection methods combined with the GLM and RF classifiers showed good performances, with predicted area under the curve (AUC), sensitivity, and specificity ranging from 0.78 to 0.97, from 0.78 to 1.00, and from 0.67 to 0.93, respectively, when compared to the results obtained from experienced radiologists when blinded to the clinical profile (AUC = 0.73 95%CI = 0.62-0.84), as well as when the clinical data were consulted (AUC = 0.75 95%CI = 0.65-0.85). CONCLUSIONS: CECT images integrated with radiomics have great potential in differentiating uterine leiomyomas from leiomyosarcomas. Such a tool can be used to mitigate the risks of eventual surgical spread in the case of leiomyosarcoma and allow for safer fertility-sparing treatment in patients with benign uterine lesions.

Cardiac Magnetic Resonance to Predict Cardiac Mass Malignancy: The CMR Mass Score.

BACKGROUND: Multimodality imaging is currently suggested for the noninvasive diagnosis of cardiac masses. The identification of cardiac masses' malignant nature is essential to guide proper treatment. We aimed to develop a cardiac magnetic resonance (CMR)-derived model including mass localization, morphology, and tissue characterization to predict malignancy (with histology as gold standard), to compare its accuracy versus the diagnostic echocardiographic mass score, and to evaluate its prognostic ability. METHODS: Observational cohort study of 167 consecutive patients undergoing comprehensive echocardiogram and CMR within 1-month time interval for suspected cardiac mass. A definitive diagnosis was achieved by histological examination or, in the case of cardiac thrombi, by histology or radiological resolution after adequate anticoagulation treatment. Logistic regression was performed to assess CMR-derived independent predictors of malignancy, which were included in a predictive model to derive the CMR mass score. Kaplan-Meier curves and Cox regression were used to investigate the prognostic ability of predictors. RESULTS: In CMR, mass morphological features (non-left localization, sessile, polylobate, inhomogeneity, infiltration, and pericardial effusion) and mass tissue characterization features (first-pass perfusion and heterogeneity enhancement) were independent predictors of malignancy. The CMR mass score (range, 0-8 and cutoff, ≥5), including sessile appearance, polylobate shape, infiltration, pericardial effusion, first-pass contrast perfusion, and heterogeneity enhancement, showed excellent accuracy in predicting malignancy (areas under the curve, 0.976 [95% CI, 0.96-0.99]), significantly higher than diagnostic echocardiographic mass score (areas under the curve, 0.932; P=0.040). The agreement between the diagnostic echocardiographic mass and CMR mass scores was good (κ=0.66). A CMR mass score of ≥5 predicted a higher risk of all-cause death (P<0.001; hazard ratio, 5.70) at follow-up. CONCLUSIONS: A CMR-derived model, including mass morphology and tissue characterization, showed excellent accuracy, superior to echocardiography, in predicting cardiac masses malignancy, with prognostic implications.

Prognostic Role of Early Cardiac Magnetic Resonance in Myocardial Infarction With Nonobstructive Coronary Arteries.

BACKGROUND: Cardiac magnetic resonance (CMR) plays a pivotal diagnostic role in myocardial infarction with nonobstructive coronary arteries (MINOCA). To date, a prognostic stratification of these patients is still lacking. OBJECTIVES: This study aims to assess the prognostic role of CMR in MINOCA. METHODS: The authors assessed 437 MINOCA from January 2017 to October 2021. They excluded acute myocarditis, takotsubo syndromes, cardiomyopathies, and other nonischemic etiologies. Patients were classified into 3 subgroups according to the CMR phenotype: 1) presence of late gadolinium enhancement (LGE) and abnormal mapping (M) values (LGE+/M+); 2) regional ischemic injury with abnormal mapping and no LGE (LGE-/M+); and 3) nonpathological CMRs (LGE-/M-). The primary outcome was the presence of major adverse cardiovascular events (MACE). The mean follow-up was 33.7 ± 12.0 months and CMR was performed on average at 4.8 ± 1.5 days from the acute presentation. RESULTS: The final cohort included 198 MINOCA; 116 (58.6%) comprised the LGE+/M+ group. During follow-up, MACE occurred significantly more frequently in MINOCA LGE+/M+ than in the LGE+/M- and normal-CMR (LGE-/M-) subgroups (20.7% vs 6.7% and 2.7%; P = 0.006). The extension of myocardial damage at CMR was significantly greater in patients who developed MACE. In multivariable Cox regression, %LGE was an independent predictor of MACE (HR: 1.123 [95% CI: 1.064-1.185]; P < 0.001) together with T2 mapping values (HR: 1.190 [95% CI: 1.145-1.237]; P = 0.001). CONCLUSIONS: In MINOCA with early CMR execution, the %LGE and abnormal T2 mapping values were identified as independent predictors of adverse cardiac events at ∼3.0 years of follow-up. These parameters can be considered as high-risk markers in MINOCA.

Frozen elephant trunk technique using hybrid grafts: 15-year outcomes from a single-centre experience.

OBJECTIVES: The purpose of the study is to compare the short- and long-term outcomes of the frozen elephant trunk (FET) technique based on 2 different hybrid grafts implanted from January 2007 to July 2022. METHODS: The study includes patients who underwent an elective or emergency FET procedure. Short-term, long-term mortality and freedom from thoracic endovascular aortic repair (TEVAR) were the primary end points. Analyses were carried out separately for the periods 2007-2012 and 2013-2022. RESULTS: Of the 367 enrolled, 49.3% received E-Vita Open implantation and 50.7% received Thoraflex Hybrid implants. Overall mean age was 61 years [standard deviation (SD) = 11] and 80.7% were male. The average annual volume of FET procedures was 22.7 cases/year. Compared to E-Vita Open, patients implanted with Thoraflex Hybrid grafts were more likely to receive distal anastomosis in zone 2 (68.3% vs 11.6%, P < 0.001) with a shorter stent portion, mean = 103mm (SD = 11.3) vs mean = 149 mm (SD = 12.7; P < 0.001) and they underwent a reduced visceral ischaemia time, mean = 42.5 (SD = 14.2) vs mean= 61.0 (SD = 20.2) min, P < 0.001. In the period 2013-2022, overall survival at 1, 2 and 5 years was 74.8%, 72.5% and 63.2% for Thoraflex and 73.2%, 70.7% and 64.1% for E-Vita, without significant differences between groups (log-rank test = 0.01, P = 0.907). Overall freedom from TEVAR at 1, 2 and 5 years was 66.7%, 57.6% and 39.3% for Thoraflex and 79%, 69.7% and 66% for E-Vita, with significant differences between groups (log-rank test = 5.28, P = 0.029). In a competing risk analysis adjusted for chronic/residual aortic syndromes and stent diameter, the Thoraflex group was more likely to receive TEVAR during follow-up (subdistribution hazard ratio SHR = 2.12, 95% confidence interval 1.06-4.22). CONCLUSIONS: The FET technique addresses acute and chronic arch disease with acceptable morbidity and mortality. Downstream endovascular reinterventions are very common during follow-up.

Staging TEVAR after FET - an exception or the rule?

Purpose: Frozen elephant trunk (FET) was born as an ideal one-step procedure to treat complex arch and descending thoracic aorta pathology. It was then proved that it frequently needs reintervention, which can often be performed by thoracic endovascular aortic repair (TEVAR) extension since FET provides a safe proximal landing zone. We hereby describe our experience in TEVAR extension after FET, its main indications, technique, and outcomes. Methods: Between 2007 and 2022, 371 patients underwent FET at our center. Of these, 119 needed TEVAR extension. Some required more than one TEVAR, with a total of 154 procedures. The preoperative characteristics, indications, and outcomes were analyzed retrospectively. Results: Of 154 TEVAR procedures, 15 were performed in an urgent setting. Mean time from FET to TEVAR was 22,2 ± 28,73 months. Two patients died in the operating room; no others died during the hospital stay. Survival after 1, 2, 5, and 10 years was 96.2%, 93.9%, 90.1%, and 70.5% respectively. There was no statistically significant difference in the rates of TEVAR extension for patients in which a Thoraflex™ vs E-vita™ graft was used, nor for zone 2 vs zone 3 anastomosis and stent length. Conclusion: Though TEVAR extension is often required after FET, it is a safe and effective procedure with excellent post-operative outcomes in the short-, mid-, and long-term and allows successful treatment of complex aortic pathologies. Rigorous and specialized follow-up after FET is central to identify the right moment to intervene.

Overcoming Underpowering in the Outcome Analysis of Repaired-Tetralogy of Fallot: A Multicenter Database from the CMR/CT Working Group of the Italian Pediatric Cardiology Society (SICPed).

Background: Managing repaired tetralogy of Fallot (TOF) patients is still challenging despite the fact that published studies identified prognostic clinical or imaging data with rather good negative predictive accuracy but weak positive predictive accuracy. Heterogeneity of the initial anatomy, the surgical approach, and the complexity of the mechanism leading to dilation and ventricular dysfunction explain the challenge of predicting the adverse event in this population. Therefore, risk stratification and management of this population remain poorly standardized. Design: The CMR/CT WG of the Italian Pediatric Cardiology Society set up a multicenter observational clinical database of repaired TOF evaluations. This registry will enroll patients retrospectively and prospectively assessed by CMR for clinical indication in many congenital heart diseases (CHD) Italian centers. Data collection in a dedicated platform will include surgical history, clinical data, imaging data, and adverse cardiac events at 6 years of follow-up. Summary: The multicenter repaired TOF clinical database will collect data on patients evaluated by CMR in many CHD centers in Italy. The registry has been set up to allow future research studies in this population to improve clinical/surgical management and risk stratification of this population.

Survival and reoperation in acute aortic syndromes-a single-centre experience of 912 patients.

OBJECTIVES: Acute aortic syndromes are associated with poor outcomes, despite diagnostic and therapeutic advances. We analysed trends in volumes and outcomes from 2000 to 2021. METHODS: The study population includes 494 type A acute aortic syndromes (TAAAS) (54.2%) and 418 type B acute aortic syndromes (TBAAS) (45.8%). Primary outcomes were in-hospital mortality, long-term survival and freedom from aortic reoperation. RESULTS: Regardless the type of acute aortic syndrome, patient volumes increased over time. Patients with TBAAS were older, more likely to have comorbid conditions and previous cardiac surgery (P < 0.001), while cerebrovascular accidents were more frequent in TAAAS (P < 0.05). Among TAAAS, 143 (28.9%) required total arch and 351 (71.1%) hemiarch replacement. TBAAS management was medical therapy in 182 (43.5%), endovascular in 198 (47.4%) and surgical in 38 (9.1%) cases. Overall in-hospital mortality was 14.6% [18.2% in TAAAS (95% confidence interval (CI) 14.4-21.2%) vs 10.7% in TBAAS (95% CI 7.8%-13.7%); P = 0.0027]. After propensity score adjustment, in-hospital mortality exhibited a significantly decreasing trend from 2000 to 2021 (P < 0.001) in TAAAS and TBAAS. 1-, 5- and 10-year survival was 74.2%, 62.2% and 45.5% in TAAAS and 75.4%, 60.7% and 41.0% in TBAAS (P = 0.975), with no differences among treatment strategies. The adjusted cumulative reoperation risk at 10 years was more than two-fold in TBAAS versus TAAAS (9.5% vs 20.5%, hazard ratio (HR) = 2.30, 95% I 1.31-4.04). CONCLUSIONS: In the last decades, better patient triage and surgical/endovascular techniques led to substantial improvements in the management of acute aortic syndrome, with reduction in early mortality and reoperation rate. However, long-term mortality is still >50%.

Performance of Prognostic Scoring Systems in MINOCA: A Comparison among GRACE, TIMI, HEART, and ACEF Scores.

Background: the prognosis of patients with myocardial infarction with non-obstructive coronary arteries (MINOCA) is not benign; thus, prompting the need to validate prognostic scoring systems for this population. Aim: to evaluate and compare the prognostic performance of GRACE, TIMI, HEART, and ACEF scores in MINOCA patients. Methods: A total of 250 MINOCA patients from January 2017 to September 2021 were included. For each patient, the four scores at admission were retrospectively calculated. The primary outcome was a composite of all-cause death and acute myocardial infarction (AMI) at 1-year follow-up. The ability to predict 1-year all-cause death was also tested. Results: Overall, the tested scores presented a sub-optimal performance in predicting the composite major adverse event in MINOCA patients, showing an AUC ranging between 0.7 and 0.8. Among them, the GRACE score appeared to be the best in predicting all-cause death, reaching high specificity with low sensitivity. The best cut-off identified for the GRACE score was 171, higher compared to the cut-off of 140 generally applied to identify high-risk patients with obstructive AMI. When the scores were tested for prediction of 1-year all-cause death, the GRACE and the ACEF score showed very good accuracy (AUC = 0.932 and 0.828, respectively). Conclusion: the prognostic scoring tools, validated in AMI cohorts, could be useful even in MINOCA patients, although their performance appeared sub-optimal, prompting the need for risk assessment tools specific to MINOCA patients.

A Radiomic-Based Machine Learning Model Predicts Endometrial Cancer Recurrence Using Preoperative CT Radiomic Features: A Pilot Study.

BACKGROUND: Current prognostic models lack the use of pre-operative CT images to predict recurrence in endometrial cancer (EC) patients. Our study aimed to investigate the potential of radiomic features extracted from pre-surgical CT scans to accurately predict disease-free survival (DFS) among EC patients. METHODS: Contrast-Enhanced CT (CE-CT) scans from 81 EC cases were used to extract the radiomic features from semi-automatically contoured volumes of interest. We employed a 10-fold cross-validation approach with a 6:4 training to test set and utilized data augmentation and balancing techniques. Univariate analysis was applied for feature reduction leading to the development of three distinct machine learning (ML) models for the prediction of DFS: LASSO-Cox, CoxBoost and Random Forest (RFsrc). RESULTS: In the training set, the ML models demonstrated AUCs ranging from 0.92 to 0.93, sensitivities from 0.96 to 1.00 and specificities from 0.77 to 0.89. In the test set, AUCs ranged from 0.86 to 0.90, sensitivities from 0.89 to 1.00 and specificities from 0.73 to 0.90. Patients classified as having a high recurrence risk prediction by ML models exhibited significantly worse DSF (p-value < 0.001) across all models. CONCLUSIONS: Our findings demonstrate the potential of radiomics in predicting EC recurrence. While further validation studies are needed, our results underscore the promising role of radiomics in forecasting EC outcomes.

[Management of acute type B aortic dissection: treatment and follow-up]. / Gestione della dissezione aortica acuta di tipo B: trattamento e follow-up.

Aortic dissection is a life-threatening condition caused by a tear in the tunica intima which creates a false lumen into the aortic wall. Acute type B aortic dissection (TBAD) is defined by the presence of the entry tear in the aorta distal to the left subclavian artery, without ascending aorta and arch involvement, and accounts for 25-40% of all aortic dissections. Optimal medical therapy (OMT), focused on blood pressure and heart rate control, remains the gold standard treatment, especially for patients with uncomplicated TBAD, while complicated dissections require surgical therapy. Recent studies have shown that a considerable number of patients treated only with OMT develop late aorta-related complications that increase morbidity and mortality, as well as the need for surgical intervention. During the last decades, emerging evidence indicates that thoracic endovascular aortic repair (TEVAR) is safe and effective in the treatment of TBAD, both complicated and uncomplicated, with improved long-term survival outcomes and aortic remodeling in combination with OMT compared to OMT alone. However, in cases of acute uncomplicated TBAD the optimal timing for TEVAR is not entirely clarified and there is lack of long-term evidence. Therefore, the role of pre-emptive TEVAR for these patients is still uncertain and the management of acute uncomplicated TBAD remains challenging.

The Role of Non-Invasive Multimodality Imaging in Chronic Coronary Syndrome: Anatomical and Functional Pathways.

Coronary artery disease (CAD) is one of the major causes of mortality and morbidity worldwide, with a high socioeconomic impact. Currently, various guidelines and recommendations have been published about chronic coronary syndromes (CCS). According to the recent European Society of Cardiology guidelines on chronic coronary syndrome, a multimodal imaging approach is strongly recommended in the evaluation of patients with suspected CAD. Today, in the current practice, non-invasive imaging methods can assess coronary anatomy through coronary computed tomography angiography (CCTA) and/or inducible myocardial ischemia through functional stress testing (stress echocardiography, cardiac magnetic resonance imaging, single photon emission computed tomography-SPECT, or positron emission tomography-PET). However, recent trials (ISCHEMIA and REVIVED) have cast doubt on the previous conception of the management of patients with CCS, and nowadays it is essential to understand the limitations and strengths of each imaging method and, specifically, when to choose a functional approach focused on the ischemia versus a coronary anatomy-based one. Finally, the concept of a pathophysiology-driven treatment of these patients emerged as an important goal of multimodal imaging, integrating 'anatomical' and 'functional' information. The present review aims to provide an overview of non-invasive imaging modalities for the comprehensive management of CCS patients.

Long Term Follow-Up of Patients with Systemic Right Ventricle and Biventricular Physiology: A Single Centre Experience.

BACKGROUND: A progressively increasing prevalence of congenital heart disease (CHD) in adulthood has been noticed in recent decades; CHD cases with a systemic right ventricle have a poorer outcome. METHODS: Seventy-three patients with SRV evaluated in an outpatient clinic between 2014 and 2020 were enrolled in this study. Thirty-four patients had a transposition of the great arteries treated with an atrial switch operation; 39 patients had a congenitally corrected transposition of the great arteries (ccTGA). RESULTS: Mean age at the first evaluation was 29.6 ± 14.2 years; 48% of the patients were female. The NYHA class at the visit was III or IV in 14% of the cases. Thirteen patients had at least one previous pregnancy. In 25% of the cases, complications occurred during pregnancy. Survival free from adverse events was 98.6% at one year and 90% at 6-year follow-up without any difference between the two groups. Two patients died and one received heart transplantation during follow-up. The most common adverse event during follow-up was the presence of arrhythmia requiring hospitalization (27.1%), followed by heart failure (12.3%). The presence of LGE together with lower exercise capacity, higher NYHA class and more dilated and/or hypokinetic RV predicted a poorer outcome. Quality of life was similar to the QoL of the Italian population. CONCLUSIONS: Long-term follow-up of patients with a systemic right ventricle is characterized by a high incidence of clinical events, prevalently arrhythmias and heart failure, which cause most of the unscheduled hospitalizations.

Stereotactic Arrhythmia Radioablation (STAR): A Multidisciplinary Narrative Minireview of Preclinical Studies.

The aim of this narrative review of the literature was to collect and analyze the results of the published preclinical studies on stereotactic arrhythmia radioablation (STAR) in the treatment of refractory cardiac arrhythmias. A literature search was conducted on PubMed using the following terms: ("stereotactic" OR "SBRT" OR "SABR" OR "radioablation" OR "radiosurgery") AND ("arrhythmia" OR "tachycardia"). Preclinical and pathological reports published in English without time limit, comprising studies of STAR in animal models and histological analyzes of explanted animal and human hearts were included. The analyzed studies confirm that doses lower than 25 Gy seem to produce sub-optimal therapeutic results whereas doses >35 Gy are less safe in terms of radiation-induced toxicity. However, long-term results (>1 year) are still missing and reporting outcomes based on low dose irradiation (≤15 Gy). Finally, STAR proved to be an effective therapy in the analyzed studies despite the irradiation of rather different cardiac targets. Therefore, additional studies are needed to: 1) compare the outcomes of STAR at doses of 25 Gy versus 30 Gy; 2) evaluate the long-term results (>1 year) in animal models irradiated at doses similar to those used in the clinic; 3) define the optimal target.

Paraglottic Space Invasion in Glottic Laryngeal Cancer: A Clinical-Pathological Study.

OBJECTIVE: This study aims to prospectively compare endoscopic, radiological, and pathological features of a cohort of patients with glottic laryngeal squamous cell carcinoma (LSCC) undergoing open partial horizontal laryngectomy (OPHL) type II/III or total laryngectomy to better understand the reliability of preoperative endoscopy and computed tomography (CT) to predict the inferior paraglottic space (iPGS) involvement. METHODS: We prospectively compared the endoscopic, radiological, and pathological findings in patients with glottic LSCC who underwent OPHL II/III, or total laryngectomy. RESULTS: Endoscopy achieved a diagnostic accuracy of 87.2% for the anterior iPGS (iPGSa) and 86.1% for the posterior iPGS (iPGSp). There was no statistically significant difference in terms of histopathologic iPGSa involvement between reduced (85%-17/20 pts) and absent (92%-24/26 pts) vocal cord mobility (p = 0.39). CT alone did not improve the diagnostic performance of the endoscopy, reaching a diagnostic accuracy of 62.9% and 73.7% for the iPGSa and iPGSp, respectively. When endoscopy and CT were combined, the diagnostic performance improved for the iPGSp, achieving a sensitivity (Se), specificity (Spe), positive predictive value (PPV), and negative predictive value (NPV) of 100%, 89.8%, 68.7%, and 100%, respectively. On the contrary, the combination of CT and endoscopy improved only the Se and NPV for the iPGSa with respect to the sole endoscopic assessment. CONCLUSIONS: Whenever motility impairment is present, a histopathologic invasion of the iPGS should be suspected. Endoscopic assessment of laryngeal motility achieved a satisfactory value of Se, Spe, PPV, and NPV in predicting the involvement of the iPGS. CT scan is still the mainstay imaging technique in the clinical staging of patients with LSCC. LEVEL OF EVIDENCE: 4 Laryngoscope, 133:1184-1190, 2023.

Short term outcome of myocarditis and pericarditis following COVID-19 vaccines: a cardiac magnetic resonance imaging study.

To evaluate clinical and cardiac magnetic resonance (CMR) short-term follow-up (FU) in patients with vaccine-associated myocarditis, pericarditis or myo-pericarditis (VAMP) following COVID-19 vaccination. We retrospectively analyzed 44 patients (2 women, mean age: 31.7 ± 15.1 years) with clinical and CMR manifestations of VAMP, recruited from 13 large tertiary national centers. Inclusion criteria were troponin raise, interval between the last vaccination dose and onset of symptoms < 25 days and symptoms-to-CMR < 20 days. 29/44 patients underwent a short-term FU-CMR with a median time of 3.3 months. Ventricular volumes and CMR findings of cardiac injury were collected in all exams. Mean interval between the last vaccination dose and the onset of symptoms was 6.2 ± 5.6 days. 30/44 patients received a vaccination with Comirnaty, 12/44 with Spikevax, 1/44 with Vaxzevria and 1/44 with Janssen (18 after the first dose of vaccine, 20 after the second and 6 after the "booster" dose). Chest pain was the most frequent symptom (41/44), followed by fever (29/44), myalgia (17/44), dyspnea (13/44) and palpitations (11/44). At baseline, left ventricular ejection fraction (LV-EF) was reduced in 7 patients; wall motion abnormalities have been detected in 10. Myocardial edema was found in 35 (79.5%) and LGE in 40 (90.9%) patients. Clinical FU revealed symptoms persistence in 8/44 patients. At FU-CMR, LV-EF was reduced only in 2 patients, myocardial edema was present in 8/29 patients and LGE in 26/29. VAMPs appear to have a mild clinical presentation, with self-limiting course and resolution of CMR signs of active inflammation at short-term follow-up in most of the cases.

Inter-observer variability of expert-derived morphologic risk predictors in aortic dissection.

OBJECTIVES: Establishing the reproducibility of expert-derived measurements on CTA exams of aortic dissection is clinically important and paramount for ground-truth determination for machine learning. METHODS: Four independent observers retrospectively evaluated CTA exams of 72 patients with uncomplicated Stanford type B aortic dissection and assessed the reproducibility of a recently proposed combination of four morphologic risk predictors (maximum aortic diameter, false lumen circumferential angle, false lumen outflow, and intercostal arteries). For the first inter-observer variability assessment, 47 CTA scans from one aortic center were evaluated by expert-observer 1 in an unconstrained clinical assessment without a standardized workflow and compared to a composite of three expert-observers (observers 2-4) using a standardized workflow. A second inter-observer variability assessment on 30 out of the 47 CTA scans compared observers 3 and 4 with a constrained, standardized workflow. A third inter-observer variability assessment was done after specialized training and tested between observers 3 and 4 in an external population of 25 CTA scans. Inter-observer agreement was assessed with intraclass correlation coefficients (ICCs) and Bland-Altman plots. RESULTS: Pre-training ICCs of the four morphologic features ranged from 0.04 (-0.05 to 0.13) to 0.68 (0.49-0.81) between observer 1 and observers 2-4 and from 0.50 (0.32-0.69) to 0.89 (0.78-0.95) between observers 3 and 4. ICCs improved after training ranging from 0.69 (0.52-0.87) to 0.97 (0.94-0.99), and Bland-Altman analysis showed decreased bias and limits of agreement. CONCLUSIONS: Manual morphologic feature measurements on CTA images can be optimized resulting in improved inter-observer reliability. This is essential for robust ground-truth determination for machine learning models. KEY POINTS: • Clinical fashion manual measurements of aortic CTA imaging features showed poor inter-observer reproducibility. • A standardized workflow with standardized training resulted in substantial improvements with excellent inter-observer reproducibility. • Robust ground truth labels obtained manually with excellent inter-observer reproducibility are key to develop reliable machine learning models.