Support Vector Machine for Stratification of Cognitive Impairment Using 3D T1WI in Patients with Type 2 Diabetes Mellitus

Diabetes Metabolic Syndrome and Obesity, Год журнала: 2025, Номер Volume 18, С. 435 - 451

Опубликована: Фев. 1, 2025

Purpose: To explore the potential of MRI-based radiomics in predicting cognitive dysfunction patients with diagnosed type 2 diabetes mellitus (T2DM). Patients and Methods: In this study, data on 158 T2DM were retrospectively collected between September 2019 December 2020. The participants categorized into a normal function (N) group (n=30), mild impairment (MCI) (n=90), dementia (DM) (n=38) according to Chinese version Montréal Cognitive Assessment Scale-B (MoCA-B). Radiomics features extracted from brain tissue except ventricles sulci 3D T1WI images, support vector machine (SVM) model was then established identify CI N groups, MCI DM respectively. models evaluated based their area under receiver operating characteristic curve (AUC), Precision (P), Recall rate (Recall, R), F1-score, Support. Finally, ROC curves plotted for each model. Results: study consisted 68 cases group, 54 training set 14 verification set, 128 included 90 sets 38 sets. consistency inter-group intra-group two physicians 0.86 0.90, After selection, there 11 optimal distinguish 12 DM. test AUC SVM classifier 0.857 accuracy 0.830 distinguishing N, while 0.821 Conclusion: MRI exhibits high efficacy diagnosis evaluation its severity among T2DM. Keywords: dysfunction, radiomics, magnetic resonance imaging, machine, mellitus,

Язык: Английский

---

Using interpretable rule-learning artificial intelligence to optimally differentiate adrenal pheochromocytomas from adenomas with CT radiomics

Abdominal Radiology, Год журнала: 2025, Номер unknown

Опубликована: Март 26, 2025

Язык: Английский

---

Comparative Evaluation of Machine Learning-Based Radiomics and Deep Learning for Breast Lesion Classification in Mammography

Diagnostics, Год журнала: 2025, Номер 15(8), С. 953 - 953

Опубликована: Апрель 9, 2025

Background: Breast cancer is the second leading cause of cancer-related mortality among women, accounting for 12% cases. Early diagnosis, based on identification radiological features, such as masses and microcalcifications in mammograms, crucial reducing rates. However, manual interpretation by radiologists complex subject to variability, emphasizing need automated diagnostic tools enhance accuracy efficiency. This study compares a radiomics workflow machine learning (ML) with deep (DL) approach classifying breast lesions benign or malignant. Methods: matRadiomics was used extract features from mammographic images 1219 patients CBIS-DDSM public database, including 581 cases 638 masses. Among ML models, linear discriminant analysis (LDA) demonstrated best performance both lesion types. External validation conducted private dataset 222 evaluate generalizability an independent cohort. Additionally, EfficientNetB6 model employed comparison. Results: The LDA achieved mean AUC 68.28% 61.53% In external validation, values 66.9% 61.5% were obtained, respectively. contrast, superior performance, achieving 81.52% 76.24% masses, highlighting potential DL improved accuracy. Conclusions: underscores limitations ML-based diagnosis. Deep proves be more effective approach, offering enhanced supporting clinicians improving patient management.

Язык: Английский

---

Investigation of Radiomic Feature Normalizations, Feature Selection, and Modeling (Preprint)

Опубликована: Дек. 31, 2024

Radiomics is a field that extracts additional information from patient’s medical images through quantitative analysis in search of clinically useful patterns and relationships. Using current standard-of-care images, we can obtain patient-specific noninvasive measures which give us in-sight into disease progression treatment response may not be visually apparent. However, there are barriers must overcome before widespread clinical implementation. This instability radiomic feature values hampers research progression. Researchers currently addressing this the establishment imaging standardizations reporting requirements. it has been shown each aspect hand-crafted radiomics workflow influences other aspects. Understanding impact importance detail every only its respective role, but also combination with rest workflow. study aims to aid characterization commonly used techniques for normalization, selection, modeling by generating investigating unique workflows. The overarching goal explore how component uniquely impacts overall model performance. Computed Tomography 302 patients non-small cell lung cancer were investigated features extraction gross tumor volume Pyradiomics. Each resultant separated 80%:20% training &:testing cohorts programmatically varying either selection method, number features, or model. Thusly, over 1900 workflows created signature sets (RSSs) ability differences judged characterize capabilities via area under curve (ROC-AUC) statistically log-rank. Relative abilities eleven normalizations individually assessed sub-grouping RSS normalization employed determining average, standard deviation median ROC-AUC. We similarly six selections, nine models , selected features. determined best their consistently generate RSSs relatively higher performance both testing datasets along relative significant RSSs. possesses ability. important note none methods demonstrated clear dominance. Best performing moderately increased training/testing : Results indicate thorough experimental determination key work-flow aspects provides essential insights interplay between

Язык: Английский

---