Fix: Scientific Report https://doi.org/10.1038/S41598-024-66840-1, published online on July 9, 2024
The original version of the article contained an error.
First, in the methodology section, under the subheading of the “research population”, the Aneux morphological database was not declared and cited according to the terms of use. the result,
“The dataset used in this study was carefully provided by Zenodo.org.33provided access to the geometry of over 700 cerebral aneurysms extracted from patients in Sheffield, Milan, Geneva and Barcelona. ”
“The dataset used in this study was graciously provided by the AneuX morphology database, an open-access, multi-centric database combining data from three European projects: AneuX project (www.aneux.ch), @neurIST project (www.aneurist.org), and Aneurisk (http://ecm2.mathcs.emory.edu/aneuriskweb/index)30provides access to the geometry of over 700 cerebral aneurysms extracted from patients in Sheffield, Milan, Geneva and Barcelona. ”
References 33 listed below:
33. N. Juchler, Bijlenga, Philippe, & Hirsch, Sven. , “Aneux Molphology Database (v1.0), 2022.
30. N. Juchler, S. Schilling, P. Bijlenga, V. Kurtcuoglu, and S. Hirsch, “Size of shape cards: Image-based morphology analysis reveals that 3D shapes identify intracranial aneurysm disease conditions where anaerobic is superior to aneurysm size.” front. New Roll. 13809391 (2022).
As a result of the change, the reference has been changed.
Second, in the methodology section, cross-validation of models was omitted under the “machine learning” subheading. As a result,
“This approach to model selection and hyperparameter tuning increased the robustness and accuracy of predictions for the classification of non-ruptured cerebral aneurysms. The MLP model used three hidden layers with adaptive learning rates and identity activation functions.”
“This approach to model selection and hyperparameter tuning increased the robustness and accuracy of predictions for the classification of non-ruptured cerebral aneurysms. Additionally, we performed five-fold cross-validation on all models to minimize the effect of random division and improve the reliability of the results.
Third, the results and discussion sections have been updated to reflect more commonly accepted terms under the latent head of “dominant features.” The subheading title has been revised to “The Importance of Functions,” and as a result,
“The SVM model identifies the first five dominant features as EI (elliptic index), SR (size ratio), I (irregularity), UI (unsynthesis index), and IR (ideal rounding), which are new parameters introduced in this study.”
“The SVM model identifies the first five key features as EI (Ellipticity Index), SR (Size Ratio), I (Irregularity), UI (Unsynthesis Index), and IR (ideal Rounding), a new parameter introduced in this study.”
“We currently make a brief comparison between previous and current studies, focusing on the test data sets used in all studies. To facilitate this analysis, we refer to presenting the results of six comparable studies, as previously shown.
As the range of parameters considered increases, we expect a shift in relative importance assigned to each parameter. Additionally, increasing the size of the dataset increases the reliability of the results. Among the significance parameters, size ratios appear as a focus on recurrence, highlighting their inherent importance in assessing the risk of rupture. Once again, we emphasized the importance of recall scores given the inherent sensitivity to medical data. It is noteworthy that our study is a metric that unfortunately lacks from previous studies and thus achieves metrics that limit direct comparisons.
Table 3 shows the results of six comparable studies along with the results of our own research. As previously shown, we sought to incorporate a comprehensive set of morphological parameters to ensure the robustness of our findings.
As the range of parameters considered increases, we expect a shift in relative importance assigned to each parameter. Additionally, increasing the size of the dataset increases the reliability of the results. Among the significance parameters, size ratios appear as a focus on recurrence, highlighting their inherent importance in assessing the risk of rupture. Once again, we emphasized the importance of recall scores given the inherent sensitivity to medical data. It is noteworthy that our study is a metric that unfortunately lacks from previous studies and thus achieves metrics that limit direct comparisons. ”
“We currently make a brief comparison of previous studies and current studies, focusing on the test data sets utilized in all studies. To facilitate this analysis, we are focusing on Table 3, which shows the results of six comparable studies along with the results of our own research.
As the range of parameters considered increases, we expect a shift in the relative importance assigned to each parameter. Additionally, increasing the size of the dataset increases the reliability of the results. Among the significance parameters, size ratios appear as a focus on recurrence, highlighting their inherent importance in assessing the risk of rupture. Once again, we emphasized the importance of recall scores given the inherent sensitivity to medical data. In particular, our study achieves excellent recall scores, which unfortunately is absent metric in previous works, limiting direct comparisons. ”
Furthermore, in the “Conclusion” section,
“The neck circumference ranked sixth and seventh in the MLP as the dominant feature of SVM, further highlighting the value of exploring previous meaningless features.”
“The neck circumference ranked sixth and seventh in the MLP as an important feature in SVM, further highlighting the value of investigating previous meaningless features.”
Additionally, Figure 8 and the legends for each have been updated. The original Figure 8 and accompanying legend are displayed below.
The dominant feature of two top performance models.
The original article has been revised.
